| tested for HIV within the past 12 months using Kenya DHS [message #12599] |
Tue, 20 June 2017 14:46  |
catherine3h
Messages: 10
Registered: February 2015
Location: United States
|
Member |
|
|
|
To Whom It May Concern: I am conducting an analysis looking at predictors of HIV testing within the past 12 months using the female Kenya DHS dataset from 2008-09. I am seeking help deciding on how to construct my final dependent outcome variable. The descriptive statistics / responses to the questions "when were you last tested for HIV" (v781) contain a lot of missing data. I suspect this is largely due to a skip pattern...For example if women have never heard of HIV or have never been tested for HIV, than they were not asked the timing question. However, the way that the variable is coded in the dataset, there are simply categories for tested in the previous 12 months, tested between 12-23 months ago, tested prior to 24 months ago or missing. I cannot decipher between actual missing data and cases where the skip pattern applied and some of the respondents did not need to be asked that question because it was not applicable. I am wondering if I should construct my final outcome variable as 1=tested within the past 12 months; 0=tested prior to 12 months ago;missing=missing. Or, if I should construct my final outcome variable as 1=tested within the past 12 months and 0=tested prior to 12 months ago & missing. The later approach retains all of the observations within the dataset. I know that this decision is related to the population that I want to examine. But I am wondering 1) How I can distinguish between missing and not applicable as I described above and 2) Is there a recommendation of how I should treat this variable in my final analysis? Thank you.
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
The DHS Program User Forum
Members
Search
Help
Register
Login
Home
