Need help modelling data

[Zak Issa]

I need helping modelling this data over stata. Im finding it difficult creating dummy for those that Matched, Ontario, and CdnBorn. Help would be appreciated.

[Joseph Coveney]

Im finding it difficult creating dummy for those that Matched, Ontario, and CdnBorn.

Maybe something like this?

clear *

input byte(matched immigrant ontario) int count
0 0 0 2660 // `=7660 - 5000'
0 0 1 2445 // `=4425 - 1980'
0 1 0 3295 // `=4455 - 1160'
0 1 1 3710 // `=5040 - 1330'
1 0 0 5000
1 0 1 1980
1 1 0 1160
1 1 1 1330
end

label define Immigrant 0 CdnBorn 1 Immigrant
label values immigrant Immigrant

label define Ontario 0 "Rest of Canada" 1 Ontario
label values ontario Ontario

label define Matched 0 Unmatched 1 Matched
label values matched Matched

tabulate matched immigrant if !ontario [fweight=count]
tabulate matched immigrant if ontario [fweight=count]

exit

[Zak Issa]

Maybe something like this?

clear *

input byte(matched immigrant ontario) int count
0 0 0 2660 // `=7660 - 5000'
0 0 1 2445 // `=4425 - 1980'
0 1 0 3295 // `=4455 - 1160'
0 1 1 3710 // `=5040 - 1330'
1 0 0 5000
1 0 1 1980
1 1 0 1160
1 1 1 1330
end

label define Immigrant 0 CdnBorn 1 Immigrant
label values immigrant Immigrant

label define Ontario 0 "Rest of Canada" 1 Ontario
label values ontario Ontario

label define Matched 0 Unmatched 1 Matched
label values matched Matched

tabulate matched immigrant if !ontario [fweight=count]
tabulate matched immigrant if ontario [fweight=count]

exit

Wow thank you! Also, is the Canadian Born-Immigrant match gap (difference in the probability of matching occupations) in Ontario statistically different from that in the rest of Canada? Still haven't gotten around to learning how to use the condition probability function on Stata.

[Joseph Coveney]

Also, is the Canadian Born-Immigrant match gap (difference in the probability of matching occupations) in Ontario statistically different from that in the rest of Canada? Still haven't gotten around to learning how to use the condition probability function on Stata.

I'm not sure what the condition probability function is, but how about

glm matched i.immigrant##i.ontario [fweight=count], family(binomial) link(identity)

With that sample size, it would be a surprise that even the most trivial discrepancy isn't statistically different from zero.