Status:Open    Asked:Feb 10, 2019 - 01:00 PM

3 Level Hierarchical Models using IPUMS data

I am not sure if this is the right place to ask this question. I am putting it here as I am using IPUMS data. There might be something about the data structure which I don't understand and might be creating the issue. Several similar studies have been done in the past using DHS data but no one faced similar issues.

I have pooled 3-4 waves each from 25 countries making the total samples as 95. No PSU information is missing.

About the model

My dependent variable is neghaz (negative of height for age (cm/months)) which is continuous in nature. My regression specification includes several control variables including square terms and interaction terms. The specification also includes variables that have been calculated at PSU/cluster level (Mean Employment Rate in the Cluster, etc) and also variables at country level (GDP, Average Life Expectancy etc.).


I am trying to evaluate the following 3 level hierarchical model (respondents <- clusters <- surveys)

mixed neghaz $controlset [pw=perweight] || idhspsu: || sample:

The model failed to converge. After that I tried a null model. The null model also failed to converge. I am not able to understand why null model fails to converge when there are 95 surveys and every survey has 300 clusters at least.

I also tried the null model after converting neghaz in to a dichotomous variable (xtmelogit) stunted which takes the value 1 if the child is stunted (chronic malnutrition). The converge failed again

Afterwards, I tried running 2 level models with PSUs and Surveys independently. The models worked with the full control set. However, the standard errors were different in the two models.

ICC for model with PSU – 0.98; ICC for model with survey – 0.02

Can somebody help me to understand why is convergence failing and how to fix it?

Can I safely neglect the survey random effects in this case?

Is there any other way of combining the survey effects (Random /Fixed) along with the PSU random effects?

I also tried models with only survey fixed effects (i.survey with normal ols) However, the standard errors were different. What model shall I finally use in such a case?

Sorry for the long post.

Do you have the same question? Follow this Question


Sorry, I am not equipped to answer this question. I suggest that you post your question to the DHS Users Forum; perhaps someone there can answer it. The content and structure of the variables in your model should be the same, regardless of whether you are using the original DHS files or the IPUMS version of the data.

Good luck.


Feb 10, 2019 - 01:11 PM

Report it


negative values of wtsupp 1975 and before
Why are not all the "not in the labour force" categories filled for the EMPST...
I am getting strange numbers for NIU persons in the sixties, especially 1962-...
Health Insurance Unit in ACS versus MEPS
Login   |   Register

Recently Active Members

View More »

Share |