EQLS 2007 - Methodology: weighting
In order to ensure that the results of the second European Quality of Life Survey are representative, three types of weighting have been applied to the data:
- Design weights. The data is weighted in order to account for the unequal probability of selection of respondents. This compensates for the fact that, in a one-person household, the probability of being selected is 100%, whereas it falls to 33% in a three-adult person household.
- Non-response weight. This weight corrects for possible biased estimations arising from different types of eligible respondents which have different response rates. In this stage the variables taken into consideration in the weighting procedure are age, gender, urbanisation level, region (NUTS 2 or corresponding national regional classification), household size (adults in size of household).
- International weights. this final step in the weighting is applied in order to be able to generate estimations for groupings of countries (EU27, EU15, NMS12, CC3). The weights of all respondents in each country are multiplied by the proportion that this country represents in the total adult population in the respective cross-national area. This removes the bias that occurs due to a proportionally greater number of interviews in smaller countries. For international weighting, the official population figures provided by Eurostat or national statistic offices are used.