Who Refuses to Answer the Question about the Income and How Can We Reduce the Item Non-Response Bias By Using the Propensity Score Adjustment?

The mail goal of this presentation is to find out whether the refuses to question of income are random or not as well as how can we eliminate the effect of item non-response in point estimation.

Firstly, using the Hungarian and Polish data set of ESS 2008, it will be demonstrated that the likelihood to refuse is not random, but rather proportional to the declared level of income. In this part of presentation I will introduce the basic principles of propensity score adjustment (PSA) as a weighting scheme (see Matsuo et al. 2010, Lee 2006). This procedure is normally based on the logistic regression, but I will demonstrate the usefulness of a credit-scoring model for such purposes. In fact it is also based on logistic regression, but it helps to choose the relevant set of predictors as well as to illustrate and understand the nature of income refuses.

Secondly, based on the data from the “Polish General Social Survey”, an assessment will be provided of whether PSW or PSA leads to lower total survey error (TSE). By removing the known values of income I will consider three patterns of missingness: (a) the random one, (b) the systematic one without 10% of the lowest income values and (c) the systematic one without 10% of the highest values.

Findings are four-fold: (1) PSA is much more effective when missingness mechanism is systematic, however PSW is slightly more effective when non-response is random; (2) PSA increases variance a little bit more than PSW, but (3) PSA decreases bias much more efficiently than PSW. Taking (2) and (3) together, it turns out that (4) PSA estimator seems to be better on the ground that it implies much smaller TSE.