| Stunting and maternal education [message #28990] |
Mon, 08 April 2024 16:06  |
Morrilu
Messages: 2
Registered: March 2024
|
Member |
|
|
Hey, I was wondering if someone could clear things up for me. I am looking at the impact of maternal education on the prevalence of stunting in children under 5 in Zimbabwe.
When completing the regression I am unsure about whether to use OLS (regress) or logistic regression (logistic or logit). so could someone please clarify which model is better to capture this relationship.
so far for one of my first regressions I have (Regress stunting educ_yrs wealth_index birth_intervals household_members) where wealth, intervals and household members are all control variables. The coefficient for this is -0.0149. How would you interpret this.
Later on I then use logistic regression as the maternal education is not capturing the level of education attained which is split into 5 sections. Heere I use the stata code (logistic stunting ib2.maternal_education##urban wealth_index birth_intervals household_members).
I get how to interpret the odds ratio as it is the likelihood compared to the reference group.
To summarize, what's the difference between logistic regression (specifically logit) and linear regressions and which one should I use. The second bit is how to interpret the linear regression coefficient.
|
|
|
|
The DHS Program User Forum
Members
Search
Help
Register
Login
Home
