<!DOCTYPE HTML PUBLIC "-//W3C//DTD HTML 3.2//EN">
<HTML>
<HEAD>
<META HTTP-EQUIV="Content-Type" CONTENT="text/html; charset=iso-8859-1">
<META NAME="Generator" CONTENT="MS Exchange Server version 6.5.7638.1">
<TITLE>Counterintuitive results</TITLE>
</HEAD>
<BODY>
<!-- Converted from text/plain format -->
<BR>
<P><FONT SIZE=2>Dear Subscribers,<BR>
<BR>
I've carried out the following model:<BR>
<BR>
> library("sandwich")<BR>
> set.seed(2014)<BR>
> med.fit <- glm(estuprimas ~ edad_c + sexo + regalf + deprinf, family="binomial" ,data=child65)<BR>
> out.fit <- glm(benvii ~ edad_c + sexo + regalf + deprinf + estuprimas, family="binomial" ,data=child65)<BR>
> med.out <- mediate(med.fit, out.fit, treat = "deprinf", mediator = "estuprimas", robustSE = TRUE, sims=1000, control.value = "no", treat.value = "s\xed")<BR>
> summary(med.out)<BR>
<BR>
Causal Mediation Analysis<BR>
<BR>
Quasi-Bayesian Confidence Intervals<BR>
<BR>
Estimate 95% CI Lower 95% CI Upper p-value<BR>
ACME (control) -0.012543 -0.029801 0.000732 0.07<BR>
ACME (treated) -0.011498 -0.027808 0.000723 0.07<BR>
ADE (control) -0.049137 -0.125894 0.031768 0.27<BR>
ADE (treated) -0.048092 -0.121987 0.031410 0.27<BR>
Total Effect -0.060635 -0.137208 0.021339 0.15<BR>
Prop. Mediated (control) 0.169334 -0.848191 1.686179 0.21<BR>
Prop. Mediated (treated) 0.147997 -0.854962 1.689611 0.21<BR>
ACME (average) -0.012021 -0.028888 0.000728 0.07<BR>
ADE (average) -0.048615 -0.124610 0.031589 0.27<BR>
Prop. Mediated (average) 0.158665 -0.851577 1.687895 0.21<BR>
<BR>
Sample Size Used: 657<BR>
<BR>
Simulations: 1000<BR>
<BR>
<BR>
If I understand well, this means that, of the total association of having experienced infant deprivation vs not having experienced it (variable deprinf) on having aged well (-0.06 (OR=0.94; CI95%= 0.87-1.02)) , 15.9% is via the mediator having primary education or more vs less than primary education (estuprimas). This does not make sense to me, because having primary education or more is a protective factor for aging well, as you can see in this model:<BR>
<BR>
Call:<BR>
glm(formula = benvii ~ edad_c + sexo + regalf + deprinf + estuprimas,<BR>
family = "binomial", data = child65)<BR>
<BR>
Deviance Residuals:<BR>
Min 1Q Median 3Q Max <BR>
-1.6054 -0.8433 -0.5597 1.0327 2.6291 <BR>
<BR>
Coefficients:<BR>
Estimate Std. Error z value Pr(>|z|) <BR>
(Intercept) 7.29876 1.20323 6.066 1.31e-09 ***<BR>
edad_c -0.10262 0.01619 -6.339 2.31e-10 ***<BR>
sexomujer -0.89359 0.18639 -4.794 1.63e-06 ***<BR>
regalfMedia -0.63054 0.27055 -2.331 0.01977 * <BR>
regalfBaja -0.71116 0.23645 -3.008 0.00263 **<BR>
deprinfsí -0.29268 0.23603 -1.240 0.21497 <BR>
estuprimas 0.33773 0.18910 1.786 0.07411 . <BR>
<BR>
<BR>
I'd appreciate some insight into this result.<BR>
<BR>
Thank you very much.<BR>
<BR>
Angel Rodríguez-Laso</FONT>
</P>
</BODY>
</HTML>