2. EXPLORATION OF THE NONRESPONDENTS IN THE EGLHS
23
refusal rate is 69,5%. This seems to be a poor result" if we restrict ourselves to looking only
at these percentages. We have to keep in mind, however, that this is a study of a special
group. We have got information about some "dark chapter persons" who usually do not
appear in surveys at all.
FIGURE 4 below provides a visual image of the sample development. Colour attributes
are used especially to enable a graphical contrast of the proportion of completed interviews
versus nonrespondents.
FIGURE 4:
Sample Development in the East German Life History Study
- Nonrespondents versus Completed Interviews -
N
4500
candidates for
4000
a neutral loss
nonresponse study
3500
7. panel mortality
3000
u no contact
illness, other reasons
2500
nonrespondent
a refused any information
interviews
2000
nrefusal
1500
sutilisable interviews
wave 1
1000
wave 2 interviews
500
o
1991/92 (wave 1) 1996/97 (wave 2) 1997 (nonresponse
study)
The left bar is a stack of the initial drawn sample: its total length represents N-4750 persons.
The neutral loss is given in black, non-contacts in grey, refusals of any kind of information
(only wave 1) in light red, illness together with other reasons (in wave 1) and non utilisable
interviews are given in grey stripes. The green bar indicates the final amount of utilisable
interviews. In wave 2 we also see the number of persons lost by panel mortality (green
stripes). In all three studies, the most obvious blocks are the refusals (in red) and the
completed interviews (in green). In wave 1 about one third of the initial sample are refusals.
This proportion is clearly much lower in wave 2, because only the participants were contacted
5 years later. We see that the wave 2 stack contains only a small red band.
* Groves/Couper (1998:50) mention that special nonresponse studies are often confronted with low response
rates.