Analyzing the lives of LGBTI people - survey approaches to LGBTI persons, couples and families 1 |
|
| Chair | Dr Stephanie Steinmetz (Sociology Department, University of Amsterdam ) |
| Coordinator 1 | Ms Mirjam Fischer (Sociology Department, University of Amsterdam) |
| Coordinator 2 | Ms Nancy Bates (Research and Methodology Directorate, U.S. Census Bureau) |
The Socio-Economic Panel Study (SOEP) is an annual household panel survey in Germany, covering currently about 28.000 adult respondents in 16.000 households. While information about the household structure and in many cases also self-reports on partnership biographies allows inferences about the sexual orientation of respondents in cohabitating partnerships since wave 1 in 1984, the survey also asks respondents in 2016 to self-report their sexual orientation (straight, lesbian, gay, bi, and other). This presentation focuses on these two approaches to measure sexual orientation in ongoing household panel surveys and discusses practical implications and limitations.
We first compare (under)coverage of household-inferred and self-reported sexual orientation in different subgroups of the population (cohorts, family types, education, etc.). We then investigate interviewer effects as well as other survey related factors of measuring sexual orientation (modes of data collection, presence of others, experience with the survey, etc.). Third, based on a composite measure of sexual orientation and drawing on the interdisciplinary questionnaire of the SOEP, we report exemplary differences in income, personality, health, and political attitudes between LGB respondents and heterosexual respondents. Finally, we discuss non-response bias in these population differences as well as methods to compensate for selective non-response in measures of sexual orientation (e.g., Heckman selection models).
Recently in the U.S, a great deal of attention has been paid to the measurement and collection of sexual orientation and gender identity (SOGI). In 2011, the National Institutes of Medicine recommended that SOGI be added to demographic questions in federally-funded surveys. In 2013, the National Health Interview Survey became the first nationally representative federal survey to collect sexual orientation. In 2014, the Office of Management and Budget (OMB) formulated an interagency workgroup on improving measurement of SOGI in federal surveys.
One topic garnering attention is whether SOGI can be successfully collected via proxy. To acquire empirical evidence on the subject, the Census Bureau sponsored a survey as part of the 2015-16 Joint Program in Survey Methodology (JPSM) Practicum. The primary research goal was to collect SOGI for all members of a household aged 16 and older using a single household informant. This methodology mimics how demographics are collected for surveys such as the Current Population Survey (CPS) and American Community Survey (ACS) where one household member typically answers questions on behalf of all. In this paper, we summarize the objectives, methodologies, and results from the Practicum.
The Practicum survey consisted of a voluntary online survey administered using Amazon’s Mechanical Turk (MTurk). The survey included a household roster, SOGI and other demographics including age, sex, race, and Hispanic origin and questions about employment status, earnings, and income. The SOGI questions were included in the demographic section of the survey, which occurs at the beginning of the survey. The survey was completed by a single household member providing an opportunity to evaluate the performance of SOGI questions in a proxy response environment.
The survey consisted of two stages, a “cognitive” pre-test and a main survey. The cognitive survey collected self-reported and proxy data on SOGI and included follow-up questions about sensitivity, confidence and comfort levels when reporting for others. The main survey collected data on SOGI, employment status and income, but also contained a split panel gender identity wording experiment. The first experimental panel collected GI using the “two-part” method (sex assigned at birth and current gender identity), asking sex assigned at birth first. The second experimental panel consisted of three GI questions: first asking current gender identity, next if someone specifically identified as transgender, and asking last about sex assigned at birth. The cognitive test included 500 respondents and the main survey included 5,000 respondents.
The following research questions are addressed:
• Item Nonresponse: What is the rate of nonresponse to the SOGI questions among those who answer on their own behalf? Compared to standard demographics? Compared to historically sensitive questions (earnings and income)?
• Proxy versus Self-Response: What is the rate of nonresponse to SOGI questions when asked by proxy?
• Sensitivity and Accuracy: How confident are proxy reporters in their SOGI answers? How comfortable are they when reporting for others?
• Question Wording: Which GI question wording yielded the lowest measurement error?
There is an increasing demand for data on LGBTI persons, couples and families, which is not biased by non-probability recruitment strategies. In order to cater to (parts of) this demand, we started the project Unions in Context (UNICON): a web-mode population survey among individuals in same-sex and mixed-sex couples in the Netherlands. Households were approached via postal mail and both partners were invited to take part in the web survey. The sample of households was drawn randomly from the population registers in 22 Dutch municipalities, which vary in terms of their size, geographical location and degree of urbanization. Households with same-sex couples (with and without children) were oversampled to ensure sufficient observations for comparison with mixed-sex couples. Our aim was to include not only individuals who are married or in a registered partnership (whose information is readily available in the registers) but also individuals who live together with their partners unmarried. UNICON is a multi-topic survey we are investigating a wide range of issues, including everyday experiences of same-sex couples, child well-being, division of household labor, attitudes and values, civic participation, the parental home as socialization context, and more. In this presentation, the sampling design of the survey will be presented alongside with the first preliminary results. Specifically, challenges with regards to reaching the target population will be discussed.
Between February and November 2015, the French Institute for Demographic Studies (INED) conducted a quantitative survey on violence and gender relations (Virage) on a sample of 27,000 men and women aged 20-69. To complement the representative sample of the general population, a targeted search was organized via the internet to increase the number of lesbian, gay, bisexual and transgender respondents (Virage-LGBT). Between December 2015 and March 2016, LGBT individuals consulting certain websites (dating sites, LGBT information sites) were invited to take part in the survey on a voluntary basis. From the very outset, the survey was criticized by people who did not identify with any of the proposed categories, notably those of "man" and "woman". This was the case for certain intersexuals, certain transsexuals and some people who defined themselves as non-binary in terms of gender or sexuality. In response to these criticisms, several changes were made to the questionnaire, and three questions on gender and sexual identity were added.
This paper looks at the issues underlying these criticisms. It is not simply a question of imposed statistical categories, but also of the dissonance between a survey on violence based on a gender relations approach, and respondents who contest the pertinence of the gender categories used. Although, in contrast to other surveys on violence, we sought to identify precisely the sex of the victims and perpetrators, we were criticized for reproducing the violence by assigning a sex or a sexuality to the respondents. In a context where gender categories are contested, can we and should we statistically record these categories? What are the socio-demographic characteristics of the people who express these criticisms?
To answer these questions, we will make use of the Virage LGBT database, and the open questions in the questionnaire, in which people were able to express their opinions. These verbatim accounts were coded according to the nature and register of the remarks. Our aim is to show that critical remarks take two forms: a demand for recognition of singular identities and a refusal to be placed in a specific category, a desire to remain indeterminate. In sum, our objective is to explore the consequences of the contestation of gender categories for sociological analysis of gender, quantitative analysis in particular.