Statalist


[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

st: dropped interaction term in post hoc probing of significant moderation effect using svy regress


From   Ishtar Govia <ishtargovia@gmail.com>
To   statalist@hsphsun2.harvard.edu
Subject   st: dropped interaction term in post hoc probing of significant moderation effect using svy regress
Date   Mon, 30 Mar 2009 13:12:59 -0500

Dear List,

I am using Stata 10/SE 10.1 for Macs. I have a question concerning a dropped interaction term. I am using survey data so I use the -svy- commands

I am doing post hoc probing of a significant interaction effect using the Aiken & West 1991 simple slopes method, as delineated in Holmbeck 2002. I created a two new conditional variables based on the centered moderator (high and low emotional support, variables hiemotspfm and loemotspfm respectively) and then computed interactions of each of those new conditional variables with the independent variable (the variable for the interaction with the high emotional support and IV interaction below is called hiemot_subtle). I then attempted to run 2 post-hoc multiple linear regressions, each including the IV main effect, one of the conditional variables, and the interaction of the IV and conditional moderator variable.

However, I was stumped in running the first of those two post hoc regressions. When I ran the following, however, the interaction term was dropped. I have tried to run the regression with only the new conditional moderator main effect and it works. However, when I try to run the regression with only the interaction between the new conditional moderator and the IV main effect, it still drops the interaction. Does anyone have any ideas why this might have occurred, and more importantly if/how I can make the regression run without having the interaction dropped?

Thanks for your consideration,

Ishtar Govia
ishtargovia@gmail.com

. svy, subpop (s2subpop): regress log_kessler6_sum age female hhinc forgnbrn haitian sepdivwid nevmarr otherstate outofusa C_subtle_avg hiemotspfm hiemot_subtle if !missing(age, female, hhinc, forgnbrn, haitian, sepdivwid, nevmarr, otherstate, outofusa, C_subtle_avg, C_emotspfm, subtle_emot)
(running regress on estimation sample)

Survey: Linear regression

Number of strata = 19 Number of obs = 550 Number of PSUs = 44 Population size = 64.977893 Subpop. no. of obs = 550 Subpop. size = 64.977893 Design df = 25 F( 11, 15) = . Prob > F = . R-squared = 0.2971

------------------------------------------------------------------------------
             |             Linearized
log_kessle~m | Coef. Std. Err. t P>|t| [95% Conf. Interval] ------------- +---------------------------------------------------------------- age | -.0012483 .0018973 -0.66 0.517 -. 0051559 .0026592 female | -.1549944 .0492037 -3.15 0.004 -.2563314 -.0536575 hhinc | -1.88e-06 6.11e-07 -3.08 0.005 -3.14e-06 -6.22e-07 forgnbrn | -.0346815 .0702836 -0.49 0.626 -. 1794334 .1100703 haitian | .0294622 .0609838 0.48 0.633 -. 0961362 .1550606 sepdivwid | .1145513 .0839624 1.36 0.185 -. 0583725 .287475 nevmarr | .1836943 .0487229 3.77 0.001 . 0833476 .284041 otherstate | -.0144171 .0655521 -0.22 0.828 -. 1494241 .1205899 outofusa | .0085318 .0553012 0.15 0.879 -. 1053631 .1224266 C_subtle_avg | .1138018 .0358374 3.18 0.004 .0399933 . 1876103 hiemotspfm | -.0892076 .0468532 -1.90 0.068 -. 1857036 .0072884
hiemot_sub~e |  (dropped)
_cons | 2.25171 .1138259 19.78 0.000 2.017281 2.486139
------------------------------------------------------------------------------




*
*   For searches and help try:
*   http://www.stata.com/help.cgi?search
*   http://www.stata.com/support/statalist/faq
*   http://www.ats.ucla.edu/stat/stata/



© Copyright 1996–2014 StataCorp LP   |   Terms of use   |   Privacy   |   Contact us   |   What's new   |   Site index