The DHS Program User Forum: Other countries » Labeling mistakes in exposure to mass media variables in Zambia 2007 stata dataset

Home » Countries » Other countries » Labeling mistakes in exposure to mass media variables in Zambia 2007 stata dataset

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

Labeling mistakes in exposure to mass media variables in Zambia 2007 stata dataset [message #2663]

Sun, 03 August 2014 12:00

ykang12
Messages: 8
Registered: July 2014
Location: USA

Member

I am Yunhee, graduate student of Johns Hopkins School of Public Health.
I have been doing data analysis with Zambia 2007 DHS dataset (stata format) and I wanted to know exposure to mass media among adolescent groups by tabulating such variables (v157,v158, v159). But I found labeling mistakes below and questions/answers seemed to be inconsistent with questions in DHS questionnaires. Could you help me clear about this?

< STATA dataset>
. tab v157

frequency of reading |
newspaper or magazine | Freq. Percent Cum.
------------------------+-----------------------------------
no education, preschool | 883 55.60 55.60
primary | 233 14.67 70.28
secondary | 261 16.44 86.71
higher | 209 13.16 99.87
9 | 2 0.13 100.00
------------------------+-----------------------------------
Total | 1,588 100.00

: their answers were recorded as 0,1,2,3,9, which is not consistent with Zambia questionnaire itself and even compared with other countries' dataset.
< DHS questionnaires>
Do you read a newspaper or magazine almost every day, at least once a week, less than once a week or not at all?
ALMOST EVERY DAY . . . . . . . . . . .1
AT LEAST ONCE A WEEK . . . . . . . . .2
LESS THAN ONCE A WEEK . . . . . . . . 3
NOT AT ALL . . . . . . . . . . . . . .4

Report message to a moderator

Re: Labeling mistakes in exposure to mass media variables in Zambia 2007 stata dataset [message #2668 is a reply to message #2663]

Mon, 04 August 2014 08:45

Bridgette-DHS
Messages: 3208
Registered: February 2013

Senior Member

I am looking at the frequencies in the file: zmir51fl.dta, and can't see what you have in your post. Please send me the complete name of the file you are using, and I will run some frequencies to compare with the result you are having.

[Updated on: Mon, 31 August 2020 16:23]

Report message to a moderator

Re: Labeling mistakes in exposure to mass media variables in Zambia 2007 stata dataset [message #12649 is a reply to message #2668]

Wed, 28 June 2017 09:57

dab000@mail.harvard.edu is currently offline

dab000@mail.harvard.edu
Messages: 2
Registered: July 2016

Member

Hello Bridgette,

I am having similar problems to what is described above for V157-V159 of the Kenya DHS for 2003 and 2008 - the response options in the dataset range from 0 to 3, rather than the options in the questionnaires, which range from 1 to 4 in the surveys, and tabulations of these responses do not cleanly map onto published results. For example, the final report for the Kenya DHS 2003 survey says that 22.5% of women read the newspaper at least once a week, 28.8% watch television at least once a week, and 75.1% listen to the radio once a week. However, this does not match the results I have calculated below.

I am not having problems reproducing other results, and I am also not having a problem reproducing these statistics in KDHS 2014 (although there is still a discrepancy in the response options in the dataset vs. the response options in the questionnaire.

Any advice or insights into this issue?
index.php?t=getfile&id=748&private=0

Attachment: Screen Shot 2017-06-28 at 9.48.19 AM.png
(Size: 115.38KB, Downloaded 686 times)
Attachment: Screen Shot 2017-06-28 at 9.48.13 AM.png
(Size: 61.13KB, Downloaded 723 times)

Report message to a moderator

Re: Labeling mistakes in exposure to mass media variables in Zambia 2007 stata dataset [message #12766 is a reply to message #12649]

Tue, 11 July 2017 10:35

Bridgette-DHS
Messages: 3208
Registered: February 2013

Senior Member

Following is a response from Senior DHS Stata Specialist, Tom Pullum:

Here is the weighted distribution of the responses to radio listening in the Kenya 2003 survey:

index.php?t=getfile&id=758&private=0

"Listening to the radio once a week" means listening at least that often. It is the sum of codes 2 and 3, which is 75.18%. If the report says 75.1%, I would consider that to be a match. (The difference of a tenth of a percent is annoying but is probably due to the treatment of code 9.)

I don't know why codes 0 to 3 were used rather than codes 1 to 4, for this survey. We occasionally encounter survey-specific or country-specific inconsistencies of this sort. Your problem in getting a match was that your consolidation of codes was not appropriate for the 0 to 3 coding.

I assume that this will explain the problem getting a match for newspaper and TV too, and for Kenya 2008 as well as Kenya 2003.

Attachment: tab-v158.jpg
(Size: 35.54KB, Downloaded 689 times)

Report message to a moderator

Previous Topic:	Philippines 2013 - Wealth Quintiles
Next Topic:	Birth certificate variable in Togo 2013-14 DHS

Goto Forum:

-=] Back to Top [=-

[ Syndicate this forum (XML) ] [

] [

]

Current Time: Wed Dec 11 16:15:48 Coordinated Universal Time 2024