The DHS Program User Forum
Discussions regarding The DHS Program data and results
Home » Countries » Egypt » 2015 dataset, female genital cutting (FGC 2015)
2015 dataset, female genital cutting [message #28822] Thu, 14 March 2024 14:25 Go to next message
gusweseb is currently offline  gusweseb
Messages: 6
Registered: March 2024
Member
Hi!
I'm currently working on my thesis exploring if there is an association between attitudes among women, the perception of male attitudes, and further women's intentions to circumcise their daughters. The analysis is performed using logistic regression for years between 2000-2015 and it works well for years 2000-2014 however, y.2015 differs from the rest. First, variable c.307 " circumcision of (name) in the future" differs from how the question is formulated in the other year's datasets; "intends to have daughter(s) circumcised in future". However, I have renamed them all as "circum" and further separated the variables for attitudes among women and men and the perceptions of male attitudes among women and men only to include women's attitudes. Even though it looks like the coding is the same between the years it does not work to run a LR using the dataset from 2015. The result says "no observations" r(2000)when including the variable "circum". Additionally, it does not work to run a Logistic regression comparing men's attitudes towards the continuation of FGC to women's perceptions of male attitudes for the year 2015.

What could I possibly do wrong here?
Grateful for help.
Best, Ebba
Re: 2015 dataset, female genital cutting [message #28844 is a reply to message #28822] Mon, 18 March 2024 15:28 Go to previous messageGo to next message
Bridgette-DHS is currently offline  Bridgette-DHS
Messages: 3028
Registered: February 2013
Senior Member

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

Are you using the Egypt 2015 survey? That was a "Special" survey. It does not appear to have standard recode files or a final report. I find an "IQ" file with a series of variables a503 to a517 related to FGC.

If you rename and recode these variables to match those in the standard surveys, and append the files, you should be able to make the kinds of comparisons you want. I don't know why you are getting the message "no observations" for your logit regression, but there are many possibilities. If you are using svyset, for example, it is possible that you are asking for a variable that is not defined, or is NA, for this survey. Can you provide more information about how you are pooling the surveys?

Re: 2015 dataset, female genital cutting [message #28848 is a reply to message #28844] Tue, 19 March 2024 04:36 Go to previous messageGo to next message
gusweseb is currently offline  gusweseb
Messages: 6
Registered: March 2024
Member
Alright, so is it possible to perform any kind of analysis with the data from 2015?
So far I have tried to append children's data (containing information about mother's intention to circumcise their daughters) with women's datafile (containing attitudes towards continuing FGC among women and men).

My two questions are:
1. Is there an association between women's attitudes, the perception of male attitudes, and the intention to circumcise their daughter?
2. Do men's attitudes towards continuing FGC deviate from women's perceptions of male attitudes towards continuing FGC?

I have split the variable for Attitudes and perceptions so it contains observations for women and men separately. However performing LR for Attitudes, perceptions, and intentions does not work. LR with Attitudes and perceptions work fine however it does not seem like the same women are asked about attitudes, perceptions, and intention for circumcising? That might be the problem, or should I be able to perform this analysis for 2015? it has worked well in the other datasets.

The other problem is that I cannot do a regression comparing men's attitudes towards continuing FGC with the perception of men's attitudes among women. This should work, as the variables are still in the same dataset but somehow it also says "no observations".
Re: 2015 dataset, female genital cutting [message #28867 is a reply to message #28848] Wed, 20 March 2024 02:26 Go to previous messageGo to next message
gusweseb is currently offline  gusweseb
Messages: 6
Registered: March 2024
Member
Hi!
I'm looking at Egypt from y. 1995-2015 though I only have problems with the data from y.2015. That's strange I can find data covering men's attitudes in file EGIQ73FL.DTA (Women's file). It seems like there is in the variable questioning "Should the practice of FGC continue or stop?". I separated the variable by gender and did the same for "Do you think the other gender want this practice to continue or stop?" to get women's perception of male attitudes. I then took the mean of the two variables to compare proportions since performing an LR was not possible. Would this be an appropriate way to analyze the data?

The children's file is named EGCH73DT.DTA and my supervisor advised me to match the data based on the mother's line number to compare intention to circumcise with attitudes and perception of attitudes hence these variables are located in the women's file. However, I have some trouble with how to do this if you have any advice that would be greatly appreciated.

I hope I made myself clear and that we have the same data, otherwise, I'm a bit confused over what data I have found.
Re: 2015 dataset, female genital cutting [message #28868 is a reply to message #28867] Wed, 20 March 2024 08:49 Go to previous messageGo to next message
Bridgette-DHS is currently offline  Bridgette-DHS
Messages: 3028
Registered: February 2013
Senior Member

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

I see a511 and a512 in the IQ file and I think they are the two variables you are referring to. a511 ("shall the practice of female circumcision should be continued or stopped") gives the woman's personal attitude. a511 ("do you think that the other gender wants this practice to continue or to stop") does not refer specifically to the husband, but to men, in general. At any rate, a512 is a perception. It cannot be interpreted as the actual attitude of the husband or of men in general.

I have looked at the CH file. You can easily put the mother's data onto the child's record, with the following two lines:

use "...EGCH73FL.DTA" , clear
merge 1:m hpsu hnumber eligline using "...EGIQ73FL.DTA"

You would save this file with a new name, perhaps EGBR73FL.DTA, although this is definitely not a standard BR file. I see that it includes children in the age range 0-14.

We are still checking whether there were interviews with men as part of the 1995 survey. We will post a followup response as soon as possible.
Re: 2015 dataset, female genital cutting [message #28870 is a reply to message #28868] Wed, 20 March 2024 10:02 Go to previous messageGo to next message
gusweseb is currently offline  gusweseb
Messages: 6
Registered: March 2024
Member
Hi! Thanks a lot! I tried to match the dataset based on this though it only says; variables hpsu hnumber eligline do not uniquely identify observations in the master data".

What may I do wrong?

Best, Ebba
Re: 2015 dataset, female genital cutting [message #28871 is a reply to message #28868] Wed, 20 March 2024 12:00 Go to previous messageGo to next message
gusweseb is currently offline  gusweseb
Messages: 6
Registered: March 2024
Member
Hi! I finally managed to merge the datasets only to get this output:


Result Number of obs
-----------------------------------------
Not matched 27,549
from master 10,878 (_merge==1)
from using 16,671 (_merge==2)

Matched 0 (_merge==3)
-----------------------------------------

What might this mean?
Best, Ebba
Re: 2015 dataset, female genital cutting [message #28879 is a reply to message #28871] Thu, 21 March 2024 11:20 Go to previous messageGo to next message
Bridgette-DHS is currently offline  Bridgette-DHS
Messages: 3028
Registered: February 2013
Senior Member
Following is a response from Senior DHS staff member, Tom Pullum:

Since my last response to your questions, I have reached a better understanding of this survey. I had thought that the IQ file was a file of women, but it actually includes men as well as women. Sex is given by the variable "agend". (Maybe you already knew this, but I didn't!)

You can compare attitudes and perceived attitudes for women and men with tables such as the following:
tab a511 agend [iweight=weight/1000000], col
tab a511 a512 if agend==1 [iweight=weight/1000000], col
tab a511 a512 if agend==2 [iweight=weight/1000000], col
It appears that men are more favorable to FGC than women, but not by a large margin. With these tables you can compare the stated attitude for men with the women's perception of the men's attitude. However, you cannot cross-tabulate those variables because the stated attitude for men comes from the cases with agend=1 and the women's perception of the men's attitude comes from cases with agend=2.

You cannot reliably identify women and men who are partners, and form a couples (CR) file, but I believe you can find other ways to compare women and men in the same household or subpopulation.
Re: 2015 dataset, female genital cutting [message #28887 is a reply to message #28879] Fri, 22 March 2024 16:04 Go to previous message
gusweseb is currently offline  gusweseb
Messages: 6
Registered: March 2024
Member
Thanks yes, that's what I have been doing :)
Previous Topic: issue with Husband/partner age
Goto Forum:
  


Current Time: Sat Apr 13 09:46:11 Coordinated Universal Time 2024