Tuesday 14th July      Wednesday 15th July      Thursday 16th July      Friday 17th July     

Tuesday 14th July, 16:00 - 17:30 Room: O-206

Large convenience samples, quantitative research and comparative analysis
Convenor	Mr Patrick Festy (National Institute for Demographic Studies, France )

Session Details

Convenience sampling is often used in qualitative research where the objective is not to measure behaviors or opinions but to detect relationships between different phenomena. It is generally considered as non-scientific in quantitative research, despite some usage when surveyed populations are difficult or costly to reach and/or identify. It is nevertheless regaining popularity due to the development of online surveys where respondents are incented to answer freely accessible questionnaires. The massive number of respondents that can be easily obtained orientates towards quantitative analysis of the material, the more so as most questions are inevitably closed-ended. Even international comparisons of large samples become possible. An EU agency has recently attracted more than 90,000 respondents self identified as lesbians, gays, bisexuals or trans, in 28 countries, the largest sample ever assembled on this population. In a continent of 400,000,000 adults, the number of LGBT people is several millions and the selection bias in the sample risks to be massive, but who is able to resist the temptation of quantitative analysis of such a large and original material?
What is possible and impossible on such data? What lessons can be drawn from qualitative analysis methods, which could be relevant for large samples? Which is more or less biased: simple cross tabulations or more sophisticated analysis? Are meaningless national results open to meaningful cross-national comparisons?

Paper Details

1. An online LGBT survey – surveying a difficult to reach population
Dr Sabine Springer (FRA European Union Agency for Fundamental Rights)
Dr Vida Beresneviciute (FRA European Union Agency for Fundamental Rights)

In 2012 the European Union Agency for Fundamental Rights carried out a European LGBT survey covering 28 countries, using an web-based opt-in approach. As traditional probability sampling cannot be used for hard-to-reach groups as the LGBT population, a convenience sample was considered a valid first approach for the exploration of this under-researched population. This paper will discuss the objectives, methodological precautions and measures taken to ensure high data quality and to reduce selection bias as far as possible. The use of the results for policy development as well as for scientific purposes will be discussed.

2. Perception of discrimination experiences on grounds of sexual orientation in the EU LGBT survey
Ms Marie Digoix (Ined)
Dr Elisabeth Morand (Ined)

The aim of this paper is to investigate how to use in social research the vast amount of original data collected in 2012 EU-FRA LGBT Survey. We mostly focus on the trends in different countries in a comparative perspective in the area of discrimination experiences perception in confronting the dataset with external variables reflecting diverse economic and societal indicators in respect to discrimination and family laws in selected countries

3. Do surveys correctly cover voters who are over 80 years old?
Professor Jean-yves Dormagen (Université de Montpellier)
Professor Laura Michel (Université de Montpellier)

Our paper questions the quality of electoral studies’ sampling design, focusing on age. We show that the oldest category of the actual electorate is often under-represented in such quota-based surveys. It’s crucial since those over 80 years old now account for 10% of French registered voters. The data reviewed comes from 5 electoral panels fielded between 2002 and 2012 and a barometer tracking opinions in 2012. The studies include between 4,000 and 40,000 respondents, enabling a refinement of age sub-categories. We also analyse how these representativity biases may affect the quality of the answers