The DHS Program User Forum
Discussions regarding The DHS Program data and results
Home » Countries » India » India 2019-2021 Standard DHS survey (Survey not in country report)
India 2019-2021 Standard DHS survey [message #28687] Fri, 23 February 2024 05:00 Go to next message
UnseenDisability is currently offline  UnseenDisability
Messages: 14
Registered: February 2024
Member
Hello everyone,

I would like to do research on disability, specifically hearing loss, which is included in the household characteristics section of the DHS Standard survey. In other DHS standard surveys in which disability questions are included, the variable is called "dhis4" and includes values that represent levels of hearing loss in household members. In the Indian 19-20 survey, this variable does not exist. There is four variables associated with hearing loss, and my understanding is that each one is for one household member other than the responder (only up to four were included - sh33a1a, sh33b1a, sh33c1a, and sh33d1a) and they is labeled as " person 01 (02,03,04):disablity - hearing " and the values are yes and no. I have searched for the actual survey in the country report, but it does not appear to be included.

Can anyone guide me as what this represents? What questions were asked and what constitutes a "yes"? I have already written to the two e-mail addresses on the country report for guidance, but have not received a response. I would be grateful for any guidance.

Thanks,
SA
Re: India 2019-2021 Standard DHS survey [message #28688 is a reply to message #28687] Fri, 23 February 2024 09:46 Go to previous messageGo to next message
fred.arnold@icf.com is currently offline  fred.arnold@icf.com
Messages: 84
Registered: May 2021
Senior Member
You are correct that the information in NFHS-5 on disabilities does not follow the standard DHS practices. The NFHS-5 disability questions were decided on by the Government of India. You can find the NFHS-5 questionnaires in Volume 2 of the final report at here (https://dhsprogram.com/pubs/pdf/FR375/FR375_II.pdf). The relevant questions are in the Household Questionnaire (Questions 31-33). Question 31 asks whether any usual member of your household including yourself has any disability. Note that the respondent to the household questionnaire is included in that question. If there is one or more person with a disability, then Question 32 asks for the name and line number of each person with one of more disabilities. There are four lines in all, so if there are more than four usual residents of the household with a disability, only four persons are included. Finally, for each of the persons in Question 32, the questionnaire asks for each person with a disability (up to four persons), what type of disability the person has. The response categories are as follows:
HEARING A
SPEECH B
VISUAL C
MENTAL D
LOCOMOTOR E
OTHER F

For each person with a disability, one or more of the disabilities may be recorded. If code A is recorded, it means that the person has a hearing disability, but no information is requested about the seriousness of the hearing disability.
Re: India 2019-2021 Standard DHS survey [message #28689 is a reply to message #28688] Fri, 23 February 2024 11:05 Go to previous messageGo to next message
UnseenDisability is currently offline  UnseenDisability
Messages: 14
Registered: February 2024
Member
Thank you so much, this is so helpful - exactly what I needed to know!

SA
Re: India 2019-2021 Standard DHS survey [message #29443 is a reply to message #28688] Thu, 20 June 2024 06:32 Go to previous messageGo to next message
UnseenDisability is currently offline  UnseenDisability
Messages: 14
Registered: February 2024
Member
Now that I have received ethics permission to move on with my project, I have more questions... My understanding is that in other datasets, the hearing impairment questions are located in the household member recode (PR) datasets and that this information is located in the household dataset in India.

Is this correct?

Is this information also available in the household member dataset for India (PR)?

If it's only in the household dataset, then essentially the only information we have is that there is hearing loss in (up to 4) household members, but we have no other individual information about those household members in addition to any information in the household dataset.

I'd like to merge this dataset with two other DHS datasets from other countries (India 2019-2021, Nepal 2022 and Pakistan 2017). Is this feasible? Or should I restructure my project to only use Nepal and Pakistan OR Indian datasets?

thanks in advance for any advice.
Re: India 2019-2021 Standard DHS survey [message #29449 is a reply to message #29443] Thu, 20 June 2024 11:36 Go to previous messageGo to next message
Bridgette-DHS is currently offline  Bridgette-DHS
Messages: 3199
Registered: February 2013
Senior Member
Following is a response from Senior DHS staff member, Tom Pullum:

The disability variables are attached to the household record in the HR file. Unfortunately, in the construction of the PR file for the NFHS-5, the disability variables were just copied onto each record in the PR file. They cannot be used in the current format. This issue came up in April and I prepared a Stata program (attached) to extract and reshape these variables from the HR file, so you can analyze them for individual household members. If you use this program, which is just a text file, you will need to change the paths. The resulting file can be merged back with the PR file so you can get the age, sex, etc., of the people with disabilities.

The cases in other DHS datasets are completely different, so you cannot MERGE the India data with the Nepal data, for example. However, you can compare the surveys.


Re: India 2019-2021 Standard DHS survey [message #29450 is a reply to message #29449] Thu, 20 June 2024 15:53 Go to previous messageGo to next message
UnseenDisability is currently offline  UnseenDisability
Messages: 14
Registered: February 2024
Member
Thank you so much for the swift answer - it is very helpful indeed.

I believe that I used the wrong terminology - when I wrote "merge", I meant "append". If I manage to use your program to reshape the variables, I should be able to append the India PR file to my other two PR files, and then include the data in my analysis. The hearing difficulty variable has different values than the NP and PK datasets, but I can recode them all so that they are binary and useful for my analysis.

Is it correct, then, that once this program is run, I can append the India file to the others?

Thanks in advance
Re: India 2019-2021 Standard DHS survey [message #29454 is a reply to message #29450] Fri, 21 June 2024 09:35 Go to previous messageGo to next message
Bridgette-DHS is currently offline  Bridgette-DHS
Messages: 3199
Registered: February 2013
Senior Member

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

I recommend that you first merge the constructed file with IAPR7EFL.dta, to get relevant covariates, and then yes, you can append the files from the other countries. Note that when you append, the labels will be over-written by the labels from the last file in the append. You may want to do some of your analysis on the separate files.

Let us know if you have any problems. Thanks for using the disability data!
Re: India 2019-2021 Standard DHS survey [message #29484 is a reply to message #29454] Wed, 26 June 2024 04:41 Go to previous messageGo to next message
UnseenDisability is currently offline  UnseenDisability
Messages: 14
Registered: February 2024
Member
Thank you!

It is quite disappointing that the disability data is used so rarely. Of course, there are perhaps other issues that are considered more pressing, but I hope to be able to contribute to disability research.

I see this line in the do file: * Only 26,435 unweighted cases with any kind of disability

What does this mean for analysis? I am appending two other datasets (Pakistan and Nepal) to include in my analysis. Should I not weigh the data? The script I have currently included is as follows:
*Set the survey design

gen weight = hv005/1000000 //scaling the weights if needed --> DO I NEED TO?
svyset [pw=weight], psu(hv021) strata (hv023)

Do you have a recommendation on this?

In my analysis, hearing loss (a recategorization of the variable in PK and NP to coordinate with the binary nature of the IN dataset) is my primary outcome.

Thank you in advance for advice on this.
Re: India 2019-2021 Standard DHS survey [message #29485 is a reply to message #29484] Wed, 26 June 2024 10:55 Go to previous messageGo to next message
Bridgette-DHS is currently offline  Bridgette-DHS
Messages: 3199
Registered: February 2013
Senior Member

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

Yes, you should use the weights and svy adjustments as you described. I would not rescale the weights. If you rescale the weights to represent the pooled population, India will overwhelm the other countries. I suggest giving distributions separately for each country.

I inserted the comment "only 26,435" because that number seemed low in a sample with 2,843,917 cases in the PR file (although not everyone is eligible for the disability questions). In your comparisons with other countries, it will be important to say something about the proportion of the total sample that is reported to have a disability, as well as analyzing the types of disabilities and their relation to covariates such as age and sex.

Re: India 2019-2021 Standard DHS survey [message #29487 is a reply to message #29485] Wed, 26 June 2024 17:32 Go to previous message
UnseenDisability is currently offline  UnseenDisability
Messages: 14
Registered: February 2024
Member
Thank you very much!
Previous Topic: Regarding skip pattern in occupation questions (IR) file of DHS 5 India
Next Topic: MR file in NFHS1 and NFHS2
Goto Forum:
  


Current Time: Sun Nov 24 05:22:24 Coordinated Universal Time 2024