4. Modelling Attrition at MCS2 using Family Income and
Income Response in MCS1 as Predictors
Here we consider the impact of family income and income non-response at sweep
one on attrition at sweep two. These results are summarised in Table 8. The
dependent variable here is 0 if a response is obtained and 1 if a non response is
obtained. Those families where the cohort member died or the family has emigrated
are not included in this model. Column one considers family income as measured at
sweep one. Family income at sweep one does have an important impact on drop out
at sweep two. Larger family income predicts less unit non response at sweep two.
Columns two and three consider unit non response at sweep two using item non
response with regard to income at sweep one. For both the main and the partner
respondent income non response at sweep one predicts unit non response at sweep
two.
Table 8: Predicting Attrition at Sweep Two from Family Income at Sweep One.
|
Unit Non-Response at Sweep Two | |||
|
(1) |
(2) |
(3) | |
|
Family Income |
0.66 | ||
|
(0.63 - 0.70)** | |||
|
Main Income Non- |
17 | ||
|
(1.3 - 2.2)** | |||
|
Partner Income Non- |
17 | ||
|
(1.4 - 2.1)** | |||
|
Observations |
16790 |
8205 |
11464 |
Note: 95% confidence intervals in parentheses * significant at 5%; ** significant at 1%, all explanatory variables
measured at sweep one
17
17