Help-Seeking Behaviour of Domestic Violence_Myanmar DHS (2015-16) [message #29762] |
Wed, 31 July 2024 07:26 ![Go to next message Go to previous message](/theme/default/images/down.png) |
Soe Myat Htet
Messages: 4 Registered: July 2024
|
Member |
|
|
Dear Moderator,
May I discuss two questions regarding "Weighting"?
(1) I am doing DHS data analysis using "R". Before going data analysis, I did weight the selected variables after managing missing values and mutating some variables. Then, I compared the results of unweighted and weighted frequencies. The unusual differences in 7 categories were found in one variable "v101" which had 15 categories. For example, (n = 60) before weight and (n = 5.11) after weight. I checked again and again. Also, I checked in Stata. It showed similar. Is it possible? Should I weight or not? I used R codes as below:
IRdata$wt<- IRdata$v005/1000000
mysurvey <- svydesign(id = IRdata$v021, data = IRdata, strata = IRdata$v022,
weights = IRdata$wt, nest = T)
options (survey.lonely.psu = "adjust")
svytable(~ v101, mysurvey)
(2) In relation to weighting logistic regression analysis. Some researchers say that it should be weighted all analysis (descriptive, bivariate, and multivariate analysis). However, some researchers say that it should be considered for descriptive data analysis only. Please kindly give me your suggestion.
Thank you,
Soe
|
|
|