The name is absent

respondent across the two sweeps this increases the likelihood of an income
response at sweep two. At sweep two, Wald tests on social class, NVQ levels,
ethnicity and country show that only ethnicity and country have an important effect on
model fit.

Table 6 presents multinomial logistic regressions for non response at sweep one to
consider the relative impacts of the explanatory variables on refusals and don’t
knows separately. The dependent variable is 2 if the main respondent refused to
respond, 1 if they didn’t know their income and 0 if they provided income data (this is
the reference category). For the main respondent (columns one and two) self
employed status remains important only for those who report that they do not know
their income. In addition to self employment, social class is an important predictor of
a don’t know response for the main respondent. For those main respondents who
refuse to report their income only the Northern Ireland category is a significant
predictor.

Columns three and four of Table 6 present estimates for the same analysis for the
partner respondent. A don’t know response to the income question is predicted by
self employment status, working for a small employer, not being educated to first
degree level or equivalent (NVQ Level 4) and living in Northern Ireland. A refusal is
predicted by self employment status, being an older respondent, not having an
occupation considered to be semi routine and routine, not being educated beyond
GCSE (NVQ Level 2), having a larger family, belonging to an ethnic minority group
and living in Northern Ireland.

Given the importance of self-employment status in predicting income non-response,
Table 7 considers whether a change in self-employment status leads to a change in
income response behaviour. The table presents separate panels for the main and
partner respondents. We find an increase in self-employment across the two
sweeps, especially for the partners. For the main respondent, and as predicted by
Table 4, moving out of self-employment leads to a greater likelihood of an income
response at sweep two whereas moving into self-employment reduces that
likelihood. The picture for the partners is somewhat different in that both movement
out of and into self-employment increase the chance of non-response at sweep two.
The numbers of respondents who change employment status are, however, rather
small.

10

<<   11   12   13   14   15   16   17   >>

More intriguing information