This left the following employee sample sizes: HES 1997, 2615; HES 1998, 2675; NZIS 1997, 11,604; NZIS 1998, 10,993.
Three methods were used to impute work experience variables for the employees in the HES and NZIS samples:
j] is a vector of indicator variables for each quarter in the case of the HES regressions, and a single dummy variable for the year 1997 in the case of the NZIS regressions.
Results for the HES and NZIS samples are given in Tables 7 and 8 respectively.
Industries and occupations were defined at 2-digit level in the HES regressions and at 3-digit level in the NZIS equations.
2 The NZIS regressions included 72 industry and 57 occupational group dummies, defined at 3-digit level.
Table 2: Male-female employee educational gaps, 1984 and 1998 Males HES HES NZIS 1984 1998 1998 Highest qualification % % % No qualifications 35.
For the NZIS sample, the only information available for predicting participation independently of wages was the educational level of the respondent's spouse.