The investigation of the over time (and cross-national) comparability of sample survey data is a prerequisite for their analysis. In this context, in this dissertation, the over time comparability of methodological issues of the Greek Household Budget Surveys (HBS) conducted in 1957/1958 to 2015 is examined. The main issues of the sampling design were investigated based on current literature and practice in order to demonstrate that the requirements for the comparabilityof the sample surveys refer to all aspects of the survey design. Also, the substantive units’ definitions, the measurement of the demographic and social characteristics and the coding of goods and services measuring individual consumption were investigated in order to demonstrate the requirement for their similarity and standardization. The results of the sampling design analysis for the successive Greek HBSs and their implementation shows that, due to the extensive use of substitution in dealing with non-response, all surveys depart from the probability assumption and therefore their comparability is questionable and therefore this practice should be abandoned. Also, the HBS questionnaire should be translated into the most dominant other language, i.e. in Albanian,according to the methods proposed in the literature,so as not to exclude a sub-population from the survey. The results of the investigation for the similarity and standardization of the substantive units definitions show that, the introduction of a reference person requires further systematic scrutiny of their demographic and social characteristics in particular, for the married reference persons in order to determine the impact or not of the dominance rules while analyzing data from these surveys.The results of the investigation on the over time comparability of the measurements of the demographic and social characteristics and the coding of goods and services measuring individual consumption demonstrate that their use requires even less variationsand more detail for comparability in order to achieve similarity and standardization of measurements, so as to avoid deviations in comparisons.Furthermore, the extensive use of the demographic and social characteristics in social research demonstrates the need for more detailed examination of the survey instructions issued for the interviewers and more attention should be paid to the coding system and the use of individual consumption codes for the comparability of the measurement of goods and services.
In order to demonstrate the complexity of decisions required for ensuring the comparability of measurements over time, a detailed analysis of bread, meat, pasta and rice consumption was carried out. The findings show that, the use of different classifications on goods and services as recorded in the HBS requires a detailed code check in order to allow for over time comparability. Although the analysis presented is not exhaustive, this methodological study contributes to the growing research on the over time comparability of sample survey data that is used extensively in social research. The methodology presented may be easily applied to ensure the cross-national comparability of sample survey data.
Stefanos Giakoumatos, Professor TEI Peloponnese