Statalist

test post

2006-05-11T12:36:00.000-04:00

Testing

Re: st: RE: RE: RE: list subjects with a similar value

2006-03-22T20:44:00.000-05:00

>Personally if you want to be able to revert to older versions of files >I'd recommend simply creating a copy before doing major revisions and >simply append a date in numeric format at some point to the filename >(before '.' would be most appropriate).

Thank you Neil. Are there resources for systematic and efficient "best practices" in do-file colding, that beginning Stata users can emulate and integrate into their own work? Michael

* * For searches and help try: * http://www.stata.com/support/faqs/res/findit.html * http://www.stata.com/support/statalist/faq * http://www.ats.ucla.edu/stat/stata/

st: simultaneous equation with qualitative variables

2006-03-22T20:43:00.000-05:00

how can i use stata to estimate simultaneous equations with qualitative variables as follows: y1=a1+b1y2+c1x1+e1 y2=a2+b2y1+c2x2+e2 y1*=1 if y1>0 y1*=0 if y1<=0 y2*=1 if y2>0 y2*=0 if y2<=0

* * For searches and help try: * http://www.stata.com/support/faqs/res/findit.html * http://www.stata.com/support/statalist/faq * http://www.ats.ucla.edu/stat/stata/

RE: st: RE: RE: RE: list subjects with a similar value

2006-03-22T20:23:00.000-05:00

MS-Word I understand to be a word processor. Recall that many members of Statalist do not use and certainly are not expert in Windows or Microsoft products generally.

In Vim, the way that seems to be most natural is a Unix way. You can have two files open in two windows and set it so that differences are highlighted. So, one could be a previous version and the other a working version.

I gather that Word behaves differently.

In general, good text editors will have something loosely similar. None that I know of regards it as a virtue to emulate Word.

Nick n.j.cox@durham.ac.uk

Michael McCulloch > Sorry; I was referring to changes in coding that one writes > in a do-file. > >Changes to what?

> > > Jennifer response brings to mind a question that recently > > > occurred to me: > > > Is there a Stata-compatible text editor that, like MS-WORD, > > > can highlight changes? > > > > > >Whoops, of course you're right. When I play with things to > > > work out code > > > >I usually don't keep the changes, so I mangle working datasets > > > >willy-nilly and didn't think to change the conditioned keep to a > > > >conditioned list. > > > >The use of bysorting and _N is much neater and more flexible.

* * For searches and help try: * http://www.stata.com/support/faqs/res/findit.html * http://www.stata.com/support/statalist/faq * http://www.ats.ucla.edu/stat/stata/

Re: st: RE: RE: RE: list subjects with a similar value

2006-03-22T20:16:00.000-05:00

On 3/23/06, Michael McCulloch <mm@pinest.org> wrote: > Jennifer response brings to mind a question that recently occurred to me: > Is there a Stata-compatible text editor that, like MS-WORD, can highlight > changes?

Although I don't use it myself there are version control features in Emacs. These work with programming control systems such as CVS or SVN, and I'm not sure what your mileage would be writing do/ado files under such schemas, but I suspect it is possible (don't quote me on that though :).

This isn't exactly the same as M$-words document tracking (which personally I find hideously hard to follow, particularly when there are multiple authors making revisions, I've seen some docs that end up looking like the old TV-test screens :-), but it does allow you to track the changes that you are making. One of the main problems (as I see it) is that to write do/ado-files you need a _text_ editor, and word is not a text-editor, but a word-processor, so all the colour changes that you see are essentially mark-ups of the original text, and such mark-ups would render the Stata code uninterpretable.

See http://www.gnu.org/software/emacs/manual/html_node/Version-Control.html for more on Emacs' VC system.

Personally if you want to be able to revert to older versions of files I'd recommend simply creating a copy before doing major revisions and simply append a date in numeric format at some point to the filename (before '.' would be most appropriate).

HTH's

Neil -- "The best safety device in climbing is always situated between your ears" - Ross Weiter, Perth Rock Climbing Guide (2002)

Email - nshephard@gmail.com / neilshep@cyllene.uwa.edu.au Website - http://slack.ser.man.ac.uk/ Blog - http://slack---line.blogspot.com/ Flickr - http://www.flickr.com/photos/slackline/

* * For searches and help try: * http://www.stata.com/support/faqs/res/findit.html * http://www.stata.com/support/statalist/faq * http://www.ats.ucla.edu/stat/stata/

RE: st: RE: RE: RE: list subjects with a similar value

2006-03-22T20:05:00.000-05:00

Sorry; I was referring to changes in coding that one writes in a do-file.

At 04:47 PM 3/22/2006, you wrote: >Changes to what? > >Nick >n.j.cox@durham.ac.uk > >Michael McCulloch > > > Jennifer response brings to mind a question that recently > > occurred to me: > > Is there a Stata-compatible text editor that, like MS-WORD, > > can highlight changes? > > > >Whoops, of course you're right. When I play with things to > > work out code > > >I usually don't keep the changes, so I mangle working datasets > > >willy-nilly and didn't think to change the conditioned keep to a > > >conditioned list. > > >The use of bysorting and _N is much neater and more flexible. > >* >* For searches and help try: >* http://www.stata.com/support/faqs/res/findit.html >* http://www.stata.com/support/statalist/faq >* http://www.ats.ucla.edu/stat/stata/

Best wishes, Michael

____________________________________

Michael McCulloch Pine Street Clinic 124 Pine Street, San Anselmo, CA 94960-2674 tel 415.407.1357 fax 415.485.1065 email: mm@pinest.org web: www.pinest.org www.pinestreetfoundation.org

* * For searches and help try: * http://www.stata.com/support/faqs/res/findit.html * http://www.stata.com/support/statalist/faq * http://www.ats.ucla.edu/stat/stata/

RE: st: RE: RE: RE: list subjects with a similar value

2006-03-22T19:49:00.000-05:00

Changes to what?

Nick n.j.cox@durham.ac.uk

Michael McCulloch > Jennifer response brings to mind a question that recently > occurred to me: > Is there a Stata-compatible text editor that, like MS-WORD, > can highlight changes? > >Whoops, of course you're right. When I play with things to > work out code > >I usually don't keep the changes, so I mangle working datasets > >willy-nilly and didn't think to change the conditioned keep to a > >conditioned list. > >The use of bysorting and _N is much neater and more flexible.

* * For searches and help try: * http://www.stata.com/support/faqs/res/findit.html * http://www.stata.com/support/statalist/faq * http://www.ats.ucla.edu/stat/stata/

st: RE: RE: RE: Macro display format

2006-03-22T18:11:00.000-05:00

It's a good question.

I guess: -scatter- calls -graph- calls ... something that clears your r-class results.

-graph- needs all sorts of little calculations to work out what to show. It tends to do this on the fly, but either way, you can lose your r-class results.

Whatever it is, it is quite deep down.

Nick n.j.cox@durham.ac.uk

Alex Ogan

> Sorry if this is a silly question. > > I did the following sequence of commands. I closed the > scatter as soon > as it opened. No other commands. > > Why does r(mean) go away after you use it in the scatter with the > formatting extended function? > > . sysuse auto, clear > (1978 Automobile Data) > > . quietly summ length > > . di `r(mean)' > 187.93243 > > . di `: di %9.1f `r(mean)'' > 187.9 > > . di `: di %9.1f `r(mean)'' > 187.9 > > . twoway scatter length mpg, text(200 35 "The mean of length is:`: di > %2.1f `r(mean)''") > > . di `: di %9.1f `r(mean)'' > > > . di `r(mean)' > > > . >

* * For searches and help try: * http://www.stata.com/support/faqs/res/findit.html * http://www.stata.com/support/statalist/faq * http://www.ats.ucla.edu/stat/stata/

st: "sureg" with long-format data

2006-03-22T18:04:00.000-05:00

Hello all,

I have searched the web and available Stata faqs and helpfiles for an answer to an sureg question and am unable to find one on my own. Much obliged if anyone can help me through this particular puzzle.

I have a dataset oriented lengthwise where observations are grouped by country and year. I would like to run sureg using this vertical orientation. For example, in the sureg syntax, I would like to write "sureg (yvar xvar)" as equation 1 representing the first 35 observations, followed by "(yvar xvar)" as equation 2 representing the next 35 observations, and so on. My difficulty is that without re-orienting the data width-wise (which creates a cumbersome heap of newly-named X and Y variables and stops me from being able to tweak the regressions around the edges) I have no way to distinguish for Stata that instead of each equation having different Y and X variable names, each equation has the same variable names but should contain a different set of 35 observations. My abbreviated dataset looks like:

Country Year Yvar Xvar Australia 2004 .5 14 Australia 2003 .6 17 .... .. Australia 1970 .2 10

Austria 2004 .9 35 Austria 2003 .8 14 .... .. Austria 1970 .15 14

Belgium 2004 .13 7 .... (and so on for approximately 100 countries)

Ideally, the sureg command would allow something like "by country: sureg (Yvar Xvar)" but instead, I have to reorient the data widthwise and write "sureg (YvarAustralia XvarAustralia) (YvarAustria XvarAustria)..." for about one-hundred different equations (and change all 100 when I make adjustments). Requiring this horizontal orientation to make sureg work seems remarkably inefficient in the command line, but I can't seem to find another way around it. Many thanks for any and all ideas.

Cheers, Eric

Eric K. Bielke Regulatory Economics Advisor www.telecom.co.nz Level 2, Telecom House, 68-86 Jervois Quay, P O Box 570, Wellington, New Zealand

This communication, including any attachments, is confidential. If you are not the intended recipient, you should not read it - please contact me immediately, destroy it, and do not copy or use any part of this communication or disclose anything about it. Thank you. Please note that this communication does not designate an information system for the purposes of the Electronic Transactions Act 2002.

* * For searches and help try: * http://www.stata.com/support/faqs/res/findit.html * http://www.stata.com/support/statalist/faq * http://www.ats.ucla.edu/stat/stata/

st: RE: RE: Macro display format

2006-03-22T17:59:00.000-05:00

Sorry if this is a silly question.

I did the following sequence of commands. I closed the scatter as soon as it opened. No other commands.

Why does r(mean) go away after you use it in the scatter with the formatting extended function?

. sysuse auto, clear (1978 Automobile Data)

. quietly summ length

. di `r(mean)' 187.93243

. di `: di %9.1f `r(mean)'' 187.9

. twoway scatter length mpg, text(200 35 "The mean of length is:`: di %2.1f `r(mean)''")

. di `: di %9.1f `r(mean)''

. di `r(mean)'

-----Original Message----- From: owner-statalist@hsphsun2.harvard.edu [mailto:owner-statalist@hsphsun2.harvard.edu] On Behalf Of Nick Cox Sent: Wednesday, March 22, 2006 5:51 PM To: statalist@hsphsun2.harvard.edu Subject: st: RE: Macro display format

No. You should try

` : di %9.1f `r(mean)''

See help extended_fcn.

Tip: I would go `: di %2.1f `r(mean)'' even if you are sure that format is too restrictive. You're likely to be wrong, as Stata will stretch the space to avoid damage. However, with %9.1f you are likely to get the ugly spaces that are a consequence of what you asked for.

Nick n.j.cox@durham.ac.uk

Thomas Speidel > I am trying to include the content of a macro within a graph, but I'm > having problems with the display format. > > For example: > > sysuse auto, clear > qui: summ length > twoway scatter length mpg, text(200 35 "The mean of length is: > `r(mean)'") > > How do I change the format of the macro to display something > like %9.2f? > > I tried: > ... , text(200 35 "The mean of length is: `%9.1f `r(mean)''") > > Am I missing some triple compound quote? :-)

* * For searches and help try: * http://www.stata.com/support/faqs/res/findit.html * http://www.stata.com/support/statalist/faq * http://www.ats.ucla.edu/stat/stata/

This message is intended solely for the designated recipient(s). It may contain confidential or proprietary information and may be subject to confidentiality protections. If you are not a designated recipient, you may not review, copy, or distribute this message. If you receive this in error, please notify the sender by reply e-mail and delete this message.

Arrowstreet Capital, L.P. is an Equal Opportunity Employer.

* * For searches and help try: * http://www.stata.com/support/faqs/res/findit.html * http://www.stata.com/support/statalist/faq * http://www.ats.ucla.edu/stat/stata/

st: RE: Macro display format

2006-03-22T17:52:00.000-05:00

No. You should try

` : di %9.1f `r(mean)''

See help extended_fcn.

Nick n.j.cox@durham.ac.uk

* * For searches and help try: * http://www.stata.com/support/faqs/res/findit.html * http://www.stata.com/support/statalist/faq * http://www.ats.ucla.edu/stat/stata/

st: RE: RE: RE: list subjects with a similar value

2006-03-22T17:48:00.000-05:00

Whoops, of course you're right. When I play with things to work out code I usually don't keep the changes, so I mangle working datasets willy-nilly and didn't think to change the conditioned keep to a conditioned list. The use of bysorting and _N is much neater and more flexible.

N.J.Cox wrote: >Whoa! The question was just about _listing_. >You just changed Raoul's dataset by throwing >much of it away.

* * For searches and help try: * http://www.stata.com/support/faqs/res/findit.html * http://www.stata.com/support/statalist/faq * http://www.ats.ucla.edu/stat/stata/

st: Macro display format

2006-03-22T17:43:00.000-05:00

I am trying to include the content of a macro within a graph, but I'm having problems with the display format.

For example:

sysuse auto, clear qui: summ length twoway scatter length mpg, text(200 35 "The mean of length is: `r(mean)'")

How do I change the format of the macro to display something like %9.2f?

I tried: ... , text(200 35 "The mean of length is: `%9.1f `r(mean)''")

Am I missing some triple compound quote? :-)

Thanks, Thomas

-- Thomas Speidel Statistical Associate Clinical Trials Unit Tom Baker Cancer Centre 1331 - 29th Street N.W. Calgary, AB, T2N 4N4

Tel. (403) 521-3370 Email: thomassp@cancerboard.ab.ca

This e-mail and any attachments may contain confidential and privileged information. If you are not the intended recipient, please notify the sender immediately by return e-mail, delete this e-mail and destroy any copies. Any dissemination or use of this information by a person other than the intended recipient is unauthorized and may be illegal.

* * For searches and help try: * http://www.stata.com/support/faqs/res/findit.html * http://www.stata.com/support/statalist/faq * http://www.ats.ucla.edu/stat/stata/

RE: st: list subjects with a similar value

2006-03-22T17:38:00.000-05:00

You are avoiding the command -duplicates- and doing it from first principles. That is a very good idea. -duplicates- is just a wrapper for stuff like this.

But the three steps here can be cut to two.

bysort date_of_birth : gen dob_duplicate = _N list id date_of_birth if dob_duplicate >= 2

Nick n.j.cox@durham.ac.uk

clinton.thompson@summitllc.us > there may be a more elegant way to do this, albeit this is > but one attempt: > > * obtain the number of duplicates w/in date of birth > bysort date_of_birth: gen dob_duplicate = _N > * tag each combination of DOB & the duplicates therein > egen dob_tag = tag(date_of_birth dob_duplicate) > * list the ID & DOB associated w/ each repeated DOB... > list id date_of_birth dob_tag if dob_tag > > I have a large database and would like to list the idnumber of all > > subjects with the same date of birth. How do I do this? I have tried > > .duplicate, but can figure out how to do it. Thanks.

* * For searches and help try: * http://www.stata.com/support/faqs/res/findit.html * http://www.stata.com/support/statalist/faq * http://www.ats.ucla.edu/stat/stata/

st: RE: RE: list subjects with a similar value

2006-03-22T17:35:00.000-05:00

Whoa! The question was just about _listing_. You just changed Raoul's dataset by throwing much of it away.

duplicates tag dateofbirth, gen(tag) sort dateofbirth list id dateofbirth if tag

Nick n.j.cox@durham.ac.uk

Marino, Jennifer > Maybe try something along these lines: > > duplicates tag dateofbirth, gen(tag) > drop if tag==0 > sort dateofbirth > by dateofbirth: list id

Raoul C Reulen > I have a large database and would like to list the idnumber of all > subjects with the same date of birth. How do I do this? I have tried > .duplicate, but can figure out how to do it. Thanks.

* * For searches and help try: * http://www.stata.com/support/faqs/res/findit.html * http://www.stata.com/support/statalist/faq * http://www.ats.ucla.edu/stat/stata/

Re: st: list subjects with a similar value

2006-03-22T17:30:00.000-05:00

there may be a more elegant way to do this, albeit this is but one attempt:

* obtain the number of duplicates w/in date of birth bysort date_of_birth: gen dob_duplicate = _N * tag each combination of DOB & the duplicates therein egen dob_tag = tag(date_of_birth dob_duplicate) * list the ID & DOB associated w/ each repeated DOB... list id date_of_birth dob_tag if dob_tag

note that i didn't subject this to a rigorous test...but i think it works, nonetheless. --clint

> Hi, > > I have a large database and would like to list the idnumber of all > subjects with the same date of birth. How do I do this? I have tried > .duplicate, but can figure out how to do it. Thanks. > > Raoul > > * > * For searches and help try: > * http://www.stata.com/support/faqs/res/findit.html > * http://www.stata.com/support/statalist/faq > * http://www.ats.ucla.edu/stat/stata/ >

* * For searches and help try: * http://www.stata.com/support/faqs/res/findit.html * http://www.stata.com/support/statalist/faq * http://www.ats.ucla.edu/stat/stata/

st: RE: list subjects with a similar value

2006-03-22T17:25:00.000-05:00

Maybe try something along these lines:

duplicates tag dateofbirth, gen(tag) drop if tag==0 sort dateofbirth by dateofbirth: list id

Jen Marino

-----Original Message----- From: owner-statalist@hsphsun2.harvard.edu [mailto:owner-statalist@hsphsun2.harvard.edu] On Behalf Of Raoul C Reulen Sent: Wednesday, March 22, 2006 2:14 PM To: statalist@hsphsun2.harvard.edu Subject: st: list subjects with a similar value

Hi, I have a large database and would like to list the idnumber of all subjects with the same date of birth. How do I do this? I have tried .duplicate, but can figure out how to do it. Thanks. Raoul

* * For searches and help try: * http://www.stata.com/support/faqs/res/findit.html * http://www.stata.com/support/statalist/faq * http://www.ats.ucla.edu/stat/stata/

st: list subjects with a similar value

2006-03-22T17:15:00.000-05:00

Hi, I have a large database and would like to list the idnumber of all subjects with the same date of birth. How do I do this? I have tried .duplicate, but can figure out how to do it. Thanks. Raoul

* * For searches and help try: * http://www.stata.com/support/faqs/res/findit.html * http://www.stata.com/support/statalist/faq * http://www.ats.ucla.edu/stat/stata/

Re: st: monte carlo study

2006-03-22T15:01:00.000-05:00

Rudy,

here is an example based on the help file for -simulate-

------------------------- capture program drop mcarlols program define mcarlols, rclass syntax [, obs(integer 1) ] drop _all set obs `obs' drawnorm eps x g y=1+x+eps reg y x return scalar beta = _b[x] end

simulate beta=r(beta) , reps(1000): mcarlols, obs(100)

su beta, de kdensity beta, norm --------------------------------

le 22/03/2006 20:10, Rudy Fichtenbaum a ecrit : > Stata Users: > > I am still learning my way around Stata after many years of using SAS. > In SAS it is fairly easy to write a program to do a simple Monte Carlo > study to illustrate the properties of least squares estimators. > > Is there anyone that has a simple example of a Monte Carlo Study for OLS? > > Thanks, > > Rudy > * > * For searches and help try: > * http://www.stata.com/support/faqs/res/findit.html > * http://www.stata.com/support/statalist/faq > * http://www.ats.ucla.edu/stat/stata/ >

* * For searches and help try: * http://www.stata.com/support/faqs/res/findit.html * http://www.stata.com/support/statalist/faq * http://www.ats.ucla.edu/stat/stata/

Re: st: monte carlo study

2006-03-22T14:54:00.000-05:00

In the unlikely event that you have not, do check out -findit monte carlo-. The example in STB reprints Vol 4, pg 207 may be helpful

TITLE STB-20 ssi6. Simplified Monte Carlo simulations.

Rudy Fichtenbaum wrote: > Stata Users: > > I am still learning my way around Stata after many years of using SAS. > In SAS it is fairly easy to write a program to do a simple Monte Carlo > study to illustrate the properties of least squares estimators. > > Is there anyone that has a simple example of a Monte Carlo Study for OLS? > > Thanks, > > Rudy > * > * For searches and help try: > * http://www.stata.com/support/faqs/res/findit.html > * http://www.stata.com/support/statalist/faq > * http://www.ats.ucla.edu/stat/stata/ > * * For searches and help try: * http://www.stata.com/support/faqs/res/findit.html * http://www.stata.com/support/statalist/faq * http://www.ats.ucla.edu/stat/stata/

st: Chou-Talalay method

2006-03-22T14:16:00.000-05:00

Dear all,

Does anyone have or know of a Stata program to compute the various Chou-Talalay statistics of dose-effect relationship?

Thank you, Ricardo

Ricardo Ovaldia, MS Statistician Oklahoma City, OK

__________________________________________________ Do You Yahoo!? Tired of spam? Yahoo! Mail has the best spam protection around http://mail.yahoo.com * * For searches and help try: * http://www.stata.com/support/faqs/res/findit.html * http://www.stata.com/support/statalist/faq * http://www.ats.ucla.edu/stat/stata/

st: monte carlo study

2006-03-22T14:11:00.000-05:00

Stata Users:

I am still learning my way around Stata after many years of using SAS. In SAS it is fairly easy to write a program to do a simple Monte Carlo study to illustrate the properties of least squares estimators.

Is there anyone that has a simple example of a Monte Carlo Study for OLS?

Thanks,

Rudy * * For searches and help try: * http://www.stata.com/support/faqs/res/findit.html * http://www.stata.com/support/statalist/faq * http://www.ats.ucla.edu/stat/stata/

RE: st: functions for computing prob

2006-03-22T13:58:00.000-05:00

Thanks a lot, Gary!

Lei

-----Original Message----- From: owner-statalist@hsphsun2.harvard.edu [mailto:owner-statalist@hsphsun2.harvard.edu] On Behalf Of Gary Longton Sent: Wednesday, March 22, 2006 12:47 PM To: statalist@hsphsun2.harvard.edu Subject: Re: st: functions for computing prob

It appears that I was mistaken about -tprob()-, that it does exist and appear to work, though undocumented and apparently obsolete. It looks to be the 2-tailed version of -ttail()-

. di tprob(30,1.8) .08192507

. di ttail(30,1.8) .04096253

- GL

Gary Longton wrote: > Lei Xuan wrote: > >> I am computing probabilities for t-test and z-test. >> I want to know if the functions -tprob- and -normprob- are out-of-date >> since no help files explain these functions. Are -ttail(n,t)- and >> -normal(z) >> >> right functions to compute probs? > > > see -help density functions- > or -help functions- > > Yes, the cumulative normal density function, normprob(), still works but > is out of date, no longer documented, and has been replaced by > normal(). Am not sure whether tprob() ever existed? - doesn't seem to > work and is not documented in any case. ttail(n,t) is documented under > help for functions. > > - Gary * * For searches and help try: * http://www.stata.com/support/faqs/res/findit.html * http://www.stata.com/support/statalist/faq * http://www.ats.ucla.edu/stat/stata/

* * For searches and help try: * http://www.stata.com/support/faqs/res/findit.html * http://www.stata.com/support/statalist/faq * http://www.ats.ucla.edu/stat/stata/

Re: st: functions for computing prob

2006-03-22T13:48:00.000-05:00

It appears that I was mistaken about -tprob()-, that it does exist and appear to work, though undocumented and apparently obsolete. It looks to be the 2-tailed version of -ttail()-

. di tprob(30,1.8) .08192507

. di ttail(30,1.8) .04096253

- GL

Re: st: functions for computing prob

2006-03-22T13:32:00.000-05:00

Lei Xuan wrote:

> I am computing probabilities for t-test and z-test. > I want to know if the functions -tprob- and -normprob- are out-of-date > since no help files explain these functions. Are -ttail(n,t)- and -normal(z) > > right functions to compute probs?

see -help density functions- or -help functions-

Yes, the cumulative normal density function, normprob(), still works but is out of date, no longer documented, and has been replaced by normal(). Am not sure whether tprob() ever existed? - doesn't seem to work and is not documented in any case. ttail(n,t) is documented under help for functions.

- Gary

* * For searches and help try: * http://www.stata.com/support/faqs/res/findit.html * http://www.stata.com/support/statalist/faq * http://www.ats.ucla.edu/stat/stata/

st: functions for computing prob

2006-03-22T13:16:00.000-05:00

Hi,

I am computing probabilities for t-test and z-test. I want to know if the functions -tprob- and -normprob- are out-of-date since no help files explain these functions. Are -ttail(n,t)- and -normal(z)

right functions to compute probs?

Thanks,

Lei Xuan

* * For searches and help try: * http://www.stata.com/support/faqs/res/findit.html * http://www.stata.com/support/statalist/faq * http://www.ats.ucla.edu/stat/stata/

Re: st: STATA syntax colouring in TextWrangler

2006-03-22T12:48:00.000-05:00

Cool! Works w/ BBEdit, too. You just have to create a "Languages Module" folder.

See this as well: http://dataninja.wordpress.com/2006/03/03/send-to-stata-applescript- for-textwrangler/

-- Danielle H Ferry

On Mar 22, 2006, at 9:12 AM, Ronán Conroy wrote:

> On 22 Márta 2006, at 12:47, Taavi Lai wrote: > >> Google search "TextWrangler stata syntax" gave several hits and >> one of those directed to a file Stata.plist which downloaded ok. >> I'm not a Mac user and have no acquaintance to TextWrangler so you >> have to experiment further >> the link is http://dataninja.wordpress.com/2006/02/28/stata- >> language-module-for-textwrangler/ >> >> Regards, >> Taavi > > > Splended hunting. I've installed it and it works fine, except that > it fails to recognise the single open quote correctly. You need to > fix these lines > > <key>Open Strings 2</key> > <string></string> > > to read > <key>Open Strings 2</key> > <string>`</string> > > > > I have also made a couple of personal tweaks: it now recognises all > SSC packages. I've emailed this version to DataNinja so hopefully > it will be posted on his/her site. > > Ronán Conroy > rconroy@rcsi.ie > > > > > * > * For searches and help try: > * http://www.stata.com/support/faqs/res/findit.html > * http://www.stata.com/support/statalist/faq > * http://www.ats.ucla.edu/stat/stata/

* * For searches and help try: * http://www.stata.com/support/faqs/res/findit.html * http://www.stata.com/support/statalist/faq * http://www.ats.ucla.edu/stat/stata/

Re: st: GOOGLE ANSWERS FORWARDED

2006-03-22T12:27:00.000-05:00

I could not agree more--in this case, it seems as though someone is charging for help that they are soliciting from a free source, and http://answers.google.com/answers/ says "Researchers are ready to answer your question for as little as $2.50 -- usually within 24 hours" which I suppose is why "It's pretty urgent."

This is at least as objectionable as students seeking help on graded problem sets.

On 3/22/06, n j cox <n.j.cox@durham.ac.uk> wrote: > What is this? > > I propose three simple principles: > > 1. If someone wants to join Statalist and post to it, they should > feel welcome, and they should read the FAQ to see how we operate. > Nothing new there. > > 2. If someone wants to post something to Statalist on behalf of someone > else, that's OK so long as they explicitly undertake to answer > subsidiary questions and forward the answers. > > 3. Otherwise I see zero point in answering questions like this, as > we have no assurances of an answer being seen, or "someone" or "Martin" > answering any questions it raises. The appeal to urgency is also > objectionable. > > Nick > n.j.cox@durham.ac.uk > > -------------------------------------------------------------------------------- > From: noah_kauffman@prusec.com > Subject: st: GOOGLE ANSWERS FORWARDED > Date: Wed, 22 Mar 2006 10:34:11 -0500 > > Someone @ google answers asks: > > > Ok, I'm trying to run a > fixed-effects panel in stata. > My regression: (where var1 is the > depend. variable) > > (PLEASE SEE DATA AT END) - Data > pasted from excel into stata via > editor > Stata commands were as follows: > > "tsset id year" > "gen var1 = ln(x3)" > "gen var2 = ln(var 1[_n-1])" ie. > require ln(x3 i,t-1) > "gen var3 = ln(x2)" > "gen var4 = ln(5 +x1)" > > xtreg var1 var2 var3 var4, i(id), fe > > Now for the coefficient on var2 I > get 0.177, and I'm looking for > something around the 0.5 region. > I've gone wrong somewhere - can > someone please help??? > It's pretty urgent as I need this by > Thursday morning at latest. If it > helps, I'm trying to estiamte the > model on page 1146 for OECD > countries, (but with my own data) in > Islam's paper: "Growth Empirics: > A Panel Data Approach" (1995) > I'm online all the time, so if > necessarty, I can clarify things > beforehand. > > <zap>

* * For searches and help try: * http://www.stata.com/support/faqs/res/findit.html * http://www.stata.com/support/statalist/faq * http://www.ats.ucla.edu/stat/stata/

RE: st: Naming convention, Ideas?

2006-03-22T11:21:00.000-05:00

Thank you Nick(s)! I hadn't thought of the colon approach or the major minor subcommand route. And one module certainly *is* easier to maintain, document, and support.

I am glad I posted for a query I thought might seem trivial.

Nick Winter > Maybe something that works off the notion of "generalized." > > One option would be to use the prefix approach, to create a > syntax like: > > . genmanip : merge ... > > . genmanip : append ... > > and so on. Then you have only one .ado file to maintain, easily > allowing options that apply to your command (distinct from the > append, merge, etc. options), etc. > > See -help _on_colon_parse- for a Stata command that helps > parsing that syntax. > > I'm not sure -genmanip- is a great name, but something like that? > > --Nick Winter > > > > At 10:10 AM 3/22/2006, you wrote: > >I am looking into writing a suite of wrapper data management > >commands around merge, mmerge, append, joinby, and cross that can > >either take a stata data file, gzip compressed data file or simply a > >comma or tab delimited text file as the -using- argument, e.g. > ><cmd_name> using *.dta | *.dta.gz | *.dgz | *.txt | *.cvs [, * ]. > > > >Two questions: > >1) Any ideas w/ regard to a consistent naming convention that could > >be used? as I'd like to get it right the first time. I am not very > >fond of using an integer as a suffix a la cf2, cf3 for various > >reasons (e.g not very informative, unclear if integers imply > >incremental functionality, can conflict with others' names). So far > >I thought of: > > > >- mmergeplus, appendplus, joinbyplus (but rather long) > >- aappend, jjoinby, (but look like typos, besides mmerge > already exists) > > > >2) Would anyone find these useful, i.e. should they be posted on SSC?

* * For searches and help try: * http://www.stata.com/support/faqs/res/findit.html * http://www.stata.com/support/statalist/faq * http://www.ats.ucla.edu/stat/stata/

Re: st: Mata function stata() within program

2006-03-22T11:02:00.000-05:00

Many thanks for this. I will carefully look at it. It seems that this will make the program much faster, which would be _very_ important. The snippet will often run more than a million times ...

William Gould, Stata wrote: > Ulrich Kohler <kohler@wz-berlin.de> wrote, > > > I have a Mata function which looks as follows: > > > > -------------------------------------lsq.mata-- > > (...) > > // Mata Function to extract the substitution costs from subcost-matrix > > void showhash(real rowvector R) > > { > > string scalar key1 > > st_local("key1",key1) > > key1 = strofreal(R[1,2]) > > st_local("key1",key1) > > stata("local hash1 = mod(`key1',197)") > > } > > > > (...) > >------------------------------------------------ > > and Uli notes that when he runs it, he gets an error, > > > : R = 2,3,5,4 > > : showhash(R) > > > > invalid syntax > > stata(): 3598 Stata returned error > > showhash(): - function returned error > > <istmt>: - function returned error > > r(3598); > > Alan Riley <ariley@stata.com> has already given a solution, and suggested > the line > > stata("local hash1 = mod(`key1',197)") > > be changed to read > > stata("local hash1 = mod(\`key1',197)") > > Alan's right, but his solution is too tricky for me. Moreover, his > solution shows he is still thinking an ado mode rather than Mata mode. > > My suggested solution is > > stata("local hash1 = mod(" + key1 + ", 197)") > > and, with my solution, Uli's code can be simplified to read, > > > void showhash(real rowvector R) > { > string scalar key1 > > key1 = strofreal(R[1,2]) > stata("local hash1 = mod(" + key1 + ", 197)") > } > > or even > > void showhash(real rowvector R) > { > stata("local hash1 = mod(" + strofreal(R[1,2]) + ", > 197)") } > > Let me expound on the ado versus the Mata way of thinking. > > Uli wanted to run the Stata command > > local hash1 = mod(_______, 197) > > where he substituted the value from Mata matrix R[1,2] for ______. > Why Uli wanted to do this, we don't know, nor care. > > The ado way of thinking says we substitute a macro for _____, and arrange > for the macro to contain R[1,2], so when the macro is substituted by > Stata, we obtrain the desired result. Good way of thinking, when you > are writing an ado-file. > > The Mata way of thinking is more direct: we need to construct a string > > "local hash1 = mod(_______, 197)" > > where where R[1,2] is substitued for _____, and we can just use the > standard operators to do that, > > "local hash1 = mod(" + strofreal(R[1,2]) + ", 197)" > > Here's a good rule: It's perfectly okay to obtain input from macros, or > post output to macros. That is one way Mata can communicate with > ado-files. If you have to use macros to obtain your result, however, you > are thinking ado, not Mata. There's a simpler, more direct way. > > -- Bill > wgould@stata.com > * > * For searches and help try: > * http://www.stata.com/support/faqs/res/findit.html > * http://www.stata.com/support/statalist/faq > * http://www.ats.ucla.edu/stat/stata/

-- kohler@wz-berlin.de +49 (030) 25491-361 * * For searches and help try: * http://www.stata.com/support/faqs/res/findit.html * http://www.stata.com/support/statalist/faq * http://www.ats.ucla.edu/stat/stata/

Re: st: GOOGLE ANSWERS FORWARDED

2006-03-22T10:58:00.000-05:00

What is this?

I propose three simple principles:

1. If someone wants to join Statalist and post to it, they should feel welcome, and they should read the FAQ to see how we operate. Nothing new there.

2. If someone wants to post something to Statalist on behalf of someone else, that's OK so long as they explicitly undertake to answer subsidiary questions and forward the answers.

3. Otherwise I see zero point in answering questions like this, as we have no assurances of an answer being seen, or "someone" or "Martin" answering any questions it raises. The appeal to urgency is also objectionable.

Nick n.j.cox@durham.ac.uk

-------------------------------------------------------------------------------- From: noah_kauffman@prusec.com Subject: st: GOOGLE ANSWERS FORWARDED Date: Wed, 22 Mar 2006 10:34:11 -0500

Someone @ google answers asks:

Ok, I'm trying to run a fixed-effects panel in stata. My regression: (where var1 is the depend. variable)

(PLEASE SEE DATA AT END) - Data pasted from excel into stata via editor Stata commands were as follows:

"tsset id year" "gen var1 = ln(x3)" "gen var2 = ln(var 1[_n-1])" ie. require ln(x3 i,t-1) "gen var3 = ln(x2)" "gen var4 = ln(5 +x1)"

xtreg var1 var2 var3 var4, i(id), fe

Now for the coefficient on var2 I get 0.177, and I'm looking for something around the 0.5 region. I've gone wrong somewhere - can someone please help??? It's pretty urgent as I need this by Thursday morning at latest. If it helps, I'm trying to estiamte the model on page 1146 for OECD countries, (but with my own data) in Islam's paper: "Growth Empirics: A Panel Data Approach" (1995) I'm online all the time, so if necessarty, I can clarify things beforehand.

<zap>

* * For searches and help try: * http://www.stata.com/support/faqs/res/findit.html * http://www.stata.com/support/statalist/faq * http://www.ats.ucla.edu/stat/stata/

st: GOOGLE ANSWERS FORWARDED

2006-03-22T10:46:00.000-05:00

Someone @ google answers asks:

Ok, I'm trying to run a fixed-effects panel in stata. My regression: (where var1 is the depend. variable)

(PLEASE SEE DATA AT END) - Data pasted from excel into stata via editor Stata commands were as follows:

"tsset id year" "gen var1 = ln(x3)" "gen var2 = ln(var 1[_n-1])" ie. require ln(x3 i,t-1) "gen var3 = ln(x2)" "gen var4 = ln(5 +x1)"

xtreg var1 var2 var3 var4, i(id), fe

Thanks for your help,

Martin id Year x1 x2 x3 69 1980 1.662461763 14.98616639 17366.351 69 1985 1.751607907 11.13300528 16621.06196 69 1990 1.687373049 9.304076135 27174.53416 69 1995 1.145010297 11.69968344 30926.21307 69 2000 1.287333281 13.37070116 30192.66929 70 1980 0.822325571 19.58621955 16393.97505 70 1985 0.810340975 15.2813291 13144.98494 70 1990 0.524329844 13.18757172 30948.48975 70 1995 0.535188192 16.75931281 43827.01387 70 2000 0.210425117 15.7938949 35069.97475 71 1980 0.602471123 21.51876931 18890.07907 71 1985 0.450791288 18.81103302 12464.20844 71 1990 0.109973609 17.44272923 29622.43449 71 1995 0.188184945 16.84773807 41116.6454 71 2000 0.143463565 16.66518245 33698.04446 72 1980 1.793759025 13.47103394 15883.10011 72 1985 1.13896003 10.38388336 19731.97361 72 1990 1.329878954 10.02094259 30365.61038 72 1995 1.011132436 10.51075658 29259.80686 72 2000 1.090468632 12.83921854 33940.05624 73 1980 0.477478563 15.80453557 20635.75266 73 1985 0.474309767 14.65851051 17595.00342 73 1990 0.33966525 10.00934545 38504.07008 73 1995 0.395259882 11.23964728 51179.45169 73 2000 0.16514965 7.732210879 44379.18977 74 1980 0.423406049 20.10184597 16151.36943 74 1985 0.533425877 17.03576594 16286.31437 74 1990 0.17987329 15.98985483 40788.69617 74 1995 0.318379132 14.12707098 38004.6729 74 2000 0.300608318 13.51151556 34586.6218 75 1980 0.893883928 22.8138795 19857.92277 75 1985 0.961587541 19.68265195 14615.15092 75 1990 0.522867504 19.65530273 32582.27016 75 1995 0.285397244 15.93692676 41063.60444 75 2000 0.236410263 16.32833599 34134.14132 76 1980 1.32316342 20.73041017 7904.237485 76 1985 0.922498984 19.56876374 6319.746039 76 1990 0.995140916 21.41446384 12340.93113 76 1995 0.903410832 16.741426 16436.26653 76 2000 0.442745708 17.34889112 15296.49573 77 1980 1.6190352 14.66842754 10467.50384 77 1985 0.984051537 13.11345854 9644.005897 77 1990 0.52640211 13.47451789 22004.14509 77 1995 1.573243946 15.01458907 28639.96061 77 2000 2.010459298 16.43950604 37062.08373 78 1980 0.68755784 18.83731856 12316.87681 78 1985 0.85343638 17.14138777 11127.02382 78 1990 0.326057221 19.51414315 28247.96617 78 1995 0.200326573 21.82167717 27807.58652 78 2000 -0.29931545 17.18042319 27566.33909 79 1980 0.801608658 22.79975178 13501.75246 79 1985 0.915264718 18.50416861 16474.29537 79 1990 0.723636851 19.17730187 35344.51147 79 1995 0.19249092 22.90417495 60613.2475 79 2000 -0.145385709 19.08132976 54827.36543 80 1980 1.387616421 14.94607015 19026.16294 80 1985 1.025591315 11.06662356 13313.43306 80 1990 0.75005838 9.991219054 28602.29705 80 1995 0.432385881 15.01848763 39256.06623 80 2000 0.500530203 16.40487928 34288.62992 81 1980 0.828691467 20.71323278 11601.11144 81 1985 1.311964138 20.45743036 10716.73514 81 1990 1.3691672 18.28935167 19337.00316 81 1995 1.352106185 19.59779156 25288.2837 81 2000 0.900292987 20.21200377 20582.89207 82 1980 0.587584723 15.61925349 24668.24692 82 1985 0.682825353 11.2998007 23825.56852 82 1990 0.543734919 9.963804811 42244.72488 82 1995 0.525269517 11.79609955 52537.70169 82 2000 0.674729984 16.59774348 57349.30306 83 1980 1.462021324 19.39049446 4808.80038 83 1985 0.692344139 10.83176142 3804.655031 83 1990 0.24932822 14.84951396 10872.80679 83 1995 0.165472127 18.03693854 16241.68266 83 2000 1.032193538 18.57355586 15384.17332 84 1980 1.224982892 17.50583952 9429.244045 84 1985 1.067391322 14.1583388 6892.579016 84 1990 0.755676566 12.92168201 19659.18951 84 1995 0.591949849 11.73567215 21807.56001 84 2000 0.647429503 10.58827526 20354.88874 85 1980 0.255875886 18.00839687 24331.0159 85 1985 0.220629595 16.16479109 19377.70035 85 1990 0.47660532 15.96377273 43624.07581 85 1995 0.412345421 15.32120281 44019.86808 85 2000 0.273147601 14.41038065 41946.34968 86 1980 0.496489298 15.48425823 26419.38091 86 1985 0.941437445 13.88066924 22534.61193 86 1990 0.952530759 14.10603859 51114.72784 86 1995 0.641787638 14.00834142 65537.35073 86 2000 0.224554799 17.41319943 50647.46892 87 1980 2.612506141 17.82209964 2843.972229 87 1985 3.50641537 15.35851228 2251.386322 87 1990 2.741307328 16.79774136 4417.057814 87 1995 2.819981075 18.63495028 4297.528683 87 2000 2.376735812 13.4780134 4493.247857 88 1980 0.484872291 23.73434618 14863.17762 88 1985 0.544248613 21.78829843 12241.47715 88 1990 0.171420325 23.40476948 26392.5887 88 1995 0.177399289 17.48076441 30017.22473 88 2000 0.2890547 19.05241986 37523.02325 89 1980 1.510727766 16.39963029 18387.85317 89 1985 0.907081959 12.01908341 26494.88105 89 1990 0.845319833 12.206305 35104.42998 89 1995 1.224515978 14.20890757 42224.98562 89 2000 1.382534638 12.96683686 52391.32554

* * For searches and help try: * http://www.stata.com/support/faqs/res/findit.html * http://www.stata.com/support/statalist/faq * http://www.ats.ucla.edu/stat/stata/

Re: st: Naming convention, Ideas?

2006-03-22T10:38:00.000-05:00

Maybe something that works off the notion of "generalized."

One option would be to use the prefix approach, to create a syntax like:

. genmanip : merge ...

. genmanip : append ...

and so on. Then you have only one .ado file to maintain, easily allowing options that apply to your command (distinct from the append, merge, etc. options), etc.

See -help _on_colon_parse- for a Stata command that helps parsing that syntax.

I'm not sure -genmanip- is a great name, but something like that?

--Nick Winter

At 10:10 AM 3/22/2006, you wrote: >I am looking into writing a suite of wrapper data management >commands around merge, mmerge, append, joinby, and cross that can >either take a stata data file, gzip compressed data file or simply a >comma or tab delimited text file as the -using- argument, e.g. ><cmd_name> using *.dta | *.dta.gz | *.dgz | *.txt | *.cvs [, * ]. > >Two questions: >1) Any ideas w/ regard to a consistent naming convention that could >be used? as I'd like to get it right the first time. I am not very >fond of using an integer as a suffix a la cf2, cf3 for various >reasons (e.g not very informative, unclear if integers imply >incremental functionality, can conflict with others' names). So far >I thought of: > >- mmergeplus, appendplus, joinbyplus (but rather long) >- aappend, jjoinby, (but look like typos, besides mmerge already exists) > >2) Would anyone find these useful, i.e. should they be posted on SSC? > > >Patrick Joly > >* >* For searches and help try: >* http://www.stata.com/support/faqs/res/findit.html >* http://www.stata.com/support/statalist/faq >* http://www.ats.ucla.edu/stat/stata/

________________________________________________________ Nicholas J. G. Winter 607.255.8819 t Assistant Professor 607.255.4530 f Department of Government nw53@cornell.edu e Cornell University falcon.arts.cornell.edu/nw53 w 308 White Hall Ithaca, NY 14853-4601

* * For searches and help try: * http://www.stata.com/support/faqs/res/findit.html * http://www.stata.com/support/statalist/faq * http://www.ats.ucla.edu/stat/stata/

Re: st: Naming convention, Ideas?

2006-03-22T10:37:00.000-05:00

My suggestion is none of these. Find a suitable name, say -pjcombine-, and then write a command with subcommands, so that your syntax is

pjcombine append <whatever> pjcombine merge <whatever> ...

Reasoning:

0. There are many precedents in Stata itself.

1. One program is easier to keep track of than several.

2. -combine- is a good word, but StataCorp lay claim to all the words in the English language. Some of us have forgotten that in the past -- "accidentally on purpose", perhaps -- and you might choose to forget that too. At worst, StataCorp may grab "your" program name and either your program has been superseded, or you need to change the name.

3. -pj- would be both modest and good PR.

Put them on SSC!

P.S. on integers: there are two conventions in use. Typically low integers (esp. 2, 3) mean versions of the command; high integers (esp. 5, 6, 7, 8) mean versions of Stata it works with. In practice that is less confusing than it seems, as although many programs go through much revision, programmers don't change names that often.

If <program><n> means "works with Stata <n>", then I suggest that should always be explicit in its help.

Nick n.j.cox@durham.ac.uk

Joly.Patrick

I am looking into writing a suite of wrapper data management commands around merge, mmerge, append, joinby, and cross that can either take a stata data file, gzip compressed data file or simply a comma or tab delimited text file as the -using- argument, e.g. <cmd_name> using *.dta | *.dta.gz | *.dgz | *.txt | *.cvs [, * ].

Two questions: 1) Any ideas w/ regard to a consistent naming convention that could be used? as I'd like to get it right the first time. I am not very fond of using an integer as a suffix a la cf2, cf3 for various reasons (e.g not very informative, unclear if integers imply incremental functionality, can conflict with others' names). So far I thought of:

- mmergeplus, appendplus, joinbyplus (but rather long) - aappend, jjoinby, (but look like typos, besides mmerge already exists)

2) Would anyone find these useful, i.e. should they be posted on SSC?

* * For searches and help try: * http://www.stata.com/support/faqs/res/findit.html * http://www.stata.com/support/statalist/faq * http://www.ats.ucla.edu/stat/stata/

st: RE: left-truncation of entry in survival analysis

2006-03-22T10:33:00.000-05:00

I am assuming that the covariates in the 2 models are all time-independent. In general, the 2 models are not equivalent. However, it is entirely possible that they might give the same parameter estimates, in at least some specific sets of data.

Survival analyses (as their name suggests) are based on events whereby one subject is observed to survive another, ie one subject at risk on a particular day survives to the end of that day and another subject at risk on the same day is dead by the end of that day. The two models are different in what is meant by "the same day" for the two subjects. In the first model, we compare the fate of Subject A on Day X of the life of Subject A with the fate of Subject B on Day X of the life of Subject B, for all pairs of Subjects A and B who were both under observation in the study on Day X of their respective lives. In the second model, we compare the fate of Subject A on Day Y of Subject A's study time (measured from Subject A's entry into the study) with the fate of Subject B on Day Y of Subject B's study time (measured from Subject B's entry into the study), for all pairs of Subjects A and B who were both under observation on Day Y of their respective study time windows.

In a specific study, it might be the case that, for each Subject A who died on Day X of his/her life and Day Y of his/her study time, the set of Subjects B who survived through the Days X of their respective lives in the study might be the same set as the set of Subjects B who survived through the Days Y of their respective study times in the study. This might especially be the case if the number of subjects is small and/or deaths in the study are sparse. For such a specific study, the two Cox regressions will give the same parameter estimates. However, this will not be the case for all studies. For instance, in some studies, there will be pairs of Subjects A and B, such that Subject A dies in the study at 100 years of age after having entered the study at 99 years of age, whereas Subject B dies in the study at 40 years of age after having entered the study at 30 years of age. In this case, the first model will assume that neither patient was observed to survive the other, whereas the second model will assume that Subject B has survived Subject A, even though Subject B died younger.

I hope this helps.

Roger

Roger Newson Lecturer in Medical Statistics POSTAL ADDRESS: Respiratory Epidemiology and Public Health Group National Heart and Lung Institute at Imperial College London St Mary's Campus Norfolk Place London W2 1PG STREET ADDRESS: Respiratory Epidemiology and Public Health Group National Heart and Lung Institute at Imperial College London 47 Praed Street Paddington London W1 1NR TELEPHONE: (+44) 020 7594 0939 FAX: (+44) 020 7594 0942 EMAIL: r.newson@imperial.ac.uk WEBSITE: http://www.imperial.ac.uk/nhli/r.newson/ Opinions expressed are those of the author, not of the institution.

-----Original Message----- From: owner-statalist@hsphsun2.harvard.edu [mailto:owner-statalist@hsphsun2.harvard.edu] On Behalf Of Sue Chinn Sent: 22 March 2006 12:47 To: statalist@hsphsun2.harvard.edu Subject: st: left-truncation of entry in survival analysis

Dear Statalist readers,

Reports of survival analysis which use age as the time scale rather than

time-on-study often 'adjust for delayed entry'. In Stata this is achieved by:

stset age, fail(died) enter(ageatentry)

(see recent e-mail from Dawn Teele, or reply to st: streg from rgutierrez@stata.com on 19th September 2002.)

However, a model fitted with the above stset gives exactly the same answer as one with

stset timeonstudy, fail(died)

provided timeonstudy=age-ageatentry (as it normally would, but might not

exactly depending how variables were calculated) and the models are exactly the same. In the second model it is usual to adjust or stratify on age, while in the first it isn't as age is taken into account, supposedly, so

researchers may not have realised the equivalence.

So, am I missing something, or are advocates of the first model deluding

themselves? Can left truncation be ignored with age as the timescale?

Thanks

Sue

Sue Chinn Professor of Medical Statistics Division of Asthma, Allergy and Lung Biology King's College London 5th Floor Capital House 42 Weston Street London SE1 3QD

tel no. 020 7848 6607 fax no. 020 7848 6605

* * For searches and help try: * http://www.stata.com/support/faqs/res/findit.html * http://www.stata.com/support/statalist/faq * http://www.ats.ucla.edu/stat/stata/

st: Naming convention, Ideas?

2006-03-22T10:12:00.000-05:00

- mmergeplus, appendplus, joinbyplus (but rather long) - aappend, jjoinby, (but look like typos, besides mmerge already exists)

2) Would anyone find these useful, i.e. should they be posted on SSC?

Patrick Joly

* * For searches and help try: * http://www.stata.com/support/faqs/res/findit.html * http://www.stata.com/support/statalist/faq * http://www.ats.ucla.edu/stat/stata/

Re: st: Mata function stata() within program

2006-03-22T10:01:00.000-05:00

Ulrich Kohler <kohler@wz-berlin.de> wrote,

> I have a Mata function which looks as follows: > > -------------------------------------lsq.mata-- > (...) > // Mata Function to extract the substitution costs from subcost-matrix > void showhash(real rowvector R) > { > string scalar key1 > st_local("key1",key1) > key1 = strofreal(R[1,2]) > st_local("key1",key1) > stata("local hash1 = mod(`key1',197)") > } > > (...) >------------------------------------------------

and Uli notes that when he runs it, he gets an error,

> : R = 2,3,5,4 > : showhash(R) > invalid syntax > stata(): 3598 Stata returned error > showhash(): - function returned error > <istmt>: - function returned error > r(3598);

Alan Riley <ariley@stata.com> has already given a solution, and suggested the line

stata("local hash1 = mod(`key1',197)")

be changed to read

stata("local hash1 = mod(\`key1',197)")

Alan's right, but his solution is too tricky for me. Moreover, his solution shows he is still thinking an ado mode rather than Mata mode.

My suggested solution is

stata("local hash1 = mod(" + key1 + ", 197)")

and, with my solution, Uli's code can be simplified to read,

void showhash(real rowvector R) { string scalar key1

key1 = strofreal(R[1,2]) stata("local hash1 = mod(" + key1 + ", 197)") }

or even

void showhash(real rowvector R) { stata("local hash1 = mod(" + strofreal(R[1,2]) + ", 197)") }

Let me expound on the ado versus the Mata way of thinking.

Uli wanted to run the Stata command

local hash1 = mod(_______, 197)

where he substituted the value from Mata matrix R[1,2] for ______. Why Uli wanted to do this, we don't know, nor care.

The ado way of thinking says we substitute a macro for _____, and arrange for the macro to contain R[1,2], so when the macro is substituted by Stata, we obtrain the desired result. Good way of thinking, when you are writing an ado-file.

The Mata way of thinking is more direct: we need to construct a string

"local hash1 = mod(_______, 197)"

where where R[1,2] is substitued for _____, and we can just use the standard operators to do that,

"local hash1 = mod(" + strofreal(R[1,2]) + ", 197)"

Here's a good rule: It's perfectly okay to obtain input from macros, or post output to macros. That is one way Mata can communicate with ado-files. If you have to use macros to obtain your result, however, you are thinking ado, not Mata. There's a simpler, more direct way.

-- Bill wgould@stata.com * * For searches and help try: * http://www.stata.com/support/faqs/res/findit.html * http://www.stata.com/support/statalist/faq * http://www.ats.ucla.edu/stat/stata/

st: group mean pooling VAR

2006-03-22T09:42:00.000-05:00

Dear statalist,
I'm trying to do group mean pooling of VAR regression results across an unbalanced panel (id1, year). When I try to take the first step, creating the pool of coefficients, I used statsby, but Stata tells me I haven't tsset. I get this:

. tsset id1 year
panel variable: id1, 1 to 208
time variable: year, 1966 to 2002, but with a gap

. statsby _b , by(id1) saving(varDT): varbasic ICRGliberal DirTax year, lags(1/
> 2)
(running varbasic on estimation sample)
time variable not set, use -tsset varname ...-
an error occurred when statsby executed varbasic
r(111);

Am I doing something wrong, and/or has someone on the list written a good routine for group mean pooling of VAR regressions across a panel?
Thanks.
Jim Mahon

Re: st: left-truncation of entry in survival analysis

2006-03-22T09:32:00.000-05:00

On 22 Márta 2006, at 12:46, Sue Chinn wrote:

> > So, am I missing something, or are advocates of the first model > deluding themselves? Can left truncation be ignored with age as the > timescale?

One powerful argument for using age as the time variable in survival analysis has to do with the hazard function itself. In modelling risk factors or prognostic factors, the shape of the hazard function is often treated as a high-dimensional nuisance parameter (I love that expression and use it as often as I can). However, in risk estimation the hazard function represents the absolute risk, without which the hazard ratios cannot be interpreted in real-life terms.

Using age as the time variable allows the construction of hazard functions spanning the age range from the youngest entry to the oldest exit. It allowed us, in the SCORE project, to make cardiovascular risk charts showing ten-year risk from datasets many of which contained at most 8 years of follow up.

Since the publication of Ed Korn's paper, there have been a number of papers looking to see if the modelling of risk factor effects is affected by the choice of time metric and the short answer seems to be 'not so you would notice'. I haven't been keeping up with the literature on this, but I don't know of a case where someone got very different results - does anyone else?

Korn's proposal appeals to my sense of elegance, as well as being a gift to those of us who have to deal with absolute risks.

Ronán Conroy rconroy@rcsi.ie

* * For searches and help try: * http://www.stata.com/support/faqs/res/findit.html * http://www.stata.com/support/statalist/faq * http://www.ats.ucla.edu/stat/stata/

Re: st: STATA syntax colouring in TextWrangler

2006-03-22T09:15:00.001-05:00

On 22 Márta 2006, at 12:47, Taavi Lai wrote:

> Google search "TextWrangler stata syntax" gave several hits and one > of those directed to a file Stata.plist which downloaded ok. I'm > not a Mac user and have no acquaintance to TextWrangler so you have > to experiment further > the link is http://dataninja.wordpress.com/2006/02/28/stata- > language-module-for-textwrangler/ > > Regards, > Taavi

Splended hunting. I've installed it and it works fine, except that it fails to recognise the single open quote correctly. You need to fix these lines

<key>Open Strings 2</key> <string></string>

to read <key>Open Strings 2</key> <string>`</string>

I have also made a couple of personal tweaks: it now recognises all SSC packages. I've emailed this version to DataNinja so hopefully it will be posted on his/her site.

Ronán Conroy rconroy@rcsi.ie

* * For searches and help try: * http://www.stata.com/support/faqs/res/findit.html * http://www.stata.com/support/statalist/faq * http://www.ats.ucla.edu/stat/stata/

Re: st: STATA syntax colouring in TextWrangler

2006-03-22T09:15:00.000-05:00

On 22 Márta 2006, at 12:47, Taavi Lai wrote:

Splended hunting. I've installed it and it works fine, except that it fails to recognise the single open quote correctly. You need to fix these lines

<key>Open Strings 2</key> <string></string>

to read <key>Open Strings 2</key> <string>`</string>

I have also made a couple of personal tweaks: it now recognises all SSC packages. I've emailed this version to DataNinja so hopefully it will be posted on his/her site.

Ronán Conroy rconroy@rcsi.ie

* * For searches and help try: * http://www.stata.com/support/faqs/res/findit.html * http://www.stata.com/support/statalist/faq * http://www.ats.ucla.edu/stat/stata/

st: RE: problem with the F test

2006-03-22T08:42:00.000-05:00

If you do this interactively, or look at a log in scml format, the missing F-stat should be in blue, indicating a hyperlink. If so, you can click on it to get a description of possible causes for the missing statistic.

HTH.

--Mark

> -----Original Message----- > From: owner-statalist@hsphsun2.harvard.edu > [mailto:owner-statalist@hsphsun2.harvard.edu] On Behalf Of > mduque@cedes.org > Sent: 21 March 2006 23:28 > To: statalist@hsphsun2.harvard.edu > Subject: st: problem with the F test > > I have a problem with the F test when I select the option > ROBUST for the estimation . > > I'm just running OLS regressions and when I select "robust" > for the estimation of the variances the value of the F test > is missing ! (I have 72 obs and 6 regresors) > > If I don't select that option then I don't have the problem > in the same estimation. > > What can it be? > > > * > * For searches and help try: > * http://www.stata.com/support/faqs/res/findit.html > * http://www.stata.com/support/statalist/faq > * http://www.ats.ucla.edu/stat/stata/ > >

* * For searches and help try: * http://www.stata.com/support/faqs/res/findit.html * http://www.stata.com/support/statalist/faq * http://www.ats.ucla.edu/stat/stata/

Re: st: problem with the F test

2006-03-22T08:41:00.000-05:00

been there recently (or close)... -whelp j_robustsingular- might helpful

mduque@cedes.org wrote: > I have a problem with the F test when I select the option ROBUST for the estimation . > > I'm just running OLS regressions and when I select "robust" for the estimation of the variances the value of the F test is missing ! (I have 72 obs and 6 regresors) > > If I don't select that option then I don't have the problem in the same estimation. > > What can it be? > > > * > * For searches and help try: > * http://www.stata.com/support/faqs/res/findit.html > * http://www.stata.com/support/statalist/faq > * http://www.ats.ucla.edu/stat/stata/ > * * For searches and help try: * http://www.stata.com/support/faqs/res/findit.html * http://www.stata.com/support/statalist/faq * http://www.ats.ucla.edu/stat/stata/

Re: st: STATA syntax colouring in TextWrangler

2006-03-22T08:37:00.000-05:00

On Mar 22, 2006, at 1:47 PM, Taavi Lai wrote:

> the link is http://dataninja.wordpress.com/2006/02/28/stata- > language-module-for-textwrangler/

Thank you - it worked perfectly.

Steffen * * For searches and help try: * http://www.stata.com/support/faqs/res/findit.html * http://www.stata.com/support/statalist/faq * http://www.ats.ucla.edu/stat/stata/

st: left-truncation of entry in survival analysis

2006-03-22T07:49:00.001-05:00

Dear Statalist readers,

Reports of survival analysis which use age as the time scale rather than time-on-study often 'adjust for delayed entry'. In Stata this is achieved by:

stset age, fail(died) enter(ageatentry)

(see recent e-mail from Dawn Teele, or reply to st: streg from rgutierrez@stata.com on 19th September 2002.)

However, a model fitted with the above stset gives exactly the same answer as one with

stset timeonstudy, fail(died)

provided timeonstudy=age-ageatentry (as it normally would, but might not exactly depending how variables were calculated) and the models are exactly the same. In the second model it is usual to adjust or stratify on age, while in the first it isn't as age is taken into account, supposedly, so researchers may not have realised the equivalence.

So, am I missing something, or are advocates of the first model deluding themselves? Can left truncation be ignored with age as the timescale?

Thanks

Sue

Sue Chinn Professor of Medical Statistics Division of Asthma, Allergy and Lung Biology King's College London 5th Floor Capital House 42 Weston Street London SE1 3QD

tel no. 020 7848 6607 fax no. 020 7848 6605

* * For searches and help try: * http://www.stata.com/support/faqs/res/findit.html * http://www.stata.com/support/statalist/faq * http://www.ats.ucla.edu/stat/stata/

Re: st: STATA syntax colouring in TextWrangler

2006-03-22T07:49:00.000-05:00

Google search "TextWrangler stata syntax" gave several hits and one of those directed to a file Stata.plist which downloaded ok. I'm not a Mac user and have no acquaintance to TextWrangler so you have to experiment further the link is http://dataninja.wordpress.com/2006/02/28/stata-language-module-for-textwrangler/

Regards, Taavi

Steffen Hokland wrote: > I've started to do STATA programming, and have decided to use my > 'normal' text editor - TextWrangler (Apple Mac OS-X). I've been able > to track down a syntax colouring module for TextWrangler (by Ben > Hulley), but all links refering to it (e.g. > homepage.mac.com/benhulley) seem to be dead. Can anyone direct me to a > module for syntax colouring in TextWrangler (or another texteditor for > that matter)? > > Steffen > * > * For searches and help try: > * http://www.stata.com/support/faqs/res/findit.html > * http://www.stata.com/support/statalist/faq > * http://www.ats.ucla.edu/stat/stata/ * * For searches and help try: * http://www.stata.com/support/faqs/res/findit.html * http://www.stata.com/support/statalist/faq * http://www.ats.ucla.edu/stat/stata/

st: problem with the F test

2006-03-22T07:37:00.001-05:00

I have a problem with the F test when I select the option ROBUST for the estimation .

I'm just running OLS regressions and when I select "robust" for the estimation of the variances the value of the F test is missing ! (I have 72 obs and 6 regresors)

If I don't select that option then I don't have the problem in the same estimation.

What can it be?

* * For searches and help try: * http://www.stata.com/support/faqs/res/findit.html * http://www.stata.com/support/statalist/faq * http://www.ats.ucla.edu/stat/stata/

Re: st: STATA syntax colouring in TextWrangler

2006-03-22T07:37:00.000-05:00

For "STATA" read "Stata", passim.

Steven Samuels asked essentially the same question on 17 January.

Some while ago I made a public appeal for information on this point on Statalist and got no replies, public or privste. Also, emails to Ben Hulley's last known email address at NIH elicited no reply (although they did not bounce).

Thus the link was removed from

http://fmwww.bc.edu/repec/bocode/t/textEditors.html

However, that FAQ does report syntax colouring for many other text editors.

Nick n.j.cox@durham.ac.uk

Steffen Hokland

-------------------------------------------------------------------------------- I've started to do STATA programming, and have decided to use my 'normal' text editor - TextWrangler (Apple Mac OS-X). I've been able to track down a syntax colouring module for TextWrangler (by Ben Hulley), but all links refering to it (e.g. homepage.mac.com/ benhulley) seem to be dead. Can anyone direct me to a module for syntax colouring in TextWrangler (or another texteditor for that matter)? * * For searches and help try: * http://www.stata.com/support/faqs/res/findit.html * http://www.stata.com/support/statalist/faq * http://www.ats.ucla.edu/stat/stata/

st: Scandanavian Journal of Statistics Prize

2006-03-22T07:28:00.000-05:00

This may be of interest to Statalisters, especially those in the cold north.

m.p.

The Scandinavian Journal of Statistics Prize for Young Researchers

======================================

The Scandinavian Journal of Statistics Prize of 40.000 DKK will be given to the young researcher who presents the best paper at the NORDSTAT meeting http://blackwell-publishing.msgfocus.com/c/13qFGFJEY2hnBuR, to be held in Rebild Bakker, Denmark, 11-15 June 2006, provided that the paper is accepted (before or after the meeting) as a regular submission to the Scandinavian Journal of Statistics (SJS) http://blackwell-publishing.msgfocus.com/c/13qH4N3U4HvcDpM (visit http://blackwell-publishing.msgfocus.com/c/13qHbpMnQUdbN6j see author guidelines for submission details). The editor of SJS must be informed that the paper is a candidate for the Scandinavian Journal of Statistics Prize.

To qualify as a candidate for the SJS Prize, the author must have received a Ph.D. degree at most five years before the NORDSTAT meeting or be younger than 35 years of age at the time of the meeting. The young author is preferably the sole-author of the paper, but if multiple young authors are involved, the paper will also be considered. There are no restrictions on nationality.

Young researchers of any nationality are strongly encouraged to present their work at the NORDSTAT 2006 meeting and to submit their papers to the SJS and for the SJS Prize competition.

====================================== * * For searches and help try: * http://www.stata.com/support/faqs/res/findit.html * http://www.stata.com/support/statalist/faq * http://www.ats.ucla.edu/stat/stata/

Re: st: Mata function stata() within program

2006-03-22T04:24:00.000-05:00

Thank you, Alan, that solved the problem. Also thanks for the detailed explanation.

Hopefully, I can show you why I need all that next week in Mannheim

Uli

Alan Riley wrote: > Ulrich Kohler asked about calling Stata (including a local macro > > substitution) from Mata: > > I have a Mata function which looks as follows: > > > > -------------------------------------lsq.mata-- > > > > (...) > > > > // Mata Function to extract the substitution costs from subcost-matrix > > void showhash(real rowvector R) > > { > > string scalar key1 > > st_local("key1",key1) > > key1 = strofreal(R[1,2]) > > st_local("key1",key1) > > stata("local hash1 = mod(`key1',197)") > > } > > > > (...) > > > > ------------------------------------------------ > > > > The function is part of the mata-libary lsq.mlib, which compiles without > > problems. All other function in the library works well. However, if I run > > > > -showhash()- from within Mata I got the following error: > > : R = 2,3,5,4 > > : showhash(R) > > > > invalid syntax > > stata(): 3598 Stata returned error > > showhash(): - function returned error > > <istmt>: - function returned error > > r(3598); > > > > Further investigation showed that the error is connected to the -stata()- > > function. However, After running -showhash()- I can intactively run the > > same > > > > -stata()- function without any problems: > > : stata("local hash1 = mod(`key1',197)") > > : stata("display `hash1'") > > > > 3 > > `key1' is being substituted at compile time. That is, when > Stata reads the lines > > void showhash(real rowvector R) > { > string scalar key1 > st_local("key1",key1) > key1 = strofreal(R[1,2]) > st_local("key1",key1) > stata("local hash1 = mod(`key1',197)") > } > > just before they are compiled by Mata, macros are substituted. > This, what is compiled is > > void showhash(real rowvector R) > { > string scalar key1 > st_local("key1",key1) > key1 = strofreal(R[1,2]) > st_local("key1",key1) > stata("local hash1 = mod(,197)") > } > > And, > > local hash1 = mod(,197) > > produces a syntax error when it is executed later. > > The reason it works interactively afterward is that by that point > the local macro 'key1' has already been defined, and thus what > Mata sees is > > stata("local hash1 = mod(1.234,197)") > > (or rather, it sees whatever number was in R[1,2] and thus > was stored in `key1'). > > You can see the advantage of this macro substitution in > adoupdate.ado. Type -which adoupdate- and then look at the > file in a text editor (being careful to not modify it of course). > Look for CMDNAME and the local macros after it, then search lower > in the file to see how they are used. Their values get substituted, > not at run-time, but just before the lines are compiled. > > Ulrich needs to protect `key1' from being substituted just > before compilation by coding > > stata("local hash1 = mod(\`key1',197)") > > The string that will be compiled into the code to be passed to > stata() will then literally be > > local hash1 = mod(`key1',197) > > which is what Ulrich wants. > > > Alan > (ariley@stata.com) > * > * For searches and help try: > * http://www.stata.com/support/faqs/res/findit.html > * http://www.stata.com/support/statalist/faq > * http://www.ats.ucla.edu/stat/stata/

st: STATA syntax colouring in TextWrangler

2006-03-22T04:16:00.000-05:00

I've started to do STATA programming, and have decided to use my 'normal' text editor - TextWrangler (Apple Mac OS-X). I've been able to track down a syntax colouring module for TextWrangler (by Ben Hulley), but all links refering to it (e.g. homepage.mac.com/ benhulley) seem to be dead. Can anyone direct me to a module for syntax colouring in TextWrangler (or another texteditor for that matter)?

Steffen * * For searches and help try: * http://www.stata.com/support/faqs/res/findit.html * http://www.stata.com/support/statalist/faq * http://www.ats.ucla.edu/stat/stata/

st: Queries about dfl

2006-03-22T03:40:00.000-05:00

Dear Statalist users: I would really appreciate any help with a couple of issues concerning the new dfl ssc package carried out by Pedro de Azevedo. First, I would like to run the package with more than 140 explanatory variables. Yet, the logit/probit that estimates the propensity score just considers the first 30. Is there any way of increasing the number of variables? Second, I would like to estimate some characteristics (mean, deciles or even inequality measures) of the counterfactual estimated density. Is it possible with the dfl package? Thanks in advance. Hipolito Simon University of Alicante

* * For searches and help try: * http://www.stata.com/support/faqs/res/findit.html * http://www.stata.com/support/statalist/faq * http://www.ats.ucla.edu/stat/stata/

st: RE: xtabond2

2006-03-22T01:00:00.000-05:00

It is shown in the help file. Just type -help xtabond2- if it has already been installed. Thuy Le

-----Original Message----- From: owner-statalist@hsphsun2.harvard.edu [mailto:owner-statalist@hsphsun2.harvard.edu] On Behalf Of Vivian Sibbaluca Sent: Wednesday, March 22, 2006 12:28 AM To: statalist@hsphsun2.harvard.edu Subject: st: xtabond2

Dear all: I am working on a panel data model with lagged endogenous variables as regressors, blundell and bond in particular. i was wondering how am i going to incorporate lags of y in the varlist. can anyone give a syntax of this model? thanks!

* * For searches and help try: * http://www.stata.com/support/faqs/res/findit.html * http://www.stata.com/support/statalist/faq * http://www.ats.ucla.edu/stat/stata/

st: xtabond2

2006-03-22T00:29:00.000-05:00

Re: st: Re: hausman test after mlogit - negative chi-square

2006-03-21T22:09:00.000-05:00

Marilyn Ibara wrote:

> I estimated a Multinominal logit model with 7 choices. I > have 17 independent variables. I specified mlogtest after I > ran the mlogit command.

It would have been worth mentioning the _full_ command (I suspect) you used to get the post-estimation results shown in your post: -mlogtest, all-.

> I am not sure about my results. Does anyone know how it is > possible to get a negative Chi-square for the hausman test > for IIA as I did in my results? Also, how are the degrees > of freedom calculated, it appears that I should have 17 and > not what is reported.

This question is answered by Long and Freese (2006: 244-5) in the second edition of their excellent textbook, cited below my signature. Essentially, this is very common, and it is often evidence that the independence of irrelevant alternatives (IIA) has _not_ been violated. They also suggest a good tip to confirm your results here: choose a different base category and run the same -mlogtest- command.

> Furthermore, my results of the hausman test differ from the > small-hsiao test, they contradict each other, I read online > this is common, does anyone know how one determines which > one is appropriate?

Long and Freese (2006: 243), quoting Monte Carlo research conducted by Cheng and Long (2005), report that neither the Hausman-McFadden (HM) nor the Small-Hsiao (SH) test is especially useful for assessing violations of IIA. This is largely because both have poor size properties even with some kinds of data, even N > 1000 in the case of HM but also sometimes for SH when N > 500. The best advice they give is only to use -mlogit- when you can _clearly_ distinguish between the outcome categories in your dataset. Quite frankly, I couldn't agree more.

In sum, buy Long and Freese!

CLIVE NICHOLAS |t: 0(044)7903 397793 Politics |e: clive.nicholas@ncl.ac.uk Newcastle University |http://www.ncl.ac.uk/geps

References:

Cheng S and Long JS (2005) "Testing For IIA in the Multinomial Logit Model", University of Connecticut, Working Paper.

Long JS and Freese J (2nd ed, 2006) REGRESSION MODELS FOR CATEGORICAL DEPENDENT VARIABLES USING STATA, College Station, TX: Stata Press.

* * For searches and help try: * http://www.stata.com/support/faqs/res/findit.html * http://www.stata.com/support/statalist/faq * http://www.ats.ucla.edu/stat/stata/

Re: st: Mata function stata() within program

2006-03-21T19:17:00.000-05:00

Ulrich Kohler asked about calling Stata (including a local macro substitution) from Mata: > I have a Mata function which looks as follows: > > -------------------------------------lsq.mata-- > > (...) > > // Mata Function to extract the substitution costs from subcost-matrix > void showhash(real rowvector R) > { > string scalar key1 > st_local("key1",key1) > key1 = strofreal(R[1,2]) > st_local("key1",key1) > stata("local hash1 = mod(`key1',197)") > } > > (...) > > ------------------------------------------------ > > The function is part of the mata-libary lsq.mlib, which compiles without > problems. All other function in the library works well. However, if I run > -showhash()- from within Mata I got the following error: > > : R = 2,3,5,4 > : showhash(R) > invalid syntax > stata(): 3598 Stata returned error > showhash(): - function returned error > <istmt>: - function returned error > r(3598); > > Further investigation showed that the error is connected to the -stata()- > function. However, After running -showhash()- I can intactively run the same > -stata()- function without any problems: > > : stata("local hash1 = mod(`key1',197)") > : stata("display `hash1'") > 3

`key1' is being substituted at compile time. That is, when Stata reads the lines

void showhash(real rowvector R) { string scalar key1 st_local("key1",key1) key1 = strofreal(R[1,2]) st_local("key1",key1) stata("local hash1 = mod(`key1',197)") }

just before they are compiled by Mata, macros are substituted. This, what is compiled is

void showhash(real rowvector R) { string scalar key1 st_local("key1",key1) key1 = strofreal(R[1,2]) st_local("key1",key1) stata("local hash1 = mod(,197)") }

And,

local hash1 = mod(,197)

produces a syntax error when it is executed later.

The reason it works interactively afterward is that by that point the local macro 'key1' has already been defined, and thus what Mata sees is

stata("local hash1 = mod(1.234,197)")

(or rather, it sees whatever number was in R[1,2] and thus was stored in `key1').

You can see the advantage of this macro substitution in adoupdate.ado. Type -which adoupdate- and then look at the file in a text editor (being careful to not modify it of course). Look for CMDNAME and the local macros after it, then search lower in the file to see how they are used. Their values get substituted, not at run-time, but just before the lines are compiled.

Ulrich needs to protect `key1' from being substituted just before compilation by coding

stata("local hash1 = mod(\`key1',197)")

The string that will be compiled into the code to be passed to stata() will then literally be

local hash1 = mod(`key1',197)

which is what Ulrich wants.

Alan (ariley@stata.com) * * For searches and help try: * http://www.stata.com/support/faqs/res/findit.html * http://www.stata.com/support/statalist/faq * http://www.ats.ucla.edu/stat/stata/

st: Re: hausman test after mlogit - negative chi-square

2006-03-21T18:54:00.000-05:00

Hi all,

I estimated a Multinominal logit model with 7 choices. I have 17 independent variables. I specified mlogtest after I ran the mlogit command.

I am not sure about my results. Does anyone know how it is possible to get a negative Chi-square for the hausman test for IIA as I did in my results? Also, how are the degrees of freedom calculated, it appears that I should have 17 and not what is reported.

Furthermore, my results of the hausman test differ from the small-hsiao test, they contradict each other, I read online this is common, does anyone know how one determines which one is appropriate?

Here is my output. Thanks in advance for any help.

Marilyn.

**** Hausman tests of IIA assumption

Ho: Odds(Outcome-J vs Outcome-K) are independent of other alternatives.

(storing estimation results as _HAUSMAN)

**** Small-Hsiao tests of IIA assumption

Ho: Odds(Outcome-J vs Outcome-K) are independent of other alternatives.

Omitted | lnL(full) lnL(omit) chi2 df P>chi2 evidence ---------+--------------------------------------------------- ------ 2 | -3123.828 -3080.945 85.767 17 0.000 against Ho 3 | -5769.906 -5729.245 81.323 17 0.000 against Ho 4 | -5647.749 -5601.483 92.532 17 0.000 against Ho 5 | -5774.945 -5730.604 88.682 17 0.000 against Ho 6 | -5633.140 -5594.139 78.001 17 0.000 against Ho 7 | -5429.578 -5384.741 89.674 17 0.000 against Ho ------------------------------------------------------------- ------

* * For searches and help try: * http://www.stata.com/support/faqs/res/findit.html * http://www.stata.com/support/statalist/faq * http://www.ats.ucla.edu/stat/stata/

st: RE: RE: RE: FW: zinb help

2006-03-21T17:49:00.000-05:00

You appear to be using a version other than the latest. The FAQ does advise making that clear.

Sorry; I can't tell what model you have in mind. But a wild guess is that you want -inflate(_cons)-.

Nick n.j.cox@durham.ac.uk

Rajesh Tharyan > > I am seriously sorry nick I wasn't trying to be clever ... > but I did go > through help and the manuals .. I was looking at the > descriptions which does > not say that clearly..it says.. > > zinb estimates a maximum-likelihood zero-inflated negative binomial > regression of depvar on varlist,where depvar is a nonnegative count > variable. > > Of course I didint understand the syntax diagram .. my > fault.. but now I > understand that Typing [ and ] means the varlist is optional. > > So the command will be zinb (var),inflate(var).. is that right?

* * For searches and help try: * http://www.stata.com/support/faqs/res/findit.html * http://www.stata.com/support/statalist/faq * http://www.ats.ucla.edu/stat/stata/

st: RE: RE: FW: zinb help

2006-03-21T17:36:00.000-05:00

Hi,

I am seriously sorry nick I wasnt trying to be clever ... but I did go through help and the manuals .. I was looking at the descriptions which does not say that clearly..it says..

zinb estimates a maximum-likelihood zero-inflated negative binomial regression of depvar on varlist,where depvar is a nonnegative count variable.

Of course I didint understand the syntax diagram .. my fault.. but now I understand that Typing [ and ] means the varlist is optional.

So the command will be zinb (var),inflate(var).. is that right?

Thanks Rajesh

This refers to the exchange below.

The syntax diagram for -zinb- makes it clear that independent variables need not be specified: thus -zinb- may be used to fit a distribution to a response (dependent) variable alone.

So, again, do please read the help.

Nick n.j.cox@durham.ac.uk

--------------------------- 28 February 2006 Stata offers many commands for these distributions and others in the same territory.

Please use -help-, -search- and the manual before sending such queries to the list.

> I have a dataset which looks like the one below > > Individual 1 individual 2 > Time count count > 1 0 1 > 2 1 0 > 3 0 0 > 4 0 0 > 5 3 1 > 6 0 0 > 7 1 0 > > Time represents months and count represents the occurance of > an event. I > understand that count might follow a zero inflated poisson > distribution or > a negative binomial distribution. > > Are there any stata command to check if the data fits these > distributions? > (just like we have the shapiro-wilk test for normality.). > > My objective is to see if there is any increase in the counts > between say > period1 (time 1 to 3) and period2 (time 4 to7) for each > individual. And for > all individuals taken together...

---------------------------------------------------------------------------- ----

Rajesh Tharyan > I have asked this question before but got told off... but I > am no closer to > finding an answer... I know there is "nbfit" to fit a > negative binomial > distribution to a single variable by the method of maximum > likelihood. Is > there something similar to fit a zero inflated negative binomial > distribution to a single variable? I found nbvargr which fits > a poisson and > a negbinomial dist to a variable but nothing on zero inflated > distributions. > > The only commands related to zero inflated models I found were zip and > zinb... > > I am sorry if it is still a very dumb question..

* * For searches and help try: * http://www.stata.com/support/faqs/res/findit.html * http://www.stata.com/support/statalist/faq * http://www.ats.ucla.edu/stat/stata/

st: RE: FW: zinb help

2006-03-21T17:16:00.000-05:00

This refers to the exchange below.

The syntax diagram for -zinb- makes it clear that independent variables need not be specified: thus -zinb- may be used to fit a distribution to a response (dependent) variable alone.

So, again, do please read the help.

Nick n.j.cox@durham.ac.uk

--------------------------- 28 February 2006 Stata offers many commands for these distributions and others in the same territory.

Please use -help-, -search- and the manual before sending such queries to the list.

--------------------------------------------------------------------------------

* * For searches and help try: * http://www.stata.com/support/faqs/res/findit.html * http://www.stata.com/support/statalist/faq * http://www.ats.ucla.edu/stat/stata/

st: FW: zinb help

2006-03-21T16:55:00.000-05:00

Hi all,

I have asked this question before but got told off... but I am no closer to finding an answer... I know there is nbfit to fit a negative binomial distribution to a single variable by the method of maximum likelihood. Is there something similar to fit a zero inflated negative binomial distribution to a single variable? I found nbvargr which fits a poisson and a negbinomial dist to a variable but nothing on zero inflated distributions.

The only commands related to zero inflated models I found were zip and zinb...

I am sorry if it is still a very dumb question..

Regards rajesh

* * For searches and help try: * http://www.stata.com/support/faqs/res/findit.html * http://www.stata.com/support/statalist/faq * http://www.ats.ucla.edu/stat/stata/

st: RE: Meta-analysis of diagnostic tests

2006-03-21T15:51:00.000-05:00

How is anyone expected to respond to your "not convinced" comment on -metan-?

No argument, no evidence => nothing to discuss.

Nick n.j.cox@durham.ac.uk

Seligman, Hilary > Has anybody used Stata to do a meta-analysis of diagnostic tests? > Specifically, I would like to generate summary positive and negative > likelihood ratios with p-values, and an associated test of > heterogeneity. > I am not convinced that the metan command (metan tp fn fp tn, with > tp=true positive, fn=false negative, etc.), as some have suggested, > produces correct results. > > As a (less desirable) alternative, I believe that I can use a logit > command to generate a summary sensitivity/specificity. However, I am > unclear on how that would actually be coded. > > Any advice on either approach would be MUCH appreciated. >

* * For searches and help try: * http://www.stata.com/support/faqs/res/findit.html * http://www.stata.com/support/statalist/faq * http://www.ats.ucla.edu/stat/stata/

st: ratio between oprobit marginal effects

2006-03-21T15:42:00.000-05:00

Dear Statalisters: I am trying to perform a ratio of marginal effects after an ordered probit command. That is, after . oprobit y x1 x2 x3 where y = 0,1,2 I asked for the marginal effects . mfx compute, predict (outcome(1)) . mfx compute, predict (outcome(2)) which gave me, say, "dydx1" and "dydx2". Now, what I need to know is "dydx2/dydx1" with its standard error. I guess I need to use "nlcom" command, but I have no clue on how to write this last syntax statement, despite all my tries. Does anyone have any clue on this matter? Thanks.

* * For searches and help try: * http://www.stata.com/support/faqs/res/findit.html * http://www.stata.com/support/statalist/faq * http://www.ats.ucla.edu/stat/stata/

st: Meta-analysis of diagnostic tests

2006-03-21T14:26:00.000-05:00

Has anybody used Stata to do a meta-analysis of diagnostic tests? Specifically, I would like to generate summary positive and negative likelihood ratios with p-values, and an associated test of heterogeneity. I am not convinced that the metan command (metan tp fn fp tn, with tp=true positive, fn=false negative, etc.), as some have suggested, produces correct results.

As a (less desirable) alternative, I believe that I can use a logit command to generate a summary sensitivity/specificity. However, I am unclear on how that would actually be coded.

Any advice on either approach would be MUCH appreciated.

Thanks very much,

Hilary Seligman hseligman@medsfgh.ucsf.edu

* * For searches and help try: * http://www.stata.com/support/faqs/res/findit.html * http://www.stata.com/support/statalist/faq * http://www.ats.ucla.edu/stat/stata/

Re: st: SUR and random coefficients?

2006-03-21T12:48:00.000-05:00

If you don't have a solution to this with STATA you can do it with AD Model builder's random effects module. In fact you can (almost) as easily do nonlinear SUR's with re's.

Dave

>if someone has an answer to Stephan, I am a taker as well. Have been >wanting to do SUR with RCM for a while.

----Message d'origine---- >De: "Stephan Brunow" <Stephan.Brunow@mailbox.tu-dresden.de> >A: <statalist@hsphsun2.harvard.edu> >Sujet: st: SUR and random coefficients? >Date: Wed, 15 Mar 2006 12:33:43 +0100 > >Dear all, > >we want to estimate a SUR model (Zellners seemingly unrelated regression) >but also want to consider a random coefficient approach (mixed >variables/multilevel model) in this context as well. Not only the intercept >but also variables should have the random coefficient. >Does anybody knows how we can do this in STATA? Possibly, does anybody host >an ado-file? It seems as if there is no way to do it with STATA at the >moment. > >Thanks for your help, > >Stephan Brunow > > > > >* >* For searches and help try: >* http://www.stata.com/support/faqs/res/findit.html >* http://www.stata.com/support/statalist/faq >* http://www.ats.ucla.edu/stat/stata/ > > -- David A. Fournier P.O. Box 2040, Sidney, B.C. V8l 3S3 Canada Phone/FAX 250-655-3364 http://otter-rsch.com * * For searches and help try: * http://www.stata.com/support/faqs/res/findit.html * http://www.stata.com/support/statalist/faq * http://www.ats.ucla.edu/stat/stata/

Re: st: frontier

2006-03-21T12:43:00.000-05:00

Might -help frontier postestimation- be helpful

khanna1@hei.unige.ch wrote: > Dear Statalist members, > > When using the "frontier" command to fit stochastic frontier production > models to estimate (in)efficiency using cross section data, how can one > view the individual inefficiecy for each unit of observation? I am > interested in viewing both the average inefficiency as well as the > producer specific inefficiency. > > > > Thank you. > > Ms. Gauri Khanna > > * > * For searches and help try: > * http://www.stata.com/support/faqs/res/findit.html > * http://www.stata.com/support/statalist/faq > * http://www.ats.ucla.edu/stat/stata/ > * * For searches and help try: * http://www.stata.com/support/faqs/res/findit.html * http://www.stata.com/support/statalist/faq * http://www.ats.ucla.edu/stat/stata/

st: Mata function stata() within program

2006-03-21T10:52:00.000-05:00

I have a Mata function which looks as follows:

-------------------------------------lsq.mata--

(...)

// Mata Function to extract the substitution costs from subcost-matrix void showhash(real rowvector R) { string scalar key1 st_local("key1",key1) key1 = strofreal(R[1,2]) st_local("key1",key1) stata("local hash1 = mod(`key1',197)") }

(...)

------------------------------------------------

The function is part of the mata-libary lsq.mlib, which compiles without problems. All other function in the library works well. However, if I run -showhash()- from within Mata I got the following error:

: R = 2,3,5,4 : showhash(R) invalid syntax stata(): 3598 Stata returned error showhash(): - function returned error <istmt>: - function returned error r(3598);

Further investigation showed that the error is connected to the -stata()- function. However, After running -showhash()- I can intactively run the same -stata()- function without any problems:

: stata("local hash1 = mod(`key1',197)") : stata("display `hash1'") 3

Any thoughts on this?

cu Uli

st: hausman test (FE vs RE)

2006-03-21T10:41:00.000-05:00

Dear statalist members:

I have a doubt, the Hausman test for Fixed effect verus Random effects should be obtained with robust and clustered corrected standard errors or without them.

thanks all

felipe

* * For searches and help try: * http://www.stata.com/support/faqs/res/findit.html * http://www.stata.com/support/statalist/faq * http://www.ats.ucla.edu/stat/stata/

st: frontier

2006-03-21T10:24:00.000-05:00

Dear Statalist members,

When using the "frontier" command to fit stochastic frontier production models to estimate (in)efficiency using cross section data, how can one view the individual inefficiecy for each unit of observation? I am interested in viewing both the average inefficiency as well as the producer specific inefficiency.

Thank you.

Ms. Gauri Khanna

* * For searches and help try: * http://www.stata.com/support/faqs/res/findit.html * http://www.stata.com/support/statalist/faq * http://www.ats.ucla.edu/stat/stata/

Re: st: How can I change the way Stata clusters the yvars in a bar graph?

2006-03-21T07:45:00.000-05:00

Deb wrote:

I am trying to make a bar graph that contains several y variables grouped over 1 categorical variable that has 2 levels (0, 1). -------------------------------------------------------------

Stating what command gave the undesired result would be helpful, first to other statalisters, next to yourself.

Try this:

clear sysuse uslifeexp.dta gen halfcent = year>=1950 graph bar le_wmale le_wfemale le_bmale le_bfemale, over(halfcent)

Is this similar to what you want? -graph bar- takes a list of y-variables.

Svend ________________________________________________________

Svend Juul Institut for Folkesundhed, Afdeling for Epidemiologi (Institute of Public Health, Department of Epidemiology) Vennelyst Boulevard 6 DK-8000 Aarhus C, Denmark Phone, work: +45 8942 6090 Phone, home: +45 8693 7796 Fax: +45 8613 1580 E-mail: sj@soci.au.dk _________________________________________________________

* * For searches and help try: * http://www.stata.com/support/faqs/res/findit.html * http://www.stata.com/support/statalist/faq * http://www.ats.ucla.edu/stat/stata/

st: Question related to treatment-effects model

2006-03-21T07:44:00.000-05:00

Hello, I would like to ask a question related to the Stata command to run a treatment-effects model using a two-step consistent estimator. I know that the Stata command "treatreg" fits the treatment-effects model that considers the effect of an endogenously chosen binary treatment on another endogenous continuous variable, conditional on two sets of independent variables.

HOW CAN I FIT THE TREATMENT-EFFECTS MODEL OF TWO ENDOGENOUSLY CHOSEN BINARY TRATEMENT ON ANOTHER CONTINUOUS VARIABLE?

Waiting for the answer, I thanks you in advance

Noemi Pace

-- Noemi Pace

Economics Department "Tor Vergata" University via Columbia 2 00173 Rome Italy

* * For searches and help try: * http://www.stata.com/support/faqs/res/findit.html * http://www.stata.com/support/statalist/faq * http://www.ats.ucla.edu/stat/stata/

Re: st: Do file editor

2006-03-20T18:49:00.000-05:00

From what you said it seems like you're trying to do the same thing again and again? If this is the case using the -program define- command will save you a lot of hassle. E.g.:

program drop _all program define repeat_codes . <insert codes that you're repeating all over the do file> . end

After you have done this whenever you need to refer to those codes, you'll just need to write:

repeat_codes

Hope this helps.

Ada

On 3/18/06, Carter Rees <carterrees@gmail.com> wrote: > Greetings, > > Working with Stata 9.1, Windows XP. > > When I attempt to cut and paste text from within the do-file editor to > different areas of the same do-file or from another word processing program > to the do-file editor I get the following error message: > > "You will exceed the maximum number of characters the Do-file Editor can > hold (131071 bytes) if text is pasted". > > So, this leaves me with the option of re-typing a lot of redundant code. A > search of the archives and FAQ has not produced an answer. I have checked > to make sure that I am not unknowingly pasting a large amount of text left > on my clipboard. > > Thanks. > > Carter > > > * > * For searches and help try: > * http://www.stata.com/support/faqs/res/findit.html > * http://www.stata.com/support/statalist/faq > * http://www.ats.ucla.edu/stat/stata/ >

-- Ada Ma Research Fellow Health Economics Research Unit University of Aberdeen, UK. http://www.abdn.ac.uk/heru/ Tel: +44 (0) 1224 553863 Fax: +44 (0) 1224 550926

* * For searches and help try: * http://www.stata.com/support/faqs/res/findit.html * http://www.stata.com/support/statalist/faq * http://www.ats.ucla.edu/stat/stata/

RE: st: question on gllamm (multilevel logit)

2006-03-20T18:02:00.000-05:00

I'm wrong. It does, currently, say "Author". This is a bit of a legal fiction. I certainly didn't write it all.

Nick n.j.cox@durham.ac.uk

> -----Original Message----- > From: owner-statalist@hsphsun2.harvard.edu > [mailto:owner-statalist@hsphsun2.harvard.edu]On Behalf Of Nick Cox > Sent: 20 March 2006 22:52 > To: statalist@hsphsun2.harvard.edu > Subject: RE: st: question on gllamm (multilevel logit) > > > When Stas says "Nick Cox FAQ" I think he means > the Statalist FAQ mentioned at the end of every > Statalist posting. > > The Statalist FAQ is not attributed to authors > -- only its official maintainer is mentioned -- > but includes the work of Kit Baum, Bill Gould, > Ken Higbee, Marcello Pagano, David Wormuth and > various others over the years. > > Nick > n.j.cox@durham.ac.uk > > Stas Kolenikov > > > Without any further information, nobody on this list will > > be able to help you -- Nick Cox FAQ makes all sorts of > suggestions as > > to what should be included in your posting. > > > * > * For searches and help try: > * http://www.stata.com/support/faqs/res/findit.html > * http://www.stata.com/support/statalist/faq > * http://www.ats.ucla.edu/stat/stata/ >

* * For searches and help try: * http://www.stata.com/support/faqs/res/findit.html * http://www.stata.com/support/statalist/faq * http://www.ats.ucla.edu/stat/stata/

st: RE: How can I change the way Stata clusters the yvars in a bar graph?

2006-03-20T17:59:00.000-05:00

You do not specify exactly what you typed or even what command you are using. That almost always helps and never hinders.

I don't know an automated way to do what you want, but it is not too difficult to knit your own. Here is a dopey example:

u auto, clear set scheme s1color graph bar turn, over(foreign) saving(turn_foreign, replace) asyvars /* */ yla(, ang(h)) legend(off) b2title(turn) ytitle(mean values) graph bar trunk, over(foreign) saving(trunk_foreign, replace) asyvars /* */ yla(, ang(h)) ysc(off) legend(off) b2title(trunk) graph combine "turn_foreign" "trunk_foreign", imargin(zero) ycommon row(1) ysize(4) xsize(6)

The first graph should specify a -ytitle()- The second and subsequent graphs should suppress the -yscale()-

All graphs should suppress the -legend()- specify the -b2title()- specify -saving()-

The -graph combine- should join the graphs compactly insist on a common y axis

Nick n.j.cox@durham.ac.uk

Deborah L. Duffy, Ph.D. > I am trying to make a bar graph that contains several y > variables grouped over 1 > categorical variable that has 2 levels (0, 1). Stata > automatically groups all > the yvars together for each level of the categorical variable, e.g., > > Cat1: yvar1 yvar2 yvar3 yvar4 Cat2: yvar1 yvar2 yvar3 yvar4 > > I would like to rearrange the bars into 4 clusters - 1 > cluster for each yvar > containing both levels of the cat var. e.g., > > yvar1:cat1cat2 yvar2:cat1cat2 yvar3:cat1cat2 yvar4:cat1cat2 > > I can do this easily in SPSS ("clustered bar graph") but > haven't been able to > find a way to do it with Stata. I think what I need to do is > something like > the 'ascategory' option only instead of treating the yvars as > the first over > group I need them to be the second over group. Is this possible??

* * For searches and help try: * http://www.stata.com/support/faqs/res/findit.html * http://www.stata.com/support/statalist/faq * http://www.ats.ucla.edu/stat/stata/

RE: st: question on gllamm (multilevel logit)

2006-03-20T17:54:00.000-05:00

When Stas says "Nick Cox FAQ" I think he means the Statalist FAQ mentioned at the end of every Statalist posting.

The Statalist FAQ is not attributed to authors -- only its official maintainer is mentioned -- but includes the work of Kit Baum, Bill Gould, Ken Higbee, Marcello Pagano, David Wormuth and various others over the years.

Nick n.j.cox@durham.ac.uk

Stas Kolenikov > Without any further information, nobody on this list will > be able to help you -- Nick Cox FAQ makes all sorts of suggestions as > to what should be included in your posting.

* * For searches and help try: * http://www.stata.com/support/faqs/res/findit.html * http://www.stata.com/support/statalist/faq * http://www.ats.ucla.edu/stat/stata/

Re: st: question on gllamm (multilevel logit)

2006-03-20T17:48:00.000-05:00

Well a number of strange things may be happening with your model specification and your data... Check that you don't have underidentified model (in other words, that you are not estimating the same constant or same slope twice); and check that you have the most recent version of -gllamm- and all the required programs (see gllamm.org). Without any further information, nobody on this list will be able to help you -- Nick Cox FAQ makes all sorts of suggestions as to what should be included in your posting.

On 3/20/06, sj2033@columbia.edu <sj2033@columbia.edu> wrote: > I'm trying to estimate logistic multilevel model with random > coefficients, using gllamm command with 'adapt' option. When I > submit the command, STATA calculates Initial values for fixed > effects, but keeps giving error messege of "Log-likelihood cannot > be computed" when running adaptive quadrature. Does anybody know > why this happens and how to get around this problem? Thanks > * > * For searches and help try: > * http://www.stata.com/support/faqs/res/findit.html > * http://www.stata.com/support/statalist/faq > * http://www.ats.ucla.edu/stat/stata/ >

-- Stas Kolenikov http://stas.kolenikov.name

* * For searches and help try: * http://www.stata.com/support/faqs/res/findit.html * http://www.stata.com/support/statalist/faq * http://www.ats.ucla.edu/stat/stata/

st: RE: Stata 9 scatter plot problem

2006-03-20T17:16:00.000-05:00

Make sure both your executable and ados are up-to-date. Type -update query-.

Nick n.j.cox@durham.ac.uk

Jonathan A. Schwabish > I just installed Stata 9 and am getting a weird error > when running a scatter plot (simple command line: > scatter var1 var2). I get a long line of errors that > start with "(note: scheme s2color not found, using > s2color)" and then later I get "system limit > exceeded-see manual". Any ideas? Do I need to > reinstall?

* * For searches and help try: * http://www.stata.com/support/faqs/res/findit.html * http://www.stata.com/support/statalist/faq * http://www.ats.ucla.edu/stat/stata/

st: Stata 9 scatter plot problem

2006-03-20T17:12:00.000-05:00

I just installed Stata 9 and am getting a weird error when running a scatter plot (simple command line: scatter var1 var2). I get a long line of errors that start with "(note: scheme s2color not found, using s2color)" and then later I get "system limit exceeded-see manual". Any ideas? Do I need to reinstall?

Thanks, jon

st: How can I change the way Stata clusters the yvars in a bar graph?

2006-03-20T17:08:00.000-05:00

Hello all,

I am trying to make a bar graph that contains several y variables grouped over 1 categorical variable that has 2 levels (0, 1). Stata automatically groups all the yvars together for each level of the categorical variable, e.g.,

Cat1: yvar1 yvar2 yvar3 yvar4 Cat2: yvar1 yvar2 yvar3 yvar4

I would like to rearrange the bars into 4 clusters - 1 cluster for each yvar containing both levels of the cat var. e.g.,

yvar1:cat1cat2 yvar2:cat1cat2 yvar3:cat1cat2 yvar4:cat1cat2

I can do this easily in SPSS ("clustered bar graph") but haven't been able to find a way to do it with Stata. I think what I need to do is something like the 'ascategory' option only instead of treating the yvars as the first over group I need them to be the second over group. Is this possible??

Any help is much appreciated!! Deb

-- Deborah L. Duffy, Ph.D. Research Specialist, Center for the Interaction of Animals and Society School of Veterinary Medicine, Department of Clinical Studies University of Pennsylvania 3900 Delancy Street Philadelphia, PA 19104-6010

PHONE: 215-746-2089, FAX: 215-746-2090, EMAIL: duffydl@vet.upenn.edu

* * For searches and help try: * http://www.stata.com/support/faqs/res/findit.html * http://www.stata.com/support/statalist/faq * http://www.ats.ucla.edu/stat/stata/

st: question on gllamm (multilevel logit)

2006-03-20T16:54:00.000-05:00

I'm trying to estimate logistic multilevel model with random coefficients, using gllamm command with 'adapt' option. When I submit the command, STATA calculates Initial values for fixed effects, but keeps giving error messege of "Log-likelihood cannot be computed" when running adaptive quadrature. Does anybody know why this happens and how to get around this problem? Thanks * * For searches and help try: * http://www.stata.com/support/faqs/res/findit.html * http://www.stata.com/support/statalist/faq * http://www.ats.ucla.edu/stat/stata/

RE: st: corrupted png files in 64-bit Stata 9 for Windows?

2006-03-20T15:03:00.000-05:00

Anders Alexandersson (andersalex@gmail.com) asks about PNG files created by Stata 9 for 64-bit x86-64 Windows: > I'm unable to create valid PNG files with -graph export- in 64-bit > Stata 9 for Windows. Is it a bug in 64-bit Stata 9 for Windows? > > After I have created a graph, I use -graph export fig1.png, replace- > and Stata creates the file fig1.png and writes the note "(file > fig1.png written in PNG format)". But the file cannot be open in the > latest versions of Internet Explorer or Firefox on my computer using > 64-bit Windows XP (x86), service pack 1. It can be open fine with > Windows Picture and Fax Viewer and Windows Explorer displays the > thumbnail picture fine.

I have just reproduced this and verified that it is a bug. It appears that Stata 9 for 64-bit x86-64 Windows is putting valid data in PNG files it creates, but is filling in an incorrect checksum value to 'validate' that data.

Windows Explorer and Windows Picture and Fax Viewer are ignoring the checksum, and can thus display the PNG without a problem. IE and Firefox are looking at the checksum before trying to display the PNG data.

This will be fixed in the next executable update.

Alan (ariley@stata.com)

* * For searches and help try: * http://www.stata.com/support/faqs/res/findit.html * http://www.stata.com/support/statalist/faq * http://www.ats.ucla.edu/stat/stata/

RE: st: ivreset

2006-03-20T13:32:00.000-05:00

Emails crossing in the post... I think you're probably right, though.

--Mark

> -----Original Message----- > From: owner-statalist@hsphsun2.harvard.edu > [mailto:owner-statalist@hsphsun2.harvard.edu] On Behalf Of > Arne Risa Hole > Sent: 20 March 2006 18:08 > To: statalist@hsphsun2.harvard.edu > Subject: Re: st: ivreset > > Mark, > > Note, however, that it is only because of the collinearity > problem - the fact that one variable is dropped from the > model when the forecast is not rescaled - that the test > statistics differ in your OLS example. > The rescaling does not matter in a test with only the squared > and cubed forecasts for example. > > This is different in the case of IV: even when collinearity > is not a problem the test statistics differ, as the example > of running Austin's code with and without the rescaling shows. > > The solution is probably to rescale after creating the > polynomials as you suggest - this way of rescaling the > forecasts does not affect the test statistic in either case, > IV or OLS, and may solve the numerical problems that using > the actual forcast might introduce - on the other hand it > doesn't seem to sort out the collinearity so this problem remains.. > > Cheers, > Arne > > > > > On 20/03/06, Schaffer, Mark E <M.E.Schaffer@hw.ac.uk> wrote: > > Arne, > > > > I think you are onto something here. The rescaling should perhaps > > take place *after* creating the polynomials, not before. > > > > That said, Stata's official -ovtest- also rescales first, > then creates > > the polynomials. When I wrote -ivreset-, I had used replication of > > the output of -ovtest- as a check, and I used the same approach to > > rescaling yhat, namely rescale and then create polynomials (rather > > than create polynomials and then rescale). > > > > Here is an example that demonstrates that (a) rescaling > after creating > > the polynomials leaves the reset statistic unchanged, and (b) > > rescaling before creating the polynomials replicates the output of > > official -ovtest-. Note that there are collinearity > problems with (a) > > even after rescaling, which makes me hesitate... > > > > Does anyone else want to comment? > > > > --Mark > > > > *********** do code *************** > > > > use http://fmwww.bc.edu/ec-p/data/hayashi/griliches76.dta, clear > > capture drop yhat* qui regress lw s qui predict double yhat > > * yhatr=rescaled yhat > > sum yhat, meanonly > > qui gen double yhatr = (yhat-r(min))/(r(max)-r(min)) qui gen double > > yhat2=yhat^2 qui gen double yhat3=yhat^3 qui gen double > yhat4=yhat^4 > > qui gen double yhatr2=yhatr^2 qui gen double yhatr3=yhatr^3 qui gen > > double yhatr4=yhatr^4 > > * yhatrr2=rescaled yhat2 > > * yhatrr3=rescaled yhat3 > > * yhatrr4=rescaled yhat4 > > sum yhat2, meanonly > > qui gen double yhatrr2 = (yhat2-r(min))/(r(max)-r(min)) sum yhat3, > > meanonly qui gen double yhatrr3 = > (yhat3-r(min))/(r(max)-r(min)) sum > > yhat4, meanonly qui gen double yhatrr4 = > > (yhat4-r(min))/(r(max)-r(min)) sum yhat* > > > > * Unrescaled RESET > > qui regress lw s yhat2 yhat3 yhat4 > > testparm yhat* > > * yhat that is first ^2, ^3, ^4, then rescaled > > * Output identical to unrescaled RESET qui regress lw s yhatrr2 > > yhatrr3 yhatrr4 testparm yhat* > > * yhat that is first rescaled, then ^2, ^3, ^4 > > * Output different from unrescaled RESET qui regress lw s yhatr2 > > yhatr3 yhatr4 testparm yhat* > > * Stata's built-in ovtest > > * Output again different from unrescaled RESET qui regress > lw s Ovtest > > > > ************* output **************** > > > > > > . use http://fmwww.bc.edu/ec-p/data/hayashi/griliches76.dta, clear > > (Wages of Very Young Men, Zvi Griliches, J.Pol.Ec. 1976) > > > > . capture drop yhat* > > > > . qui regress lw s > > > > . qui predict double yhat > > > > . * yhatr=rescaled yhat > > . sum yhat, meanonly > > > > . qui gen double yhatr = (yhat-r(min))/(r(max)-r(min)) > > > > . qui gen double yhat2=yhat^2 > > > > . qui gen double yhat3=yhat^3 > > > > . qui gen double yhat4=yhat^4 > > > > . qui gen double yhatr2=yhatr^2 > > > > . qui gen double yhatr3=yhatr^3 > > > > . qui gen double yhatr4=yhatr^4 > > > > . * yhatrr2=rescaled yhat2 > > . * yhatrr3=rescaled yhat3 > > . * yhatrr4=rescaled yhat4 > > . sum yhat2, meanonly > > > > . qui gen double yhatrr2 = (yhat2-r(min))/(r(max)-r(min)) > > > > . sum yhat3, meanonly > > > > . qui gen double yhatrr3 = (yhat3-r(min))/(r(max)-r(min)) > > > > . sum yhat4, meanonly > > > > . qui gen double yhatrr4 = (yhat4-r(min))/(r(max)-r(min)) > > > > . sum yhat* > > > > Variable | Obs Mean Std. Dev. Min > Max > > > -------------+-------------------------------------------------------- > > yhat | 758 5.686739 .2156493 5.261107 > 6.130727 > > yhatr | 758 .4894459 .2479809 0 > 1 > > yhat2 | 758 32.38544 2.473995 27.67924 > 37.58581 > > yhat3 | 758 184.7003 21.3139 145.6234 > 230.4284 > > yhat4 | 758 1054.929 163.4247 766.1405 > 1412.693 > > > -------------+-------------------------------------------------------- > > yhatr2 | 758 .3009707 .2769761 0 > 1 > > yhatr3 | 758 .2142687 .2681644 0 > 1 > > yhatr4 | 758 .1671979 .2530434 0 > 1 > > yhatrr2 | 758 .4750582 .2497327 0 > 1 > > yhatrr3 | 758 .4607848 .2513286 0 > 1 > > > -------------+-------------------------------------------------------- > > yhatrr4 | 758 .4466593 .252763 0 > 1 > > > > . > > . * Unrescaled RESET > > . qui regress lw s yhat2 yhat3 yhat4 > > > > . testparm yhat* > > > > ( 1) yhat2 = 0 > > ( 2) yhat3 = 0 > > ( 3) yhat4 = 0 > > Constraint 1 dropped > > > > F( 2, 754) = 0.87 > > Prob > F = 0.4191 > > > > . * yhat that is first ^2, ^3, ^4, then rescaled . * Output > identical > > to unrescaled RESET . qui regress lw s yhatrr2 yhatrr3 yhatrr4 > > > > . testparm yhat* > > > > ( 1) yhatrr2 = 0 > > ( 2) yhatrr3 = 0 > > ( 3) yhatrr4 = 0 > > Constraint 1 dropped > > > > F( 2, 754) = 0.87 > > Prob > F = 0.4191 > > > > . * yhat that is first rescaled, then ^2, ^3, ^4 . * Output > different > > from unrescaled RESET . qui regress lw s yhatr2 yhatr3 yhatr4 > > > > . testparm yhat* > > > > ( 1) yhatr2 = 0 > > ( 2) yhatr3 = 0 > > ( 3) yhatr4 = 0 > > > > F( 3, 753) = 0.59 > > Prob > F = 0.6216 > > > > . * Stata's built-in ovtest > > . * Output again different from unrescaled RESET . qui regress lw s > > > > . ovtest > > > > Ramsey RESET test using powers of the fitted values of lw > > Ho: model has no omitted variables > > F(3, 753) = 0.59 > > Prob > F = 0.6216 > > > > . > > end of do-file > > > > ************************************* > > > > > -----Original Message----- > > > From: owner-statalist@hsphsun2.harvard.edu > > > [mailto:owner-statalist@hsphsun2.harvard.edu] On Behalf > Of Arne Risa > > > Hole > > > Sent: 20 March 2006 16:03 > > > To: statalist@hsphsun2.harvard.edu > > > Subject: Re: st: ivreset > > > > > > Hi Mark, > > > > > > Thank you very much for the clarifications. The rescaling > does have > > > an impact on the test statistic, however; this can be seen from > > > using Austin's code and comparing the results with and > without the > > > line: > > > > > > replace yhat = (yhat-r(min))/(r(max)-r(min)) > > > > > > So even when the correct "optimal" forecast of yhat is used > > > (yhat=X-hat*beta-hat), rescaling the forecast affects the result. > > > This is not a problem in the case of -ovtest-, however, since the > > > Reset test statistic is invariant to the rescaling in the > OLS case. > > > > > > Sorry for going on about this, but it seems to me that > since the two > > > statistics differ the correct statistic is the one without the > > > rescaling of the (yhat=X-hat*beta-hat) forecast (even though this > > > may introduce numerical precision problems in some cases). > > > > > > Cheers, > > > Arne > > > > > > On 20/03/06, Schaffer, Mark E <M.E.Schaffer@hw.ac.uk> wrote: > > > > Arne, > > > > > > > > > -----Original Message----- > > > > > From: owner-statalist@hsphsun2.harvard.edu > > > > > [mailto:owner-statalist@hsphsun2.harvard.edu] On Behalf > > > Of Arne Risa > > > > > Hole > > > > > Sent: 18 March 2006 11:43 > > > > > To: statalist@hsphsun2.harvard.edu > > > > > Subject: RE: st: ivreset > > > > > > > > > > Austin, Mark, > > > > > > > > > > Thank you both for your replies, rescaling the > forecast did the > > > > > trick (sorry for the bad formatting of my code > before, it looked > > > > > fine in my email programme). > > > > > > > > > > I understand the motivation behind the rescaling, but I'm > > > slightly > > > > > concerned about the fact that it produces a different > > > test statistic > > > > > compared to using the actual forecast. Note that this > is not the > > > > > case when using the Reset test following OLS - the test > > > statistic is > > > > > invariant to the rescaling in this case. > > > > > > > > > > I would think that since the two approaches (rescaling/ no > > > > > rescaling) produce different results, the correct test > > > statistic is > > > > > that using the actual forecast? > > > > > > > > There are two different issues here. Rescaling is one, and > > > it is, in > > > > some sense, a side issue. The problem is that sometimes > > > the yhat has > > > > large-ish values, and the higher order polynomials of yhat that > > > > are included in the artificial regression can get so > big that the > > > > regression fails for numerical precision reasons. Stata's > > > own version > > > > of the reset test, -ovtest-, also does this rescaling. > The test > > > > statistic is, of course, invariant in theory to the units > > > used and hence to rescaling. > > > > > > > > The other issue is the one you might have missed. As > > > Austin pointed > > > > out, the IV version of the RESET test cannot use > standard fitted > > > > values that would be generated by -predict- after > estimation using > > > > -ivreg- or -ivreg2-. These are yhat=X*beta-hat, and > the problem > > > > is that X includes some endogenous regressors. > > > > > > > > As the help file for -ivreset- explains, there are two > alternatives. > > > > One is to use reduced form predictions for yhat, i.e., > regress y > > > > on all the exogenous variables (including the excluded > > > instruments) and > > > > then use -predict-. The other is to get what Pesaran and > > > Taylor call > > > > the "optimal forecast" of yhat. This is not > yhat=X*beta-hat, but > > > > yhat=X-hat*beta-hat, where X-hat includes the reduced form > > > predicted > > > > values of the endogenous regressors (rather than the > actual values). > > > > The code that Austin kindly posted to Statalast implemented > > > the latter. > > > > > > > > Cheers, > > > > Mark > > > > > > > > > Cheers > > > > > Arne > > > > > > > > > > On Mar 17 2006, Schaffer, Mark E wrote: > > > > > > > > > > > Austin, Arne, > > > > > > > > > > > > > -----Original Message----- > > > > > > > From: Austin Nichols [mailto:austinnichols@gmail.com] > > > > > > > Sent: 16 March 2006 23:30 > > > > > > > To: statalist@hsphsun2.harvard.edu > > > > > > > Subject: Re: st: ivreset > > > > > > > > > > > > > > -findit ivreset- then -help ivreset- when installed has a > > > > > excellent > > > > > > > exposition that begins: > > > > > > > > > > > > > > As Pagan and Hall (1983) and Pesaran and Taylor (1999) > > > > > point out, a > > > > > > > RESET test for an IV regression cannot use the standard > > > > > IV predicted > > > > > > > values X*beta-hat, because X includes endogenous > > > > > regressors that are > > > > > > > correlated with u. > > > > > > > > > > > > > > Try this code instead: > > > > > > > > > > > > > > use http://fmwww.bc.edu/ec-p/data/hayashi/griliches76.dta > > > > > > > qui ivreg2 lw s expr tenure rns smsa (iq=med kww) predict > > > > > ytilde mat > > > > > > > b=e(b) mat li b qui regress iq s expr tenure rns smsa med > > > > > > > kww qui predict double xh gen > yhat=ytil-b[1,1]*iq+b[1,1]*xh > > > > > > > sum yhat, meanonly qui replace yhat = > > > (yhat-r(min))/(r(max)-r(min)) > > > > > > > qui gen double yhat2=yhat^2 qui ivreg2 lw s expr > > > tenure rns smsa > > > > > > > yhat2 (iq=med kww) test yhat2 qui ivreg2 lw s expr tenure > > > > > > > rns > > > > > smsa (iq=med > > > > > > > kww) ivreset > > > > > > > > > > > > > > Now, as to why > > > > > > > replace yhat = (yhat-r(min))/(r(max)-r(min)) I cannot > > > > > tell you, but > > > > > > > it's in ivreset.ado > > > > > > > > > > > > It's to rescale yhat so that when it's squared, > cubed, etc., > > > > > > it doesn't get wildly out of scale relative to the other > > > regressors. > > > > > > This can cause problems for the regression that includes > > > > > these terms. > > > > > > > > > > > > Cheers, > > > > > > Mark > > > > > > > > > > > > > > > > > > > On 16 Mar 2006 19:10:13 +0000, Arne Risa Hole wrote: > > > > > > > > I am using ivreset to do a Pesaran-Taylor Reset test > > > > > after ivreg2. > > > > > > > > However, I am not able to replicate the result from > > > > > > > > ivreset > > > > > > > manually. For example: > > > > > > > > > > > > > > > > use > http://fmwww.bc.edu/ec-p/data/hayashi/griliches76.dta > > > > > > > > > > > > > > > > > > > > > > > > > > * > > > > > > * For searches and help try: > > > > > > * http://www.stata.com/support/faqs/res/findit.html > > > > > > * http://www.stata.com/support/statalist/faq > > > > > > * http://www.ats.ucla.edu/stat/stata/ > > > > > > > > > > > > > > > > * > > > > > * For searches and help try: > > > > > * http://www.stata.com/support/faqs/res/findit.html > > > > > * http://www.stata.com/support/statalist/faq > > > > > * http://www.ats.ucla.edu/stat/stata/ > > > > > > > > > > > > > > > > > > * > > > > * For searches and help try: > > > > * http://www.stata.com/support/faqs/res/findit.html > > > > * http://www.stata.com/support/statalist/faq > > > > * http://www.ats.ucla.edu/stat/stata/ > > > > > > > > > > * > > > * For searches and help try: > > > * http://www.stata.com/support/faqs/res/findit.html > > > * http://www.stata.com/support/statalist/faq > > > * http://www.ats.ucla.edu/stat/stata/ > > > > > > > > > > * > > * For searches and help try: > > * http://www.stata.com/support/faqs/res/findit.html > > * http://www.stata.com/support/statalist/faq > > * http://www.ats.ucla.edu/stat/stata/ > > > > * > * For searches and help try: > * http://www.stata.com/support/faqs/res/findit.html > * http://www.stata.com/support/statalist/faq > * http://www.ats.ucla.edu/stat/stata/ > >

* * For searches and help try: * http://www.stata.com/support/faqs/res/findit.html * http://www.stata.com/support/statalist/faq * http://www.ats.ucla.edu/stat/stata/

RE: st: ivreset

2006-03-20T13:09:00.001-05:00

Hi all. I think I spoke too soon. The differences between the different approaches to rescaling are driven, I think, by numerical precision problems and collinearities.

Here are the 3 versions of the basic ovtest using squares and cubes: no rescaling, rescaling yhat and then squaring and cubing, and squaring and cubing followed by rescaling. This uses the code that I posted to the list earlier today. All three generate the same results. When a yhat^4 is included, as in the example I posted earlier, is when collinearities arise and different test statistics can be generated.

--Mark

. * Squared and cubed . * Unrescaled RESET . qui regress lw s yhat2 yhat3

. testparm yhat*

( 1) yhat2 = 0 ( 2) yhat3 = 0

F( 2, 754) = 0.87 Prob > F = 0.4188

. * yhat that is first ^2, ^3, then rescaled . * Output identical to unrescaled RESET . qui regress lw s yhatrr2 yhatrr3

. testparm yhat*

( 1) yhatrr2 = 0 ( 2) yhatrr3 = 0

F( 2, 754) = 0.87 Prob > F = 0.4188

. * yhat that is first rescaled, then ^2, ^3 . * Output different from unrescaled RESET . qui regress lw s yhatr2 yhatr3

. testparm yhat*

( 1) yhatr2 = 0 ( 2) yhatr3 = 0

F( 2, 754) = 0.87 Prob > F = 0.4188

> -----Original Message----- > From: owner-statalist@hsphsun2.harvard.edu > [mailto:owner-statalist@hsphsun2.harvard.edu] On Behalf Of > Schaffer, Mark E > Sent: 20 March 2006 16:57 > To: statalist@hsphsun2.harvard.edu > Subject: RE: st: ivreset > > Arne, > > I think you are onto something here. The rescaling should > perhaps take place *after* creating the polynomials, not before. > > That said, Stata's official -ovtest- also rescales first, > then creates the polynomials. When I wrote -ivreset-, I had > used replication of the output of -ovtest- as a check, and I > used the same approach to rescaling yhat, namely rescale and > then create polynomials (rather than create polynomials and > then rescale). > > Here is an example that demonstrates that (a) rescaling after > creating the polynomials leaves the reset statistic > unchanged, and (b) rescaling before creating the polynomials > replicates the output of official -ovtest-. Note that there > are collinearity problems with (a) even after rescaling, > which makes me hesitate... > > Does anyone else want to comment? > > --Mark > > *********** do code *************** > > use http://fmwww.bc.edu/ec-p/data/hayashi/griliches76.dta, > clear capture drop yhat* qui regress lw s qui predict double yhat > * yhatr=rescaled yhat > sum yhat, meanonly > qui gen double yhatr = (yhat-r(min))/(r(max)-r(min)) qui gen > double yhat2=yhat^2 qui gen double yhat3=yhat^3 qui gen > double yhat4=yhat^4 qui gen double yhatr2=yhatr^2 qui gen > double yhatr3=yhatr^3 qui gen double yhatr4=yhatr^4 > * yhatrr2=rescaled yhat2 > * yhatrr3=rescaled yhat3 > * yhatrr4=rescaled yhat4 > sum yhat2, meanonly > qui gen double yhatrr2 = (yhat2-r(min))/(r(max)-r(min)) sum > yhat3, meanonly qui gen double yhatrr3 = > (yhat3-r(min))/(r(max)-r(min)) sum yhat4, meanonly qui gen > double yhatrr4 = (yhat4-r(min))/(r(max)-r(min)) sum yhat* > > * Unrescaled RESET > qui regress lw s yhat2 yhat3 yhat4 > testparm yhat* > * yhat that is first ^2, ^3, ^4, then rescaled > * Output identical to unrescaled RESET > qui regress lw s yhatrr2 yhatrr3 yhatrr4 testparm yhat* > * yhat that is first rescaled, then ^2, ^3, ^4 > * Output different from unrescaled RESET qui regress lw s > yhatr2 yhatr3 yhatr4 testparm yhat* > * Stata's built-in ovtest > * Output again different from unrescaled RESET qui regress lw s Ovtest > > ************* output **************** > > > . use http://fmwww.bc.edu/ec-p/data/hayashi/griliches76.dta, > clear (Wages of Very Young Men, Zvi Griliches, J.Pol.Ec. 1976) > > . capture drop yhat* > > . qui regress lw s > > . qui predict double yhat > > . * yhatr=rescaled yhat > . sum yhat, meanonly > > . qui gen double yhatr = (yhat-r(min))/(r(max)-r(min)) > > . qui gen double yhat2=yhat^2 > > . qui gen double yhat3=yhat^3 > > . qui gen double yhat4=yhat^4 > > . qui gen double yhatr2=yhatr^2 > > . qui gen double yhatr3=yhatr^3 > > . qui gen double yhatr4=yhatr^4 > > . * yhatrr2=rescaled yhat2 > . * yhatrr3=rescaled yhat3 > . * yhatrr4=rescaled yhat4 > . sum yhat2, meanonly > > . qui gen double yhatrr2 = (yhat2-r(min))/(r(max)-r(min)) > > . sum yhat3, meanonly > > . qui gen double yhatrr3 = (yhat3-r(min))/(r(max)-r(min)) > > . sum yhat4, meanonly > > . qui gen double yhatrr4 = (yhat4-r(min))/(r(max)-r(min)) > > . sum yhat* > > Variable | Obs Mean Std. Dev. Min Max > -------------+-------------------------------------------------------- > yhat | 758 5.686739 .2156493 5.261107 6.130727 > yhatr | 758 .4894459 .2479809 0 1 > yhat2 | 758 32.38544 2.473995 27.67924 37.58581 > yhat3 | 758 184.7003 21.3139 145.6234 230.4284 > yhat4 | 758 1054.929 163.4247 766.1405 1412.693 > -------------+-------------------------------------------------------- > yhatr2 | 758 .3009707 .2769761 0 1 > yhatr3 | 758 .2142687 .2681644 0 1 > yhatr4 | 758 .1671979 .2530434 0 1 > yhatrr2 | 758 .4750582 .2497327 0 1 > yhatrr3 | 758 .4607848 .2513286 0 1 > -------------+-------------------------------------------------------- > yhatrr4 | 758 .4466593 .252763 0 1 > > . > . * Unrescaled RESET > . qui regress lw s yhat2 yhat3 yhat4 > > . testparm yhat* > > ( 1) yhat2 = 0 > ( 2) yhat3 = 0 > ( 3) yhat4 = 0 > Constraint 1 dropped > > F( 2, 754) = 0.87 > Prob > F = 0.4191 > > . * yhat that is first ^2, ^3, ^4, then rescaled . * Output > identical to unrescaled RESET . qui regress lw s yhatrr2 > yhatrr3 yhatrr4 > > . testparm yhat* > > ( 1) yhatrr2 = 0 > ( 2) yhatrr3 = 0 > ( 3) yhatrr4 = 0 > Constraint 1 dropped > > F( 2, 754) = 0.87 > Prob > F = 0.4191 > > . * yhat that is first rescaled, then ^2, ^3, ^4 . * Output > different from unrescaled RESET . qui regress lw s yhatr2 > yhatr3 yhatr4 > > . testparm yhat* > > ( 1) yhatr2 = 0 > ( 2) yhatr3 = 0 > ( 3) yhatr4 = 0 > > F( 3, 753) = 0.59 > Prob > F = 0.6216 > > . * Stata's built-in ovtest > . * Output again different from unrescaled RESET . qui regress lw s > > . ovtest > > Ramsey RESET test using powers of the fitted values of lw > Ho: model has no omitted variables > F(3, 753) = 0.59 > Prob > F = 0.6216 > > . > end of do-file > > ************************************* > > > -----Original Message----- > > From: owner-statalist@hsphsun2.harvard.edu > > [mailto:owner-statalist@hsphsun2.harvard.edu] On Behalf Of > Arne Risa > > Hole > > Sent: 20 March 2006 16:03 > > To: statalist@hsphsun2.harvard.edu > > Subject: Re: st: ivreset > > > > Hi Mark, > > > > Thank you very much for the clarifications. The rescaling > does have an > > impact on the test statistic, however; this can be seen from using > > Austin's code and comparing the results with and without the line: > > > > replace yhat = (yhat-r(min))/(r(max)-r(min)) > > > > So even when the correct "optimal" forecast of yhat is used > > (yhat=X-hat*beta-hat), rescaling the forecast affects the > result. This > > is not a problem in the case of -ovtest-, however, since the Reset > > test statistic is invariant to the rescaling in the OLS case. > > > > Sorry for going on about this, but it seems to me that > since the two > > statistics differ the correct statistic is the one without the > > rescaling of the (yhat=X-hat*beta-hat) forecast (even > though this may > > introduce numerical precision problems in some cases). > > > > Cheers, > > Arne > > > > On 20/03/06, Schaffer, Mark E <M.E.Schaffer@hw.ac.uk> wrote: > > > Arne, > > > > > > > -----Original Message----- > > > > From: owner-statalist@hsphsun2.harvard.edu > > > > [mailto:owner-statalist@hsphsun2.harvard.edu] On Behalf > > Of Arne Risa > > > > Hole > > > > Sent: 18 March 2006 11:43 > > > > To: statalist@hsphsun2.harvard.edu > > > > Subject: RE: st: ivreset > > > > > > > > Austin, Mark, > > > > > > > > Thank you both for your replies, rescaling the forecast did the > > > > trick (sorry for the bad formatting of my code before, > it looked > > > > fine in my email programme). > > > > > > > > I understand the motivation behind the rescaling, but I'm > > slightly > > > > concerned about the fact that it produces a different > > test statistic > > > > compared to using the actual forecast. Note that this > is not the > > > > case when using the Reset test following OLS - the test > > statistic is > > > > invariant to the rescaling in this case. > > > > > > > > I would think that since the two approaches (rescaling/ no > > > > rescaling) produce different results, the correct test > > statistic is > > > > that using the actual forecast? > > > > > > There are two different issues here. Rescaling is one, and > > it is, in > > > some sense, a side issue. The problem is that sometimes > > the yhat has > > > large-ish values, and the higher order polynomials of > yhat that are > > > included in the artificial regression can get so big that the > > > regression fails for numerical precision reasons. Stata's > > own version > > > of the reset test, -ovtest-, also does this rescaling. The test > > > statistic is, of course, invariant in theory to the units > > used and hence to rescaling. > > > > > > The other issue is the one you might have missed. As > > Austin pointed > > > out, the IV version of the RESET test cannot use standard fitted > > > values that would be generated by -predict- after estimation using > > > -ivreg- or -ivreg2-. These are yhat=X*beta-hat, and the > problem is > > > that X includes some endogenous regressors. > > > > > > As the help file for -ivreset- explains, there are two > alternatives. > > > One is to use reduced form predictions for yhat, i.e., > regress y on > > > all the exogenous variables (including the excluded > > instruments) and > > > then use -predict-. The other is to get what Pesaran and > > Taylor call > > > the "optimal forecast" of yhat. This is not yhat=X*beta-hat, but > > > yhat=X-hat*beta-hat, where X-hat includes the reduced form > > predicted > > > values of the endogenous regressors (rather than the > actual values). > > > The code that Austin kindly posted to Statalast implemented > > the latter. > > > > > > Cheers, > > > Mark > > > > > > > Cheers > > > > Arne > > > > > > > > On Mar 17 2006, Schaffer, Mark E wrote: > > > > > > > > > Austin, Arne, > > > > > > > > > > > -----Original Message----- > > > > > > From: Austin Nichols [mailto:austinnichols@gmail.com] > > > > > > Sent: 16 March 2006 23:30 > > > > > > To: statalist@hsphsun2.harvard.edu > > > > > > Subject: Re: st: ivreset > > > > > > > > > > > > -findit ivreset- then -help ivreset- when installed has a > > > > excellent > > > > > > exposition that begins: > > > > > > > > > > > > As Pagan and Hall (1983) and Pesaran and Taylor (1999) > > > > point out, a > > > > > > RESET test for an IV regression cannot use the standard > > > > IV predicted > > > > > > values X*beta-hat, because X includes endogenous > > > > regressors that are > > > > > > correlated with u. > > > > > > > > > > > > Try this code instead: > > > > > > > > > > > > use http://fmwww.bc.edu/ec-p/data/hayashi/griliches76.dta > > > > > > qui ivreg2 lw s expr tenure rns smsa (iq=med kww) predict > > > > ytilde mat > > > > > > b=e(b) mat li b qui regress iq s expr tenure rns > smsa med kww > > > > > > qui predict double xh gen yhat=ytil-b[1,1]*iq+b[1,1]*xh sum > > > > > > yhat, meanonly qui replace yhat = > > (yhat-r(min))/(r(max)-r(min)) > > > > > > qui gen double yhat2=yhat^2 qui ivreg2 lw s expr > > tenure rns smsa > > > > > > yhat2 (iq=med kww) test yhat2 qui ivreg2 lw s expr > tenure rns > > > > smsa (iq=med > > > > > > kww) ivreset > > > > > > > > > > > > Now, as to why > > > > > > replace yhat = (yhat-r(min))/(r(max)-r(min)) I cannot > > > > tell you, but > > > > > > it's in ivreset.ado > > > > > > > > > > It's to rescale yhat so that when it's squared, > cubed, etc., it > > > > > doesn't get wildly out of scale relative to the other > > regressors. > > > > > This can cause problems for the regression that includes > > > > these terms. > > > > > > > > > > Cheers, > > > > > Mark > > > > > > > > > > > > > > > > On 16 Mar 2006 19:10:13 +0000, Arne Risa Hole wrote: > > > > > > > I am using ivreset to do a Pesaran-Taylor Reset test > > > > after ivreg2. > > > > > > > However, I am not able to replicate the result > from ivreset > > > > > > manually. For example: > > > > > > > > > > > > > > use http://fmwww.bc.edu/ec-p/data/hayashi/griliches76.dta > > > > > > > > > > > > > > > > > > > > > > * > > > > > * For searches and help try: > > > > > * http://www.stata.com/support/faqs/res/findit.html > > > > > * http://www.stata.com/support/statalist/faq > > > > > * http://www.ats.ucla.edu/stat/stata/ > > > > > > > > > > > > > * > > > > * For searches and help try: > > > > * http://www.stata.com/support/faqs/res/findit.html > > > > * http://www.stata.com/support/statalist/faq > > > > * http://www.ats.ucla.edu/stat/stata/ > > > > > > > > > > > > > > * > > > * For searches and help try: > > > * http://www.stata.com/support/faqs/res/findit.html > > > * http://www.stata.com/support/statalist/faq > > > * http://www.ats.ucla.edu/stat/stata/ > > > > > > > * > > * For searches and help try: > > * http://www.stata.com/support/faqs/res/findit.html > > * http://www.stata.com/support/statalist/faq > > * http://www.ats.ucla.edu/stat/stata/ > > > > > > * > * For searches and help try: > * http://www.stata.com/support/faqs/res/findit.html > * http://www.stata.com/support/statalist/faq > * http://www.ats.ucla.edu/stat/stata/ > >

* * For searches and help try: * http://www.stata.com/support/faqs/res/findit.html * http://www.stata.com/support/statalist/faq * http://www.ats.ucla.edu/stat/stata/

Re: st: ivreset

2006-03-20T13:09:00.000-05:00

Mark,

Note, however, that it is only because of the collinearity problem - the fact that one variable is dropped from the model when the forecast is not rescaled - that the test statistics differ in your OLS example. The rescaling does not matter in a test with only the squared and cubed forecasts for example.

This is different in the case of IV: even when collinearity is not a problem the test statistics differ, as the example of running Austin's code with and without the rescaling shows.

The solution is probably to rescale after creating the polynomials as you suggest - this way of rescaling the forecasts does not affect the test statistic in either case, IV or OLS, and may solve the numerical problems that using the actual forcast might introduce - on the other hand it doesn't seem to sort out the collinearity so this problem remains..

Cheers, Arne

On 20/03/06, Schaffer, Mark E <M.E.Schaffer@hw.ac.uk> wrote: > Arne, > > I think you are onto something here. The rescaling should perhaps take > place *after* creating the polynomials, not before. > > That said, Stata's official -ovtest- also rescales first, then creates > the polynomials. When I wrote -ivreset-, I had used replication of the > output of -ovtest- as a check, and I used the same approach to rescaling > yhat, namely rescale and then create polynomials (rather than create > polynomials and then rescale). > > Here is an example that demonstrates that (a) rescaling after creating > the polynomials leaves the reset statistic unchanged, and (b) rescaling > before creating the polynomials replicates the output of official > -ovtest-. Note that there are collinearity problems with (a) even after > rescaling, which makes me hesitate... > > Does anyone else want to comment? > > --Mark > > *********** do code *************** > > use http://fmwww.bc.edu/ec-p/data/hayashi/griliches76.dta, clear > capture drop yhat* > qui regress lw s > qui predict double yhat > * yhatr=rescaled yhat > sum yhat, meanonly > qui gen double yhatr = (yhat-r(min))/(r(max)-r(min)) > qui gen double yhat2=yhat^2 > qui gen double yhat3=yhat^3 > qui gen double yhat4=yhat^4 > qui gen double yhatr2=yhatr^2 > qui gen double yhatr3=yhatr^3 > qui gen double yhatr4=yhatr^4 > * yhatrr2=rescaled yhat2 > * yhatrr3=rescaled yhat3 > * yhatrr4=rescaled yhat4 > sum yhat2, meanonly > qui gen double yhatrr2 = (yhat2-r(min))/(r(max)-r(min)) > sum yhat3, meanonly > qui gen double yhatrr3 = (yhat3-r(min))/(r(max)-r(min)) > sum yhat4, meanonly > qui gen double yhatrr4 = (yhat4-r(min))/(r(max)-r(min)) > sum yhat* > > * Unrescaled RESET > qui regress lw s yhat2 yhat3 yhat4 > testparm yhat* > * yhat that is first ^2, ^3, ^4, then rescaled > * Output identical to unrescaled RESET > qui regress lw s yhatrr2 yhatrr3 yhatrr4 > testparm yhat* > * yhat that is first rescaled, then ^2, ^3, ^4 > * Output different from unrescaled RESET > qui regress lw s yhatr2 yhatr3 yhatr4 > testparm yhat* > * Stata's built-in ovtest > * Output again different from unrescaled RESET > qui regress lw s > Ovtest > > ************* output **************** > > > . use http://fmwww.bc.edu/ec-p/data/hayashi/griliches76.dta, clear > (Wages of Very Young Men, Zvi Griliches, J.Pol.Ec. 1976) > > . capture drop yhat* > > . qui regress lw s > > . qui predict double yhat > > . * yhatr=rescaled yhat > . sum yhat, meanonly > > . qui gen double yhatr = (yhat-r(min))/(r(max)-r(min)) > > . qui gen double yhat2=yhat^2 > > . qui gen double yhat3=yhat^3 > > . qui gen double yhat4=yhat^4 > > . qui gen double yhatr2=yhatr^2 > > . qui gen double yhatr3=yhatr^3 > > . qui gen double yhatr4=yhatr^4 > > . * yhatrr2=rescaled yhat2 > . * yhatrr3=rescaled yhat3 > . * yhatrr4=rescaled yhat4 > . sum yhat2, meanonly > > . qui gen double yhatrr2 = (yhat2-r(min))/(r(max)-r(min)) > > . sum yhat3, meanonly > > . qui gen double yhatrr3 = (yhat3-r(min))/(r(max)-r(min)) > > . sum yhat4, meanonly > > . qui gen double yhatrr4 = (yhat4-r(min))/(r(max)-r(min)) > > . sum yhat* > > Variable | Obs Mean Std. Dev. Min Max > -------------+-------------------------------------------------------- > yhat | 758 5.686739 .2156493 5.261107 6.130727 > yhatr | 758 .4894459 .2479809 0 1 > yhat2 | 758 32.38544 2.473995 27.67924 37.58581 > yhat3 | 758 184.7003 21.3139 145.6234 230.4284 > yhat4 | 758 1054.929 163.4247 766.1405 1412.693 > -------------+-------------------------------------------------------- > yhatr2 | 758 .3009707 .2769761 0 1 > yhatr3 | 758 .2142687 .2681644 0 1 > yhatr4 | 758 .1671979 .2530434 0 1 > yhatrr2 | 758 .4750582 .2497327 0 1 > yhatrr3 | 758 .4607848 .2513286 0 1 > -------------+-------------------------------------------------------- > yhatrr4 | 758 .4466593 .252763 0 1 > > . > . * Unrescaled RESET > . qui regress lw s yhat2 yhat3 yhat4 > > . testparm yhat* > > ( 1) yhat2 = 0 > ( 2) yhat3 = 0 > ( 3) yhat4 = 0 > Constraint 1 dropped > > F( 2, 754) = 0.87 > Prob > F = 0.4191 > > . * yhat that is first ^2, ^3, ^4, then rescaled > . * Output identical to unrescaled RESET > . qui regress lw s yhatrr2 yhatrr3 yhatrr4 > > . testparm yhat* > > ( 1) yhatrr2 = 0 > ( 2) yhatrr3 = 0 > ( 3) yhatrr4 = 0 > Constraint 1 dropped > > F( 2, 754) = 0.87 > Prob > F = 0.4191 > > . * yhat that is first rescaled, then ^2, ^3, ^4 > . * Output different from unrescaled RESET > . qui regress lw s yhatr2 yhatr3 yhatr4 > > . testparm yhat* > > ( 1) yhatr2 = 0 > ( 2) yhatr3 = 0 > ( 3) yhatr4 = 0 > > F( 3, 753) = 0.59 > Prob > F = 0.6216 > > . * Stata's built-in ovtest > . * Output again different from unrescaled RESET > . qui regress lw s > > . ovtest > > Ramsey RESET test using powers of the fitted values of lw > Ho: model has no omitted variables > F(3, 753) = 0.59 > Prob > F = 0.6216 > > . > end of do-file > > ************************************* > > > -----Original Message----- > > From: owner-statalist@hsphsun2.harvard.edu > > [mailto:owner-statalist@hsphsun2.harvard.edu] On Behalf Of > > Arne Risa Hole > > Sent: 20 March 2006 16:03 > > To: statalist@hsphsun2.harvard.edu > > Subject: Re: st: ivreset > > > > Hi Mark, > > > > Thank you very much for the clarifications. The rescaling > > does have an impact on the test statistic, however; this can > > be seen from using Austin's code and comparing the results > > with and without the line: > > > > replace yhat = (yhat-r(min))/(r(max)-r(min)) > > > > So even when the correct "optimal" forecast of yhat is used > > (yhat=X-hat*beta-hat), rescaling the forecast affects the > > result. This is not a problem in the case of -ovtest-, > > however, since the Reset test statistic is invariant to the > > rescaling in the OLS case. > > > > Sorry for going on about this, but it seems to me that since > > the two statistics differ the correct statistic is the one > > without the rescaling of the (yhat=X-hat*beta-hat) forecast > > (even though this may introduce numerical precision problems > > in some cases). > > > > Cheers, > > Arne > > > > On 20/03/06, Schaffer, Mark E <M.E.Schaffer@hw.ac.uk> wrote: > > > Arne, > > > > > > > -----Original Message----- > > > > From: owner-statalist@hsphsun2.harvard.edu > > > > [mailto:owner-statalist@hsphsun2.harvard.edu] On Behalf > > Of Arne Risa > > > > Hole > > > > Sent: 18 March 2006 11:43 > > > > To: statalist@hsphsun2.harvard.edu > > > > Subject: RE: st: ivreset > > > > > > > > Austin, Mark, > > > > > > > > Thank you both for your replies, rescaling the forecast did the > > > > trick (sorry for the bad formatting of my code before, it looked > > > > fine in my email programme). > > > > > > > > I understand the motivation behind the rescaling, but I'm > > slightly > > > > concerned about the fact that it produces a different > > test statistic > > > > compared to using the actual forecast. Note that this is not the > > > > case when using the Reset test following OLS - the test > > statistic is > > > > invariant to the rescaling in this case. > > > > > > > > I would think that since the two approaches (rescaling/ no > > > > rescaling) produce different results, the correct test > > statistic is > > > > that using the actual forecast? > > > > > > There are two different issues here. Rescaling is one, and > > it is, in > > > some sense, a side issue. The problem is that sometimes > > the yhat has > > > large-ish values, and the higher order polynomials of yhat that are > > > included in the artificial regression can get so big that the > > > regression fails for numerical precision reasons. Stata's > > own version > > > of the reset test, -ovtest-, also does this rescaling. The test > > > statistic is, of course, invariant in theory to the units > > used and hence to rescaling. > > > > > > The other issue is the one you might have missed. As > > Austin pointed > > > out, the IV version of the RESET test cannot use standard fitted > > > values that would be generated by -predict- after estimation using > > > -ivreg- or -ivreg2-. These are yhat=X*beta-hat, and the problem is > > > that X includes some endogenous regressors. > > > > > > As the help file for -ivreset- explains, there are two alternatives. > > > One is to use reduced form predictions for yhat, i.e., regress y on > > > all the exogenous variables (including the excluded > > instruments) and > > > then use -predict-. The other is to get what Pesaran and > > Taylor call > > > the "optimal forecast" of yhat. This is not yhat=X*beta-hat, but > > > yhat=X-hat*beta-hat, where X-hat includes the reduced form > > predicted > > > values of the endogenous regressors (rather than the actual values). > > > The code that Austin kindly posted to Statalast implemented > > the latter. > > > > > > Cheers, > > > Mark > > > > > > > Cheers > > > > Arne > > > > > > > > On Mar 17 2006, Schaffer, Mark E wrote: > > > > > > > > > Austin, Arne, > > > > > > > > > > > -----Original Message----- > > > > > > From: Austin Nichols [mailto:austinnichols@gmail.com] > > > > > > Sent: 16 March 2006 23:30 > > > > > > To: statalist@hsphsun2.harvard.edu > > > > > > Subject: Re: st: ivreset > > > > > > > > > > > > -findit ivreset- then -help ivreset- when installed has a > > > > excellent > > > > > > exposition that begins: > > > > > > > > > > > > As Pagan and Hall (1983) and Pesaran and Taylor (1999) > > > > point out, a > > > > > > RESET test for an IV regression cannot use the standard > > > > IV predicted > > > > > > values X*beta-hat, because X includes endogenous > > > > regressors that are > > > > > > correlated with u. > > > > > > > > > > > > Try this code instead: > > > > > > > > > > > > use http://fmwww.bc.edu/ec-p/data/hayashi/griliches76.dta > > > > > > qui ivreg2 lw s expr tenure rns smsa (iq=med kww) predict > > > > ytilde mat > > > > > > b=e(b) mat li b qui regress iq s expr tenure rns smsa med kww > > > > > > qui predict double xh gen yhat=ytil-b[1,1]*iq+b[1,1]*xh sum > > > > > > yhat, meanonly qui replace yhat = > > (yhat-r(min))/(r(max)-r(min)) > > > > > > qui gen double yhat2=yhat^2 qui ivreg2 lw s expr > > tenure rns smsa > > > > > > yhat2 (iq=med kww) test yhat2 qui ivreg2 lw s expr tenure rns > > > > smsa (iq=med > > > > > > kww) ivreset > > > > > > > > > > > > Now, as to why > > > > > > replace yhat = (yhat-r(min))/(r(max)-r(min)) I cannot > > > > tell you, but > > > > > > it's in ivreset.ado > > > > > > > > > > It's to rescale yhat so that when it's squared, cubed, etc., it > > > > > doesn't get wildly out of scale relative to the other > > regressors. > > > > > This can cause problems for the regression that includes > > > > these terms. > > > > > > > > > > Cheers, > > > > > Mark > > > > > > > > > > > > > > > > On 16 Mar 2006 19:10:13 +0000, Arne Risa Hole wrote: > > > > > > > I am using ivreset to do a Pesaran-Taylor Reset test > > > > after ivreg2. > > > > > > > However, I am not able to replicate the result from ivreset > > > > > > manually. For example: > > > > > > > > > > > > > > use http://fmwww.bc.edu/ec-p/data/hayashi/griliches76.dta > > > > > > > > > > > > > > > > > > > > > > * > > > > > * For searches and help try: > > > > > * http://www.stata.com/support/faqs/res/findit.html > > > > > * http://www.stata.com/support/statalist/faq > > > > > * http://www.ats.ucla.edu/stat/stata/ > > > > > > > > > > > > > * > > > > * For searches and help try: > > > > * http://www.stata.com/support/faqs/res/findit.html > > > > * http://www.stata.com/support/statalist/faq > > > > * http://www.ats.ucla.edu/stat/stata/ > > > > > > > > > > > > > > * > > > * For searches and help try: > > > * http://www.stata.com/support/faqs/res/findit.html > > > * http://www.stata.com/support/statalist/faq > > > * http://www.ats.ucla.edu/stat/stata/ > > > > > > > * > > * For searches and help try: > > * http://www.stata.com/support/faqs/res/findit.html > > * http://www.stata.com/support/statalist/faq > > * http://www.ats.ucla.edu/stat/stata/ > > > > > > * > * For searches and help try: > * http://www.stata.com/support/faqs/res/findit.html > * http://www.stata.com/support/statalist/faq > * http://www.ats.ucla.edu/stat/stata/ >

* * For searches and help try: * http://www.stata.com/support/faqs/res/findit.html * http://www.stata.com/support/statalist/faq * http://www.ats.ucla.edu/stat/stata/

st: corrupted png files in 64-bit Stata 9 for Windows?

2006-03-20T12:05:00.000-05:00

I'm unable to create valid PNG files with -graph export- in 64-bit Stata 9 for Windows. Is it a bug in 64-bit Stata 9 for Windows?

After I have created a graph, I use -graph export fig1.png, replace- and Stata creates the file fig1.png and writes the note "(file fig1.png written in PNG format)". But the file cannot be open in the latest versions of Internet Explorer or Firefox on my computer using 64-bit Windows XP (x86), service pack 1. It can be open fine with Windows Picture and Fax Viewer and Windows Explorer displays the thumbnail picture fine. I do not have this problem of viewing the PNG file in my web browsers when the file is created using either 32-bit Stata 9 for Windows or 32-bit Stata 9 for Linux. All my versions of Stata 9 are updated.

Anders Alexandersson andersalex@gmail.com

* * For searches and help try: * http://www.stata.com/support/faqs/res/findit.html * http://www.stata.com/support/statalist/faq * http://www.ats.ucla.edu/stat/stata/

RE: st: ivreset

2006-03-20T11:59:00.000-05:00

Arne,

I think you are onto something here. The rescaling should perhaps take place *after* creating the polynomials, not before.

That said, Stata's official -ovtest- also rescales first, then creates the polynomials. When I wrote -ivreset-, I had used replication of the output of -ovtest- as a check, and I used the same approach to rescaling yhat, namely rescale and then create polynomials (rather than create polynomials and then rescale).

Here is an example that demonstrates that (a) rescaling after creating the polynomials leaves the reset statistic unchanged, and (b) rescaling before creating the polynomials replicates the output of official -ovtest-. Note that there are collinearity problems with (a) even after rescaling, which makes me hesitate...

Does anyone else want to comment?

--Mark

*********** do code ***************

use http://fmwww.bc.edu/ec-p/data/hayashi/griliches76.dta, clear capture drop yhat* qui regress lw s qui predict double yhat * yhatr=rescaled yhat sum yhat, meanonly qui gen double yhatr = (yhat-r(min))/(r(max)-r(min)) qui gen double yhat2=yhat^2 qui gen double yhat3=yhat^3 qui gen double yhat4=yhat^4 qui gen double yhatr2=yhatr^2 qui gen double yhatr3=yhatr^3 qui gen double yhatr4=yhatr^4 * yhatrr2=rescaled yhat2 * yhatrr3=rescaled yhat3 * yhatrr4=rescaled yhat4 sum yhat2, meanonly qui gen double yhatrr2 = (yhat2-r(min))/(r(max)-r(min)) sum yhat3, meanonly qui gen double yhatrr3 = (yhat3-r(min))/(r(max)-r(min)) sum yhat4, meanonly qui gen double yhatrr4 = (yhat4-r(min))/(r(max)-r(min)) sum yhat*

* Unrescaled RESET qui regress lw s yhat2 yhat3 yhat4 testparm yhat* * yhat that is first ^2, ^3, ^4, then rescaled * Output identical to unrescaled RESET qui regress lw s yhatrr2 yhatrr3 yhatrr4 testparm yhat* * yhat that is first rescaled, then ^2, ^3, ^4 * Output different from unrescaled RESET qui regress lw s yhatr2 yhatr3 yhatr4 testparm yhat* * Stata's built-in ovtest * Output again different from unrescaled RESET qui regress lw s Ovtest

************* output ****************

. use http://fmwww.bc.edu/ec-p/data/hayashi/griliches76.dta, clear (Wages of Very Young Men, Zvi Griliches, J.Pol.Ec. 1976)

. capture drop yhat*

. qui regress lw s

. qui predict double yhat

. * yhatr=rescaled yhat . sum yhat, meanonly

. qui gen double yhatr = (yhat-r(min))/(r(max)-r(min))

. qui gen double yhat2=yhat^2

. qui gen double yhat3=yhat^3

. qui gen double yhat4=yhat^4

. qui gen double yhatr2=yhatr^2

. qui gen double yhatr3=yhatr^3

. qui gen double yhatr4=yhatr^4

. * yhatrr2=rescaled yhat2 . * yhatrr3=rescaled yhat3 . * yhatrr4=rescaled yhat4 . sum yhat2, meanonly

. qui gen double yhatrr2 = (yhat2-r(min))/(r(max)-r(min))

. sum yhat3, meanonly

. qui gen double yhatrr3 = (yhat3-r(min))/(r(max)-r(min))

. sum yhat4, meanonly

. qui gen double yhatrr4 = (yhat4-r(min))/(r(max)-r(min))

. sum yhat*

Variable | Obs Mean Std. Dev. Min Max -------------+-------------------------------------------------------- yhat | 758 5.686739 .2156493 5.261107 6.130727 yhatr | 758 .4894459 .2479809 0 1 yhat2 | 758 32.38544 2.473995 27.67924 37.58581 yhat3 | 758 184.7003 21.3139 145.6234 230.4284 yhat4 | 758 1054.929 163.4247 766.1405 1412.693 -------------+-------------------------------------------------------- yhatr2 | 758 .3009707 .2769761 0 1 yhatr3 | 758 .2142687 .2681644 0 1 yhatr4 | 758 .1671979 .2530434 0 1 yhatrr2 | 758 .4750582 .2497327 0 1 yhatrr3 | 758 .4607848 .2513286 0 1 -------------+-------------------------------------------------------- yhatrr4 | 758 .4466593 .252763 0 1

. . * Unrescaled RESET . qui regress lw s yhat2 yhat3 yhat4

. testparm yhat*

( 1) yhat2 = 0 ( 2) yhat3 = 0 ( 3) yhat4 = 0 Constraint 1 dropped

F( 2, 754) = 0.87 Prob > F = 0.4191

. * yhat that is first ^2, ^3, ^4, then rescaled . * Output identical to unrescaled RESET . qui regress lw s yhatrr2 yhatrr3 yhatrr4

. testparm yhat*

( 1) yhatrr2 = 0 ( 2) yhatrr3 = 0 ( 3) yhatrr4 = 0 Constraint 1 dropped

F( 2, 754) = 0.87 Prob > F = 0.4191

. * yhat that is first rescaled, then ^2, ^3, ^4 . * Output different from unrescaled RESET . qui regress lw s yhatr2 yhatr3 yhatr4

. testparm yhat*

( 1) yhatr2 = 0 ( 2) yhatr3 = 0 ( 3) yhatr4 = 0

F( 3, 753) = 0.59 Prob > F = 0.6216

. * Stata's built-in ovtest . * Output again different from unrescaled RESET . qui regress lw s

. ovtest

Ramsey RESET test using powers of the fitted values of lw Ho: model has no omitted variables F(3, 753) = 0.59 Prob > F = 0.6216

. end of do-file

*************************************

* * For searches and help try: * http://www.stata.com/support/faqs/res/findit.html * http://www.stata.com/support/statalist/faq * http://www.ats.ucla.edu/stat/stata/

st: recovering estat output

2006-03-20T11:44:00.000-05:00

Abiel said

After running a regression, I am interested in capturing the results of the estat ic command so that I can work with the Bayesian information criteria field in a do file. Is there a way to read the results of this command?

Although regress is e-class, ic is r-class. After ic,

. mat list r(S)

r(S)[1,6] N ll0 ll df AIC BIC r1 74 -695.71287 -675.17278 6 1362.3456 1376.1699

Kit Baum, Boston College Economics http://ideas.repec.org/e/pba1.html

* * For searches and help try: * http://www.stata.com/support/faqs/res/findit.html * http://www.stata.com/support/statalist/faq * http://www.ats.ucla.edu/stat/stata/

st: -adoupdate-

2006-03-20T11:10:00.000-05:00

First off, this is a really nice utility. Like most users, I had multiple copies of the same stuff installed so this has helped clean up my ados in addition to helping me keep them updated.

It was pointed out that you may need to run -adoupdate, update- a number of times before it gets everything right. However, I have a list of stubborn ados which must be updated however, -adoupdate, update- does not appear to be updating them (or if it does then they change everyday)

------------------begin log--------------------------------------

Packages that need to be updated are...

[80] collin -- zdemo. Collinearity Diagnostics [95] sjlatex -- sjlatex: LaTeX files for the Stata Journal ------------------end log--------------------------------------

And today, -textbarplot- was added to this list (I run -adoupdate, update- a number of times but kept getting this list)

[126] textbarplot -- 'TEXTBARPLOT': module to produce horizontal text and bar plot

Is anyone familiar with this and does it mean I dont have the most recent copies of these ados? (I did the uninstall, reinstall but it didnt help)

Ronnie * * For searches and help try: * http://www.stata.com/support/faqs/res/findit.html * http://www.stata.com/support/statalist/faq * http://www.ats.ucla.edu/stat/stata/

Re: st: FW: p-value

2006-03-20T11:09:00.000-05:00

On 20 Márta 2006, at 15:07, Jochen Siegele wrote:

>> Dear STATA-listers, >> >> >> could someone of you please tell me which command I need to get >> the p-value(s) after doing a (multiple) linear regression in STATA 8. >> >> Thanks a lot in advance.

-regress-

Ronán Conroy rconroy@rcsi.ie

* * For searches and help try: * http://www.stata.com/support/faqs/res/findit.html * http://www.stata.com/support/statalist/faq * http://www.ats.ucla.edu/stat/stata/

Re: st: ivreset

2006-03-20T11:05:00.000-05:00

Hi Mark,

Thank you very much for the clarifications. The rescaling does have an impact on the test statistic, however; this can be seen from using Austin's code and comparing the results with and without the line:

replace yhat = (yhat-r(min))/(r(max)-r(min))

So even when the correct "optimal" forecast of yhat is used (yhat=X-hat*beta-hat), rescaling the forecast affects the result. This is not a problem in the case of -ovtest-, however, since the Reset test statistic is invariant to the rescaling in the OLS case.

Sorry for going on about this, but it seems to me that since the two statistics differ the correct statistic is the one without the rescaling of the (yhat=X-hat*beta-hat) forecast (even though this may introduce numerical precision problems in some cases).

Cheers, Arne

On 20/03/06, Schaffer, Mark E <M.E.Schaffer@hw.ac.uk> wrote: > Arne, > > > -----Original Message----- > > From: owner-statalist@hsphsun2.harvard.edu > > [mailto:owner-statalist@hsphsun2.harvard.edu] On Behalf Of > > Arne Risa Hole > > Sent: 18 March 2006 11:43 > > To: statalist@hsphsun2.harvard.edu > > Subject: RE: st: ivreset > > > > Austin, Mark, > > > > Thank you both for your replies, rescaling the forecast did > > the trick (sorry for the bad formatting of my code before, it > > looked fine in my email programme). > > > > I understand the motivation behind the rescaling, but I'm > > slightly concerned about the fact that it produces a > > different test statistic compared to using the actual > > forecast. Note that this is not the case when using the Reset > > test following OLS - the test statistic is invariant to the > > rescaling in this case. > > > > I would think that since the two approaches (rescaling/ no > > rescaling) produce different results, the correct test > > statistic is that using the actual forecast? > > There are two different issues here. Rescaling is one, and it is, in > some sense, a side issue. The problem is that sometimes the yhat has > large-ish values, and the higher order polynomials of yhat that are > included in the artificial regression can get so big that the regression > fails for numerical precision reasons. Stata's own version of the reset > test, -ovtest-, also does this rescaling. The test statistic is, of > course, invariant in theory to the units used and hence to rescaling. > > The other issue is the one you might have missed. As Austin pointed > out, the IV version of the RESET test cannot use standard fitted values > that would be generated by -predict- after estimation using -ivreg- or > -ivreg2-. These are yhat=X*beta-hat, and the problem is that X includes > some endogenous regressors. > > As the help file for -ivreset- explains, there are two alternatives. > One is to use reduced form predictions for yhat, i.e., regress y on all > the exogenous variables (including the excluded instruments) and then > use -predict-. The other is to get what Pesaran and Taylor call the > "optimal forecast" of yhat. This is not yhat=X*beta-hat, but > yhat=X-hat*beta-hat, where X-hat includes the reduced form predicted > values of the endogenous regressors (rather than the actual values). > The code that Austin kindly posted to Statalast implemented the latter. > > Cheers, > Mark > > > Cheers > > Arne > > > > On Mar 17 2006, Schaffer, Mark E wrote: > > > > > Austin, Arne, > > > > > > > -----Original Message----- > > > > From: Austin Nichols [mailto:austinnichols@gmail.com] > > > > Sent: 16 March 2006 23:30 > > > > To: statalist@hsphsun2.harvard.edu > > > > Subject: Re: st: ivreset > > > > > > > > -findit ivreset- then -help ivreset- when installed has a > > excellent > > > > exposition that begins: > > > > > > > > As Pagan and Hall (1983) and Pesaran and Taylor (1999) > > point out, a > > > > RESET test for an IV regression cannot use the standard > > IV predicted > > > > values X*beta-hat, because X includes endogenous > > regressors that are > > > > correlated with u. > > > > > > > > Try this code instead: > > > > > > > > use http://fmwww.bc.edu/ec-p/data/hayashi/griliches76.dta > > > > qui ivreg2 lw s expr tenure rns smsa (iq=med kww) predict > > ytilde mat > > > > b=e(b) mat li b qui regress iq s expr tenure rns smsa med kww qui > > > > predict double xh gen yhat=ytil-b[1,1]*iq+b[1,1]*xh sum yhat, > > > > meanonly qui replace yhat = (yhat-r(min))/(r(max)-r(min)) qui gen > > > > double yhat2=yhat^2 qui ivreg2 lw s expr tenure rns smsa yhat2 > > > > (iq=med kww) test yhat2 qui ivreg2 lw s expr tenure rns > > smsa (iq=med > > > > kww) ivreset > > > > > > > > Now, as to why > > > > replace yhat = (yhat-r(min))/(r(max)-r(min)) I cannot > > tell you, but > > > > it's in ivreset.ado > > > > > > It's to rescale yhat so that when it's squared, cubed, etc., it > > > doesn't get wildly out of scale relative to the other regressors. > > > This can cause problems for the regression that includes > > these terms. > > > > > > Cheers, > > > Mark > > > > > > > > > > On 16 Mar 2006 19:10:13 +0000, Arne Risa Hole wrote: > > > > > I am using ivreset to do a Pesaran-Taylor Reset test > > after ivreg2. > > > > > However, I am not able to replicate the result from ivreset > > > > manually. For example: > > > > > > > > > > use http://fmwww.bc.edu/ec-p/data/hayashi/griliches76.dta > > > > > > > > > > > > > > * > > > * For searches and help try: > > > * http://www.stata.com/support/faqs/res/findit.html > > > * http://www.stata.com/support/statalist/faq > > > * http://www.ats.ucla.edu/stat/stata/ > > > > > > > * > > * For searches and help try: > > * http://www.stata.com/support/faqs/res/findit.html > > * http://www.stata.com/support/statalist/faq > > * http://www.ats.ucla.edu/stat/stata/ > > > > > > * > * For searches and help try: > * http://www.stata.com/support/faqs/res/findit.html > * http://www.stata.com/support/statalist/faq > * http://www.ats.ucla.edu/stat/stata/ >

* * For searches and help try: * http://www.stata.com/support/faqs/res/findit.html * http://www.stata.com/support/statalist/faq * http://www.ats.ucla.edu/stat/stata/

st: Recovering estat output

2006-03-20T11:04:00.000-05:00

Thanks,

Abiel

* * For searches and help try: * http://www.stata.com/support/faqs/res/findit.html * http://www.stata.com/support/statalist/faq * http://www.ats.ucla.edu/stat/stata/

st: RE: FW: p-value

2006-03-20T10:55:00.000-05:00

No command. p-values are indicated by -regress- itself as Prob > F and and P > |t|. Or at least this is true of Stata.

Nick n.j.cox@durham.ac.uk

Jochen Siegele > could someone of you please tell me which command I need to > get the p-value(s) after doing a (multiple) linear regression > in STATA 8.

* * For searches and help try: * http://www.stata.com/support/faqs/res/findit.html * http://www.stata.com/support/statalist/faq * http://www.ats.ucla.edu/stat/stata/

RE: st: looping

2006-03-20T10:51:00.000-05:00

Correct.

I guess

bysort state county age race : egen totalmales = total(males * perwt) by state county age race : egen totalfemales = total(females * perwt) gen ratio = totalmales / totalfemales

Nick n.j.cox@durham.ac.uk

Scott Cunningham > On Mar 20, 2006, at 9:42 AM, n j cox wrote: > > > You should read some basic accounts of -by:-. > > > > Also, please note the advice in the Statalist FAQ: > > > > "Statalist is an international list. Please explain details that > > may make sense only in your own corner of the world." > > > > Some of the details here presuppose knowledge of your local > > situation, which I guess to be the United States. It is > likely that > > many members of Statalist do not know what "msa" means, for example. > > > > I am still in the dark on what an observation in your dataset > > looks like, or what it represents. I assume that -males- and - > > females- are numeric variables, but are they dummies or do they > > include > > counts? In either case, what is -perwt- precisely? > > Sorry about the ambiguity. I did forget my audience. I am working > with a special sample from the 2000 United States Census called the > longform. It is a sample of households - I think > approximately 5% of > the US population received the survey. An observation is a > household, but since only 5% of all households were sampled, > you must > use the person weights (-perwt-) when working with this data. So, > for instance, if a certain household is sampled and they represent 5 > households, then the perwt=5. Males and females are numeric values. > > > > > Either way, I guess that -egen, total()- with heavy use of -by()- > > or -by:- is the easiest way to get totals of males and females, > > after which you get ratios directly. It should take about 3 lines > > of Stata. The only looping would be that tacit in -egen-. > > I've been using egen to sum males and females, but was unsure of how > to implement the looping over all US county variables. This > is where > -by county- wouuld come in, correct?

* * For searches and help try: * http://www.stata.com/support/faqs/res/findit.html * http://www.stata.com/support/statalist/faq * http://www.ats.ucla.edu/stat/stata/

RE: st: ivreset

2006-03-20T10:21:00.000-05:00

Arne,

> -----Original Message----- > From: owner-statalist@hsphsun2.harvard.edu > [mailto:owner-statalist@hsphsun2.harvard.edu] On Behalf Of > Arne Risa Hole > Sent: 18 March 2006 11:43 > To: statalist@hsphsun2.harvard.edu > Subject: RE: st: ivreset > > Austin, Mark, > > Thank you both for your replies, rescaling the forecast did > the trick (sorry for the bad formatting of my code before, it > looked fine in my email programme). > > I understand the motivation behind the rescaling, but I'm > slightly concerned about the fact that it produces a > different test statistic compared to using the actual > forecast. Note that this is not the case when using the Reset > test following OLS - the test statistic is invariant to the > rescaling in this case. > > I would think that since the two approaches (rescaling/ no > rescaling) produce different results, the correct test > statistic is that using the actual forecast?

There are two different issues here. Rescaling is one, and it is, in some sense, a side issue. The problem is that sometimes the yhat has large-ish values, and the higher order polynomials of yhat that are included in the artificial regression can get so big that the regression fails for numerical precision reasons. Stata's own version of the reset test, -ovtest-, also does this rescaling. The test statistic is, of course, invariant in theory to the units used and hence to rescaling.

The other issue is the one you might have missed. As Austin pointed out, the IV version of the RESET test cannot use standard fitted values that would be generated by -predict- after estimation using -ivreg- or -ivreg2-. These are yhat=X*beta-hat, and the problem is that X includes some endogenous regressors.

As the help file for -ivreset- explains, there are two alternatives. One is to use reduced form predictions for yhat, i.e., regress y on all the exogenous variables (including the excluded instruments) and then use -predict-. The other is to get what Pesaran and Taylor call the "optimal forecast" of yhat. This is not yhat=X*beta-hat, but yhat=X-hat*beta-hat, where X-hat includes the reduced form predicted values of the endogenous regressors (rather than the actual values). The code that Austin kindly posted to Statalast implemented the latter.

Cheers, Mark

> Cheers > Arne > > On Mar 17 2006, Schaffer, Mark E wrote: > > > Austin, Arne, > > > > > -----Original Message----- > > > From: Austin Nichols [mailto:austinnichols@gmail.com] > > > Sent: 16 March 2006 23:30 > > > To: statalist@hsphsun2.harvard.edu > > > Subject: Re: st: ivreset > > > > > > -findit ivreset- then -help ivreset- when installed has a > excellent > > > exposition that begins: > > > > > > As Pagan and Hall (1983) and Pesaran and Taylor (1999) > point out, a > > > RESET test for an IV regression cannot use the standard > IV predicted > > > values X*beta-hat, because X includes endogenous > regressors that are > > > correlated with u. > > > > > > Try this code instead: > > > > > > use http://fmwww.bc.edu/ec-p/data/hayashi/griliches76.dta > > > qui ivreg2 lw s expr tenure rns smsa (iq=med kww) predict > ytilde mat > > > b=e(b) mat li b qui regress iq s expr tenure rns smsa med kww qui > > > predict double xh gen yhat=ytil-b[1,1]*iq+b[1,1]*xh sum yhat, > > > meanonly qui replace yhat = (yhat-r(min))/(r(max)-r(min)) qui gen > > > double yhat2=yhat^2 qui ivreg2 lw s expr tenure rns smsa yhat2 > > > (iq=med kww) test yhat2 qui ivreg2 lw s expr tenure rns > smsa (iq=med > > > kww) ivreset > > > > > > Now, as to why > > > replace yhat = (yhat-r(min))/(r(max)-r(min)) I cannot > tell you, but > > > it's in ivreset.ado > > > > It's to rescale yhat so that when it's squared, cubed, etc., it > > doesn't get wildly out of scale relative to the other regressors. > > This can cause problems for the regression that includes > these terms. > > > > Cheers, > > Mark > > > > > > > On 16 Mar 2006 19:10:13 +0000, Arne Risa Hole wrote: > > > > I am using ivreset to do a Pesaran-Taylor Reset test > after ivreg2. > > > > However, I am not able to replicate the result from ivreset > > > manually. For example: > > > > > > > > use http://fmwww.bc.edu/ec-p/data/hayashi/griliches76.dta > > > > > > > > > > * > > * For searches and help try: > > * http://www.stata.com/support/faqs/res/findit.html > > * http://www.stata.com/support/statalist/faq > > * http://www.ats.ucla.edu/stat/stata/ > > > > * > * For searches and help try: > * http://www.stata.com/support/faqs/res/findit.html > * http://www.stata.com/support/statalist/faq > * http://www.ats.ucla.edu/stat/stata/ > >

* * For searches and help try: * http://www.stata.com/support/faqs/res/findit.html * http://www.stata.com/support/statalist/faq * http://www.ats.ucla.edu/stat/stata/

Re: st: looping

2006-03-20T10:16:00.000-05:00

On Mar 20, 2006, at 9:42 AM, n j cox wrote:

> You should read some basic accounts of -by:-. > > Also, please note the advice in the Statalist FAQ: > > "Statalist is an international list. Please explain details that > may make sense only in your own corner of the world." > > Some of the details here presuppose knowledge of your local > situation, which I guess to be the United States. It is likely that > many members of Statalist do not know what "msa" means, for example. > > I am still in the dark on what an observation in your dataset > looks like, or what it represents. I assume that -males- and - > females- are numeric variables, but are they dummies or do they > include > counts? In either case, what is -perwt- precisely?

Sorry about the ambiguity. I did forget my audience. I am working with a special sample from the 2000 United States Census called the longform. It is a sample of households - I think approximately 5% of the US population received the survey. An observation is a household, but since only 5% of all households were sampled, you must use the person weights (-perwt-) when working with this data. So, for instance, if a certain household is sampled and they represent 5 households, then the perwt=5. Males and females are numeric values.

> > Either way, I guess that -egen, total()- with heavy use of -by()- > or -by:- is the easiest way to get totals of males and females, > after which you get ratios directly. It should take about 3 lines > of Stata. The only looping would be that tacit in -egen-.

I've been using egen to sum males and females, but was unsure of how to implement the looping over all US county variables. This is where -by county- wouuld come in, correct?

* * For searches and help try: * http://www.stata.com/support/faqs/res/findit.html * http://www.stata.com/support/statalist/faq * http://www.ats.ucla.edu/stat/stata/

st: FW: p-value

2006-03-20T10:09:00.000-05:00

> -----Ursprüngliche Nachricht----- > Von: "Jochen Siegele" <jochen.siegele@web.de> > Gesendet: 20.03.06 14:14:22 > An: statalist@hsphsun2.harvard.edu > Betreff: p-value

> Dear STATA-listers, > > > could someone of you please tell me which command I need to get the p-value(s) after doing a (multiple) linear regression in STATA 8. > > Thanks a lot in advance. > > > Jochen

* * For searches and help try: * http://www.stata.com/support/faqs/res/findit.html * http://www.stata.com/support/statalist/faq * http://www.ats.ucla.edu/stat/stata/

st: RE: Wooldridge 3rd edition: worth buying?

2006-03-20T09:57:00.000-05:00

The content between the 2e and 3e is almost identical.

(I own the 2e and tutored someone who has the 3e).

-----Original Message----- From: owner-statalist@hsphsun2.harvard.edu [mailto:owner-statalist@hsphsun2.harvard.edu] On Behalf Of Friedrich Huebler Sent: Monday, March 20, 2006 12:32 AM To: statalist@hsphsun2.harvard.edu Subject: st: Wooldridge 3rd edition: worth buying?

If someone has the second edition of "Introductory Econometrics" by Jeffrey Wooldridge, are there reasons to buy the third edition of this book? I read the book description on the publisher's website and don't see much new in the latest edition. Please share your opinion if you are familiar with both editions.

Thank you,

Friedrich Huebler

* * For searches and help try: * http://www.stata.com/support/faqs/res/findit.html * http://www.stata.com/support/statalist/faq * http://www.ats.ucla.edu/stat/stata/

Re: st: looping

2006-03-20T09:43:00.000-05:00

You should read some basic accounts of -by:-.

Also, please note the advice in the Statalist FAQ:

"Statalist is an international list. Please explain details that may make sense only in your own corner of the world."

Some of the details here presuppose knowledge of your local situation, which I guess to be the United States. It is likely that many members of Statalist do not know what "msa" means, for example.

I am still in the dark on what an observation in your dataset looks like, or what it represents. I assume that -males- and -females- are numeric variables, but are they dummies or do they include counts? In either case, what is -perwt- precisely?

Either way, I guess that -egen, total()- with heavy use of -by()- or -by:- is the easiest way to get totals of males and females, after which you get ratios directly. It should take about 3 lines of Stata. The only looping would be that tacit in -egen-.

Nick n.j.cox@durham.ac.uk

>>> Scott Cunningham

I have the following dataset:

perwt year state msa county males females age race

My goal is to create sex ratios (ie ratios of males to females) for each county. I have not attempted this before because the way that I calculated sex ratios for states and cities (ie, msa) was to write a .do file that summed up the person weights (ie, perwt) for each state/msa. This was not so bad for states, but moreso for cities, since there are 350 cities. The number of counties are in the thousands, and since I have to create sex ratios that differ by race, age and year, it would require writing out a very long .do file. So my question is: is there a looping procedure I could use, which would use the county-level variable as the variable that does the operation? For instance, if there were two counties (1 and 2), then the loop would sum all males of the given age-race criteria, all females of the same age-race criteria, take the ratio and record that for county 1, then repeat for county 2, etc.

* * For searches and help try: * http://www.stata.com/support/faqs/res/findit.html * http://www.stata.com/support/statalist/faq * http://www.ats.ucla.edu/stat/stata/

--------------------------------------------------------------------------------

Previous Article (by Date): st: single failure with mulltiple risk periods Dawn Teele Articles sorted by: [Date] [Author] [Subject]

--------------------------------------------------------------------------------

Go to Harvard School of Public Health LWGate Home Page. * * For searches and help try: * http://www.stata.com/support/faqs/res/findit.html * http://www.stata.com/support/statalist/faq * http://www.ats.ucla.edu/stat/stata/

st: looping

2006-03-20T09:10:00.000-05:00

I have the following dataset:

perwt year state msa county males females age race

* * For searches and help try: * http://www.stata.com/support/faqs/res/findit.html * http://www.stata.com/support/statalist/faq * http://www.ats.ucla.edu/stat/stata/

st: single failure with mulltiple risk periods

2006-03-20T06:28:00.000-05:00

Hello,

I have a question about using the st commands for panel data where a failure can only happen once, but where subjects go in and out of risk.

To be specific, the event in question is the ratification of a specific ILO convention. right now I have used the command:

stset date1, id(panelid) failure(cnv100) origin(joinILO==1)

to tell stata that each country is at risk of ratification only after they have become members of the ILO.

But some of the countries in the sample, for instance the United States, were members at the beginning of the risk period (1976), but then dropped out between 1980 and 1993.

Failure, then, must be conditional on the country actually being a member in each timer period. Do you have any suggestions for how to deal with this?

Thanks!

Dawn Teele Reed College '06 * * For searches and help try: * http://www.stata.com/support/faqs/res/findit.html * http://www.stata.com/support/statalist/faq * http://www.ats.ucla.edu/stat/stata/