Annex C. Sensitivity analysis
Highly educated adults are over-represented in the OECD 2020 Survey of Career Guidance for Adults (SCGA). The average share of highly-educated adults (defined as ISCED 4-6) is 68% in SCGA, which is nearly twice higher than the average share of highly educated in the population (35% on average across the six countries in the survey). This is because low-educated people tend to participate less in online surveys than those with a higher level of education (Van der Heyden et al., 2017[1]).
Table A C.1 shows results from a simple sensitivity analysis where the use of career guidance within each education group is held fixed, while the share of adults in each education group is adjusted to match the population. A weighted average is computed, multiplying the share of adults in each education group by their use of career guidance, then summing up across all education groups. The results of the sensitivity analysis show that, all other things being equal, if the education composition in the sample matched the actual education composition in the population, the share of adults who used career guidance in the last five years would be 40%. This is indeed slightly lower than the measured usage of career guidance (43%) in the sample. But it suggests that over-representation of highly educated adults does not have a large impact on the accuracy of the overall findings.
Reference
[1] Van der Heyden, J. et al. (2017), “Additional weighting for education affects estimates from a National Health Interview Survey”, European Journal of Public Health, Vol. 27/5, pp. 892-897, https://doi.org/10.1093/eurpub/ckx005.