Children nutritional status [message #29748] |
Tue, 30 July 2024 12:41 ![Go to next message Go to next message](/theme/default/images/down.png) |
Benn88
Messages: 3 Registered: November 2023
|
Member |
|
|
Hello everyone,
My name is Benn,
As an academic researcher investigating the determinants of stunting, wasting, and underweight among children under five in Haiti, I have a few clarifying questions regarding the analysis. I am utilizing the Haiti Demographic and Health Survey data, specifically the KR file.
1. When analyzing maternal body mass index as a potential determinant, should I also exclude pregnant women and women who gave birth in the preceding 2 months? I am unsure if the BMI values reported in the final DHS report have already been adjusted for these considerations.
2. In the logistic regression modeling, is it appropriate to include both birth order and birth interval as independent variables? When computing birth interval, the common practice is to exclude the first-born children. Would there be any issues with including both of these variables in the same model?
I appreciate your guidance on these methodological aspects to ensure the robustness of my analysis and the validity of the findings. Any clarification you can provide would be greatly helpful as I work on my first academic paper using this dataset.
Thank you very much
|
|
|
Re: Children nutritional status [message #29768 is a reply to message #29748] |
Wed, 31 July 2024 11:49 ![Go to previous message Go to previous message](/theme/default/images/up.png) |
Janet-DHS
Messages: 787 Registered: April 2022
|
Senior Member |
|
|
Following is a response from DHS staff member, Tom Pullum:
You do need to omit BMI for women who are pregnant or gave birth in the past two months. There is nothing in the coding of the woman's BMI that adjust for that. Mother's BMI would be coded NA for mothers who are pregnant or gave birth in the past two months. Details are given in the Guide to DHS Statistics.
You can include both birth order and the preceding birth interval--those variables are not collinear. However, as you say, birth interval is Not Applicable for bord=1, so first births will be dropped.
With these two types of omissions you will lose a substantial fraction of the cases. However, I don't think there is a good way to avoid that. I would not recommend imputing the NA values. In the part of your analysis that does not include mother's BMI and/or preceding birth interval, you will not lose cases.
|
|
|