The DHS Program User Forum: Other countries

Home » Countries » Other countries » Colombia Gender Scale (Source of Gender scale)

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

Colombia Gender Scale [message #26379]

Tue, 14 March 2023 17:01

Deidi
Messages: 14
Registered: March 2021

Member

Hi,
I would like to ask about the source of the questions in DHS Colombia serie s1302 (e.g. agrees on statement: men are head of households; agrees on statement: women need men to be happy.) I haven't been able to identify the source of them. Any leads would be much appreciated. Thanks!

[Updated on: Tue, 14 March 2023 17:05]

Report message to a moderator

Re: Colombia Gender Scale [message #26403 is a reply to message #26379]

Thu, 16 March 2023 11:19

Janet-DHS
Messages: 882
Registered: April 2022

Senior Member

Following is a response from our DHS Team, Senior Demographic and Health Research Analyst, Jeffrey Edmeades, DHS Research & Data Analysis Director, Kerry MacQuarrie, and DHS Chief of Data Processing, Guillermo Rojas:

The questions in section 13 of the questionnaire appear to be drawn mainly from the Gender Equitable Men Scale (GEMS) English version here https://www.equimundo.org/resources/measuring-gender-attitud e-using-gender-equitable-men-scale-gems-in-various-socio-cul tural-settings/. We are not aware of an established scale for question 1302. We know that some gender advocacy groups participated in the design of the Colombia DHS 2015, but we cannot reconstruct the development of the questionnaire. The Colombia surveys are conducted almost completely independently from The DHS Program and USAID.

Report message to a moderator

Re: Colombia Gender Scale [message #26721 is a reply to message #26403]

Sat, 22 April 2023 21:44

Deidi
Messages: 14
Registered: March 2021

Member

Thank you, this is helpful!

One more question: Could d109 be used as a time variable for survival analysis? I'm a little confused because I contrasted it with the IPV variable and it seems to include both those who had experienced violence and those hadn't. Thank you so much for any help with understanding this. Thanks!

Report message to a moderator

Re: Colombia Gender Scale [message #26758 is a reply to message #26721]

Wed, 26 April 2023 14:27

Janet-DHS
Messages: 882
Registered: April 2022

Senior Member

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

Yes, you could use d109 as the time variable in a survival analysis. It's ok that it includes those who have not experienced violence. You could say that those who have not experienced it have been censored. But you do not want to set up a model which assumes that eventually all women will experience domestic violence. Some percentage (a large one, I hope!) will never experience it, just as, say, some women will never have a first birth, but you can still apply a survival model.

Report message to a moderator

Re: Colombia Gender Scale [message #27742 is a reply to message #26758]

Wed, 27 September 2023 16:11

Deidi
Messages: 14
Registered: March 2021

Member

Good afternoon,

Thanks for your response. That totally makes sense. I have a follow up question for d109. I haven't being able to figure out the following: 1) Whether for a survival model, the value 0 should be coded as 1, or keep it the way it is. 2) How to solve the issue of missing values. If I omit them in the stset set up (e.g, stset years [pw = weight] if years !=.,...), it doesn't give me an error anymore, but then the issue reappears when I'm evaluating the model fit (predict cs, csnell). Is there perhaps an additional restriction (e.g. dropping some other vars) I should run so I don't have such a high number of missing obs? Any other leads, would be much appreciated. Thanks in advance.

timing of first event |
(from d105 series) in |
years after marriage | Freq. Percent Cum.
-------------------------+---------------------------------- -
0 | 6,130 36.78 36.78

.keep if v044==1 & v502>0 // (13,828 observations deleted)

. stset years [pw = weight], failure(ipv_phy_sex==1)

Survival-time data settings

Failure event: ipv_phy_sex==1
Observed time interval: (0, years]
Exit on or before: failure
Weight: [pweight=weight]

------------------------------------------------------------ --------------
24,890 total observations
8,375 event time missing (years>=.) PROBABLE ERROR
------------------------------------------------------------ --------------
16,515 observations remaining, representing
8,567 failures in single-record/single-failure data
47,273 total analysis time at risk and under observation
At risk from t = 0
Earliest observed entry t = 0
Last observed exit t = 30

Report message to a moderator

Re: Colombia Gender Scale [message #27754 is a reply to message #26721]

Fri, 29 September 2023 15:47

Janet-DHS
Messages: 882
Registered: April 2022

Senior Member

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

d109 is a timing variable, based on when the first event of d105a -d105h occurred (if any of them occurred). The NA cases are the women for whom d105a to d105h are NA. If you do "tab d105*,m" you will see that for each of the d105* variables there are 13,828 NA cases. In your work file you can just drop these cases (for example, with "drop if d105a==."). I would also drop the 152 women who are "don't know" for d109, but that's a judgment call. Then do "tab d109,m" and you see that there are 8,223 women who are censored, in the sense described before. They have not experienced violence up to the time of the survey but they are at risk of experiencing it in the future. You would construct a new variable that is 0 if d109 is "." (for those 8,223 censored cases) and 1 otherwise. (I have not used survival models myself recently and hope that this is not misleading.)

I found a youtube video that may help: https://www.youtube.com/watch?v=HnsJG42LxMo. I'm sure there are others, probably some using DHS data.

Report message to a moderator

Re: Colombia Gender Scale [message #28877 is a reply to message #26403]

Thu, 21 March 2024 08:57

Deidi
Messages: 14
Registered: March 2021

Member

Dear DHS Program Team and fellow forum users,

I hope all is well.

I'm writing because I have a question about the Colombia DHS data 2015-16. I'm using the couples dataset and the set of questions SM1054A-SM1054I about justification of IPV is consistently missing 320 observations. I was not able to find any skip pattern in the questionaire. I also checked the MR dataset and it also has a similar pattern of missingness. I wonder whether this set of questions was applied only to a specific sample? Any additional information would be much appreciated.

Thanks in advance for your help.

Best,

Deidi

[Updated on: Thu, 21 March 2024 09:37]

Report message to a moderator

Re: Colombia Gender Scale [message #28890 is a reply to message #28877]

Fri, 22 March 2024 16:07

Janet-DHS
Messages: 882
Registered: April 2022

Senior Member

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

I see that in the MR file, 554 men have NA (a dot in Stata) for sm1054a, etc., as well as for many, if not most, of the other sm105* variables. In the CR file, this reduces to 320 cases. This is less than 2% of the men, which is too small for it to be due to subsampling. I have checked and do not find any relationship with marital status. If an explanation can be found, we will post a followup.

Report message to a moderator

Re: Colombia Gender Scale [message #28899 is a reply to message #28877]

Tue, 26 March 2024 09:40

Janet-DHS
Messages: 882
Registered: April 2022

Senior Member

Following is a response from DHS staff member, Tom Pullum:
We checked with Guillermo Rojas, the DHS staff member most familiar with all of the Colombia surveys. Here is his response: "The 554 NA cases correspond to respondents for whom privacy was not possible. In question 1001 interviewers record whether there was privacy at the time of starting the "Gender Violence" section. If privacy was not granted the whole section is skipped."

These cases are probably not "missing at random (MAR)" but there are so few of them, in percentage terms, that analyses based on the non-missing cases will have a negligible bias.

Report message to a moderator

Re: Colombia Gender Scale [message #28901 is a reply to message #28899]

Tue, 26 March 2024 10:28

Deidi
Messages: 14
Registered: March 2021

Member

Thank you very much for you response and information. I really appreciate it.

Best regards,

Deidi

Report message to a moderator

Previous Topic:	Namibia 2013, mental health descriptives for 50-64 year olds
Next Topic:	Mauritania DHS 2019-2021 refugees and asylum seekers

Goto Forum:

-=] Back to Top [=-

[ Syndicate this forum (XML) ] [

] [

]

Current Time: Thu Nov 14 18:58:01 Coordinated Universal Time 2024