The DHS Program User Forum: Child Health

Home » Topics » Child Health » Child vaccination (Multinomial Regression)

Show: Today's Messages :: Show Polls :: Message Navigator

Child vaccination [message #29040]

Wed, 17 April 2024 00:24

shruti.acharya
Messages: 5
Registered: September 2023

Member

I am new to NFHS data as well as STATA. I am trying to understand the association of socio-demographic factors with childhood immunization status (which in my case is a categorical outcome with 3 categories). I am using multinomial logistic regression on STATA using the following code :
mlogit vacc_status(i.sex i.birth_order) (i.Wealth i.religion i.caste i.household_size i.v106) (i.media_exposure i.age_firstbirth )

I have recoded the variables for better identification. I wanted to know if I need to use the "svyset linearized" everytime I run my regression on STATA, and also whether I am doing it correctly.

Thanks in advance !

S Acharya

Report message to a moderator

Re: Child vaccination [message #29063 is a reply to message #29040]

Thu, 18 April 2024 16:38

Janet-DHS
Messages: 702
Registered: April 2022

Senior Member

Thank you for submitting your question. Could you please provide some more detailed information so we can better advise you?
Can you let us know:
• Which survey you are using (Include country name and year)?
• Which data files you are referring to?
• Which software you are using (Stata, SPSS, R, etc.)?
If you are trying to match a Table in a final report, please also indicate which table and which estimate you are trying to match.

Report message to a moderator

Re: Child vaccination [message #29065 is a reply to message #29063]

Thu, 18 April 2024 21:13

shruti.acharya
Messages: 5
Registered: September 2023

Member

Hi Janet !
I am using DHS data for India (NFHS-5) and using STATA for data analysis. I am using the KR (Child's RECODE) file. My outcome variable is vaccination status ( no vaccination, partial vaccination and full vaccination), which I have created referring and modifying the STATA code available on GIThub.

I am using multinomial logistic regression to understand the association between socio-demographic variables and vacc_status on STATA using the following code :
mlogit vacc_status(i.sex i.birth_order) (i.Wealth i.religion i.caste i.household_size i.v106) (i.media_exposure i.age_firstbirth )

I have recoded the variables for better identification. I wanted to know if I need to use the "svyset linearized" everytime I run my regression on STATA, and also whether I am doing it correctly.

Thanks
S Acharya

Report message to a moderator

Re: Child vaccination [message #29096 is a reply to message #29065]

Tue, 23 April 2024 13:52

Janet-DHS
Messages: 702
Registered: April 2022

Senior Member

Following is a response from DHS staff member, Tom Pullum:

If you specify svyset before an estimation command (such as mlogit) and then put "svy: " at the beginning of the estimation command (for example, "svy: mlogit..."), you do not have to re-specify svyset for the next estimation. The specification of svyset lasts until it is changed. The current svyset specification will be included in the save, if you save the dta file, which may or may not be a good thing.

Hope you know that "svyset linearized" would not be correct. You want something more like "svyset v001 [pweight=v005], strata(v022) singleunit(centered)". You could add "vce(linear)" after the comma but it's usually not necessary.

Your command, as given, will produce (if It even runs) a huge number of coefficients that will be difficult to interpret. I would recommend that you first try simpler models with just one or two covariates as a time.

Report message to a moderator

Re: Child vaccination [message #29099 is a reply to message #29096]

Tue, 23 April 2024 20:55

shruti.acharya
Messages: 5
Registered: September 2023

Member

Thank you for your response

Report message to a moderator

Previous Topic:

child immunization

Goto Forum:

-=] Back to Top [=-

[ Syndicate this forum (XML) ] [

] [

]

Current Time: Fri May 3 04:23:42 Coordinated Universal Time 2024