Re: st: logistic regression and interpreting trend across two time points

[Ángel Rodríguez Laso <angelrlaso@gmail.com>]

Ameya,

Because you work with different individuals in both surveys, you
cannot explore the association between change in breastfeeding
practices and malnutrition. You would need longitudinal data from the
same individuals to do that.

You can only describe the trends in the surveys and see if the
differences in percentages are statistically significant.

Best regards.

Angel Rodriguez-Laso

2012/11/11 Ameya Bondre <ameyabondre.jhsph@gmail.com>:
> Hello,
>
> I am studying the change in malnutrition as a response to the change
> in prevalence of exclusive breastfeeding (and a few other covariates),
> from 2009 to 2011.
>
> Both the 2009 and 2011 data are cross-sectional and independent.
> During both rounds of data collection, each respondent is a mother
> with a child 6-24 months of age. The study involves evaluating the
> sustenance of program activities in 2011, two years after the program
> ended in 2009.
>
> The data set that I am using has these binary variables - (endline =
> 2009 and followup = 2011)
>
>  underwt_endline - child measured underweight at 2009 (yes/no)
>  underwt_followup - child measured underweight at 2011 (yes/no)
>  exclbf_endline - child exclusively breastfed as told by mother (yes/no)
>  exclbf_followup - child exclusively breastfed as told by mother (yes/no)
>
> There are other similar predictors but for this email, I am limiting
> it to just one predictor.
>
> I am not able perform a logistic regression here, that is
> interpretable. I have the absolute percentages for these indicators
> for both 2009 and 2011:
>
> Prevalence of Exclusive breastfeeding increased from 30% in 2009 to 47% in 2011.
> Prevalebce of Underweight decreased from 40% in 2009 to 33% in 2011.
>
> My question is - How can we test for the association between the
> percentage change in prevalence of underweight and the percentage
> change in prevalence of exclusive breastfeeding, between 2009 and
> 2011? Can it be done through logistic regression?
>
> Thank you,
> Ameya Bondre
> *
> *   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/
*
*   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/