Understanding Society User Support

Dear Sir/Madam,

I hope you are well.

I am doing analysis on ethnic identity and am currently looking at the englang variable measured at waves a, e and f. In the documentation, I can see the question is asked first of the whole sample (wave a), then of the EMB sample (wave e), and finally of the IEMB sample (wave f).

However, some of those respondents who answered that question both at waves a and e seem to have changed their answers. In particular, 208 respondents from yes to no, and 193 in reverse (cross-tabulating a_englang and e_englang).

I had a look at the related #843 issue and would have a few follow-up questions.

(1) Shall I assume that English as a first language is a stable characteristic and these 401 responses are then not valid? If so - would you recommend to recode these cases as missing?

However, treating English as a first language as a stable characteristic doesn't explain why the question was asked a second time to those members of the EMB who participated both at wave a and e (i.e. as questions on other stable characteristics such as ethnic group would be asked only once).

(2) Alternatively, shall I assume such responses indicate that respondents might be bilingual (e.g. one parent does speak English as a first language and one doesn't) - and hence responses at both waves are valid and reflect what associations respondents made during the interview?
(3) Shall I assume that the responses at one wave overwrite the responses at the other wave and if so - which wave shall I use for these respondents?

I look forward to your reply.

Kind regards,

Ivelina