Design of the survey

• Pages: 11-16

11

The LGBTIII survey was an onli ne opt-in survey. That is

to say, unlike surveys using proba bility rando m sam-

ples, respondents were self-selected, as they volun-

teered to participa te in the survey.

This design was ado pted because it would not have

been possible to achi eve arepresentative ra ndom sam-

ple of LGBTI people across t he EU in the absence of sam-

pling frames and reliable, detailed information about

the target popula tion in terms of its size, character istics

and composition in t he survey countries.

Using the traditio nal sampling te chniques would al so

have been challengi ng because of the low preval ence

of the target group in th e population. Fur thermore,

considering different data collection modes, research

has shown that conventional fa ce-to-face surveys suf-

fer from ahighe r level of social desirab ility bias than

web-based surveys, es pecially when ask ing sensitive

questions (4). Because of social stigma, some LGBT I

respondents may cho ose to conceal being LGBTI in con-

ventional face-to-face interviews. The social desirability

bias(5) is ex pected to be lower in web-based surveys,

which are perceived as anonymous and conf‌idential.

To avoid such bias, the LGBTIII survey was desig ned to

ensure conf‌identiality and anonymity.

Given the survey’s desig n, LGBTIII su rvey data are

based on anarrower pop ulation than the who le LGBTI

population of the 30 s urveyed countries. T he survey

population consists of people who:

(4) See Heerweg (2009).

(5) Social desirability bias is the phenomenon describing the

tendency of su rvey respondents to a nswer interview

questions i n away that is socially accep table and received

favourably o r at least not negatively by o thers. In this way,

respondent s tend not to respond hone stly but in away that

they can appea r more likeable to the inter viewer or avoid

negative reactions.

•can be reached through t he online means use d, i.e.

those who have access to the interne t; and

•became aware of the sur vey, e.g. by visiting aweb-

site/app promoting the sur vey; and

•chose to complete the sur vey.

This raises anumb er of issues for consideration:

•despite the fact tha t internet penetrati on has in-

creased across the EU (6), the remaini ng internet

non-users are not usual ly ara ndom subsampl e of

the general pop ulation, as elderl y, less educated

and low-income subgrou ps are more likely to be

internet non-users(7);

•due to self-selection, t here is no control over the

selection process, a s it is not possible to determ ine

whether targeted respondents complete the sur-

vey because they may have differe nt levels of mo-

tivation and interest in participating;

•the propensity to resp ond of the different LGBTI

groups (i.e. the probab ility of the LGBTI groups par-

ticipating in the su rvey) may vary across countri es

and sociodemographic strata because of cultural

and social norms.

To reduce apossible bias, su rvey disseminati on was

targeted as broadly as p ossible. A combination of

(6) A ccording to Eurostat 2018 d ata, the level of interne t access

to household s exceeded 80% in all count ries surveyed

except for Bulga ria (72%), Greece (76%), Lithuania (7 8%),

Portugal (7 9%), North Macedonia (79%) and Ser bia (73%),

where the percentage was slightly lower.

(7) Eurostat off‌i cial statistics, su rvey on ICT (information

communications and technology) usage in households and

by individuals.



Design of the survey