Notice: On April 23, 2014, Statalist moved from an email list to a forum, based at statalist.org.
From | Richard Williams <richardwilliams.ndu@gmail.com> |
To | statalist@hsphsun2.harvard.edu, statalist@hsphsun2.harvard.edu |
Subject | Re: st: Logistic growth curve question |
Date | Sun, 13 Apr 2014 18:55:07 -0500 |
1. Is the Stata dataset easily available somewhere?2. Can you tell if the fits of the models are identical across software packages? Are d.f. the same? Things like that would help to tell you if the models are algebraically equivalent but parameterized differently, or if something must be missing from the Stata model.
At 04:07 PM 4/13/2014, Dave Garson wrote:
For instructional purposes, I am trying to replicate the classic Pothoff & Roy linear growth model in SPSS, SAS, and Stata. While there are variations on this model, the version I am replicating has the following attributes and constraints:1. y1 through y4 are distance measurements at ages 8, 10, 12, and 142. ICEPT is the intercept latent variable, whose slopes are constrained to 1.0. It is modeled as a cause of y1 through y4. 3. SLOPE is the slope latent variable, whose paths are constrained to 8, 10, 12, and 14. It is also modeled as a cause of y1 through y4.4. Error variances are constrained equal. 5. Dependent means are constrained to 0.I get the same results in SAS and SPSS but very different coefficients in Stata, with which I am less familiar. The estat framework postestimation command in Stata shows the five model aspects above are met, but coefficients differ a great deal. Below is my SAS code and my Stata code. My question is, what needs to be changed in the Stata code to give results consistent with SPSS, SAS, and the model described above?SAS SYNTAX: PROC CALIS DATA=in.royfm MODIFICATION EFFPART PLATCOV;/* Above, MODIFICATION requests Lagrange multiplier modification indices *//* Above, EFFPART requests a partition of total effects *//* Above, PLATCOV requests latent variable covariances and score coefficients */PATH /* Below, there are four time periods, y1 - y4 */ y1 y2 y3 y4 <--- ICEPT = 1.0 1.0 1.0 1.0, y1 y2 y3 y4 <--- SLOPE = 8.0 10.0 12.0 14.0;/*Above, intercept paths constrained to 1, slopes are costrained to linear growth */PVAR y1 = variance1, y2 = variance1, y3 = variance1, y4 = variance1, ICEPT = ivariance, SLOPE = svariance; /* Above, time variables are constrained to have equal error variance *//* Above, variances of ICEPT and SLOPE are freely estimated as ivariance and svariance */PCOV ICEPT SLOPE = cov1; /* Above, ICEPT and SLOPE covariance is freely estimated as cov1 */ MEAN y1 = 0, y2 = 0, y3 = 0, y4 = 0, ICEPT = imean, SLOPE = smean; /* Above, time variables constrained to a mean of 0 *//* Above, means of ICEPT and SLOPE are freely estimated as imean and smean */RUN; STATA SYNTAX sem (ICEPT -> y1@1 y2@1 y3@1 y4@1) (SLOPE -> y1@8 y2@10 y3@12 y4@14), latent(ICEPT SLOPE) cov( e.y1@variance1 e.y2@variance1 e.y3@variance1 e.y4@variance1 SLOPE*ICEPT) means(ICEPT SLOPE) noconstant Advice appreciated on list or to david_garson@ncsu.edu Best to all, Dave * * For searches and help try: * http://www.stata.com/help.cgi?search * http://www.stata.com/support/faqs/resources/statalist-faq/ * http://www.ats.ucla.edu/stat/stata/
------------------------------------------- Richard Williams, Notre Dame Dept of Sociology OFFICE: (574)631-6668, (574)631-6463 HOME: (574)289-5227 EMAIL: Richard.A.Williams.5@ND.Edu WWW: http://www.nd.edu/~rwilliam * * For searches and help try: * http://www.stata.com/help.cgi?search * http://www.stata.com/support/faqs/resources/statalist-faq/ * http://www.ats.ucla.edu/stat/stata/