Page 312 - Contributed Paper Session (CPS) - Volume 2

CPS1873 Ferenc M. et al.
the target population annually. As EU-SILC statistics (European Union Statistics
on Income and Living Conditions) are based on the data from the HBLS, a
significant share of the research questions (variables) are internationally
harmonised. Even in these cases, it is of great importance to find the best
Hungarian formulation possible, that is, whose interpretation by the
respondents and elicitation of responses is the closest to the intention of the
researchers.

2. Methodology
The cognitive questionnaire testing methodology used in the project will
be described in detail below, for its Hungarian adaptation is an important
achievement of the project. Due to the qualitative nature and special purpose
of cognitive testing, the selection of participants typically follows such
selection strategies whose primary aims include not only reaching a certain
degree of heterogeneity but also ensuring that the sample covers an
adequately wide range of the target population (Willis, 2005).
The recommendations on sample size typically range between 5-50 people
in the literature. (Snijkers, 2002, Willis, 2005, Blair et al. 2006, Beatty & Willis,
2007, Blair & Conrad, 2011, etc.). The major factors to be considered in the
composition of the sample should be the characteristics of the respondents
the survey focuses on and the complexity of the questions (Miller et al. 2014).
In case of the cognitive testing of the HBLS questionnaire we aimed at a small
sample qualitative analysis of the cognitive processes in the question-
response process.
Sampling for cognitive questionnaire test studies is usually done in a non-
representative, purposive way. In the case of the present study, interlocked
quota sample design was used (see Collins, 2015). The selection criteria
(quotas) were fitted to the characteristics of the HBLS. The final sampling
criteria were thus the following: level of education, residence (capital versus
small rural village), and main activity status. The size of the sample designed
with these carefully chosen aspects in mind was 12 people. Regarding sample
size, the budget allocated for the study was determinative, the sample design
had to be tailored to it (number of selection criteria and test subjects per
quota).
Weighing the alternatives, two of the possible recruitment channels were
deemed befitting the target population of the cognitive testing of the HBLS
questionnaire: the combination of (online) advertisement and recruitment
through personal network could ensure optimal results. Thus our criteria
enabled including hard-to-reach populations (such as residents of small rural
villages as well as active urban residents with low level education) in the study
as well as filling each quota. A reserve sample of one reserve person assigned

301 | I S I W S C 2 0 1 9