I have a problem when trying to test difference between two means in a survey with a t test. Basically I have a repeated health survey from different years and I want to test if the difference in means between them is meaningful. Using package survey I was able to get the means, along with standard errors with this code
svyby(~CADL2,by=~ageco+year,design = wcd,FUN= svymean,na.rm=TRUE)
ageco year CADL2None CADL2Limited se.CADL2None se.CADL2Limited
50-54.2007 50-54 2007 0.8717507 0.12824934 0.02652932 0.02652932
55-59.2007 55-59 2007 0.8919843 0.10801569 0.01662038 0.01662038
60-64.2007 60-64 2007 0.9056865 0.09431355 0.01556955 0.01556955
65-69.2007 65-69 2007 0.8525438 0.14745624 0.02376984 0.02376984
70-74.2007 70-74 2007 0.7534787 0.24652131 0.03419399 0.03419399
75-79.2007 75-79 2007 0.7466576 0.25334237 0.04010796 0.04010796
80-85.2007 80-85 2007 0.5690972 0.43090276 0.06083682 0.06083682
85+.2007 85+ 2007 0.3853919 0.61460811 0.08913058 0.08913058
50-54.2017 50-54 2017 0.7150962 0.28490379 0.13929132 0.13929132
55-59.2017 55-59 2017 0.8720697 0.12793025 0.04088908 0.04088908
60-64.2017 60-64 2017 0.8503688 0.14963123 0.01783197 0.01783197
65-69.2017 65-69 2017 0.7931459 0.20685411 0.01829031 0.01829031
70-74.2017 70-74 2017 0.7764609 0.22353912 0.01895070 0.01895070
75-79.2017 75-79 2017 0.6666032 0.33339681 0.02428625 0.02428625
80-85.2017 80-85 2017 0.5462507 0.45374929 0.03324155 0.03324155
85+.2017 85+ 2017 0.3223467 0.67765331 0.03956227 0.03956227
Now I would want to test if for example the mean for age cohort 50-54 in 2007 (0.8717) is significantly different from that in 2017 (0.7151). I tried svyttest but as Im quite new to this package I couldn't get it to work.
svyby(~CADL2,by=~ageco+year,design = wcd,FUN= svyttest,na.rm=TRUE)
Error in formula[[3]] : subscript out of bounds
I tried swapping variables in the formula but everytime I got this error. Do you know how I could get it to work or achieve the test between two means in any onther way?
do you want svyby( CADL2 ~ year , ~ ageco , wcd , svyttest , keep.var = F ) ?
example with data from ?svyttest
library(survey)
data(api)
dclus2<-svydesign(id=~dnum+snum, fpc=~fpc1+fpc2, data=apiclus2)
svyttest(enroll~comp.imp, dclus2)
# single subset
svyttest(enroll~comp.imp, subset(dclus2,stype=='E'))
# with svyby
svyby(enroll~comp.imp,~stype,dclus2,svyttest,keep.var=F)