2. EXPLORATION OF THE NONRESPONDENTS IN THE EGLHS 
19 
of N-1246 persons out of 2131 nonrespondents from the initial sample (this means that the 
NRS gross sample contains (1246/2131)'100-58,5%33 of the wave 1 nonrespondents). From 
this gross sample, 282 persons had to be subtracted due to out-of-date addresses, problems of 
finding persons, etc. This is reported as a whole under the title of "neutral loss" in the field 
report. The final actualised gross sample of nonrespondents consisted of N=964 persons. 
In addition to an announcement letter, an incentive of DM 50,- was offered to all 
selected nonresponse target persons. Although the majority of them once again did not 
participate (69,5% refused again), 201 nonrespondents finally gave complete interviews in 
which CATI/CAPI programmes" similar to the ones in wave 2 were used. We can define the 
response rate of the NRS study in three ways: taking all the initial nonrespondents of wave 1 
as 100%, à response rate of only 9,4% is calculated (201 out of 2131). Taking the pool of the 
NRS gross sample, we end up with a response rate of 16,1% (201 out of 1246). Defining the 
actualised gross sample as 100%, we calculate a final response rate of 20,9% (201 out of 
964). The last version is given in the infas report. 
The Nature of the Nonresponse Sample 
Summing up the preconditions of the nonresponse study, we can recognise: 
independently from the kind of calculation of the response rate, the NRS sample obviously 
cannot be assumed to be a representative selection of all nonrespondents. Thus generalisations 
of the findings will be limited. The sample is in principle biased as most of the refusals were 
again missed out and so it consequently appears to be a filtered extract of special "difficult" 
target persons. 
In the next step, we will take a look at the sample development for all parts of the 
study in more detail. It was shown above that numbers for percentages depend strongly upon 
the basis chosen as 100%. For this reason TABLE 1 follows the absolute number of target 
persons from the initially drawn sample until the final survey data set and lists the reasons 
why people dropped out. 
3 rounded to one decimal 
» With slight modifications concerning additional questions about jobs, marriages, partners and education before 
1989.