Re: regression to explore the relationship between neonatal mortality & maternal health seeking behavioo [message #20102 is a reply to message #20094] |
Fri, 25 September 2020 09:08 ![Go to previous message Go to previous message](/theme/default/images/up.png) |
Bridgette-DHS
Messages: 3107 Registered: February 2013
|
Senior Member |
|
|
Following is a response from DHS Research & Data Analysis Director, Tom Pullum:
My recommendation would be that you go ahead with a preliminary logit regression, including the weights and survey design effects, and look at the standard errors and confidence intervals in the results. You may be able to get a significant coefficient at the national level, but I have my doubts.
In logit regression, the standard error of the coefficient depends as much on the number of "yes" cases as on the number of "no" cases, and not just on the total number of cases. It's good that you are being cautious.
An important issue in this relationship is women who have difficulties during the pregnancy, that is, a higher risk pregnancy, may be more likely to seek care. This can flip the sign of the coefficient, making it appear that health care has had a negative effect.
|
|
|