Proxy Interviews in the Norwegian Labour Force Survey

In 2018 there will be a test-run of a multi-mode data collection for the labour force survey with the use of CAWI. This pilot does not permit proxy interviews, and it is therefore important to get a better understanding on who is responding with proxy interviews and how this may affect the labour force survey statistics. The sampling in the future may change from family selection to person selection.

This document summarises trends in the Norwegian Labour Force Survey between 2006 and 2017, and it summarises the impact proxy interview gives compared to self-respondent interview for persons aged 15 to 74 years. A closer look at background such as gender, age, register and survey employment data, student status and marital status etc. can help us define which group is represented most in proxy interviews. We found that younger persons who mostly are students have higher chance of proxy interviews. They are usually single persons, who haven’t settled yet. The youngest age group from 15 to 19 years has more proxy interviews than self-respondent interviews. The distribution for females and males swapped for proxy interviews during this period. Until around 2012, men were more likely to have proxy interviews, but in the recent years, women are more likely to have proxy interviews.

Odds ratio for certain groups and sub-groups can compare the odds in self-respondent and proxy interviews. The results from odds ratio replicate the tables and plots from the trends between 2006 and 2017. We also compare the odds ratios for register employment and survey employment data for self-respondent interviews compared to proxy interviews, for students and non-students at the age of 15 to 29-year olds in 2000, 2006 and 2017. There is no significant difference for students and non-students by comparing the quarters for each year. However, the odds ratios to employments, (register and survey employment), for self-respondent compared to proxy interviews have increased by the years.

In a sample survey, each person in the net sample is been given a weight of how many persons they represent in the population, and the weights estimate the employment status for the total population. Originally, the weights include both self-respondents and proxies. However, going from family selection to person selection, the proxy interviews are assumed as part of the non-response. This means that the weights are only given to self-respondents. Overall there is an increase in the labour force, employed and unemployed, and a decrease outside the labour force, when we go from the original labour force survey to assume the proxies as non-response. However, the youngest age group 15 to 24 years has a 1 to 3.5 percentage points increase in the labour force, employed and unemployed and a decrease of 3.5 percentage points outside the labour force when we exclude the proxy interviews.

Author: Lisa Li