Measuring sexual identity using direct and indirect questioning

The identification of the best strategies to measure sexual orientation is needed to inform policy makers of experience of stigma and harassment suffered by the lesbian, gay and bisexual population. However, the measurement of sexual orientation faces methodological difficulties, since sexuality is among the most sensitive topics in surveys. This experiment proposes to evaluate the self-administered UKHLS question on sexual identity against the Integrated Household Surveys (IHS) interviewer administered version. The experiment uses a two-list item count sensitive questioning (ICT) technique to obtain something akin to validation data on sexual orientation in order to evaluate the IHS and UKHLS direct questioning approaches. By repeating the two-list ICT longitudinally and rotating allocation of the sensitive item to lists, respondent’s sexual identity can be directly ascertained, permitting a validated micro-level analysis.

Together with the item count lists outlined under the experiment concerning the measurement of attitudes to immigrants Social desirability bias in attitudes towards immigration, the ordering of item counts was randomised across respondents.  The statements which respondents counted also had a randomised order across respondents. Randomisation occurred during the interview with associated variables indicating the randomised orderings, their variable and value labels, specified in the questionnaire.

The experiment was a 2×2 design where half of respondents received either the ONS direct sexual identity question or the UKHLS direct sexual identity question.  The crossed-treatment assigned respondents to one of a pair of item-counts concerning sexual attraction, sexual behaviour and sexual identity respectively.

The 2×2 design:

		ff_ictsexw8
		Group A	Group B
ff_sexidw8	1 = UKHLS protocol	1 = Lists: A, B+S1, C, D+S2, E, F+S3 Plus the UKHLS direct question	2 = Lists: A+S1, B, C+S2, D, E+S3, F Plus the UKHLS direct question
	2 = IHS protocol	1 = Lists: A, B+S1, C, D+S2, E, F+S3 Plus IHS direct question	2 = Lists: A+S1, B, C+S2, D, E+S3, F Plus HIS direct question

The longitudinal element:

		IP8	IP9
ff_ictsexw8	1 = Group A	List A	List A + S1
		List B + S1	List B
		List C	List C + S2
		List D + S2	List D
		List E	List E + S3
		List F + S3	List F
		Plus Direct Question A or B
	2 = Group B	List A + S1	List A
		List B	List B + S1
		List C + S2	List C
		List D	List D + S2
		List E + S3	List E
		List F	List F + S3
		Plus Direct Question A or B

The variables used for this experiment are in record h_indresp and i_indresp: ictsexa, ictsexbs, ictsexao, ictsexbso, ictsexabso, ictsexc, ictsexds, ictsexco, ictsexdso, ictsexcdso, ictsexe, ictsexfs, ictsexeo, ictsexfso, ictsexefso, ictsexas, ictsexb, ictsexaso, ictsexbo, ictsexasbo, ictsexcs, ictsexd, ictsexcso, ictsexdo, ictsexcsdo, ictsexes, ictsexf, ictsexeso, ictsexfo, ictsexesfo,  sexuor, sidqn