A proxy response is a response made on behalf of the sampled graduate by someone other than the graduate. It is an indicator of accuracy as information given by a proxy may be less accurate than information given by the desired respondent. However, if the respondent is unavailable, someone in their household or family (who is therefore likely to know them well) may be able to offer some useful information about their activity. Since our users value high levels of completeness, we viewed the risks to accuracy and reliability as acceptable, if we could seek to minimise them.
Our survey therefore uses the following strategy to minimise proxy responses. During the first half of the field period for each cohort (approximately six or seven weeks) proxy responses are not sought by telephone interviewers. During the second half of the field period, interviewers are advised to collect responses from third parties, where possible, and where a suitable proxy respondent (defined as a partner, relative, carer or close friend) is available. Only the mandatory questions are asked of proxies, and subjective questions are excluded. We do not collect proxy responses from the graduates of English Further Education Colleges as a matter of policy. We also make sure that responses collected from third parties do not exceed 10% of a provider’s target population, limiting the impact on data quality.
In the third year of surveying, we received a total of 1,455 proxy responses. Proxy responses for by cohort for the second and third years of the survey are displayed in Table 6 (below):
|Cohort||Number of proxy responses,
|Number of proxy responses,
Given how few proxy responses we have obtained, and the controls in place to manage these, as well as a lack of feedback from users, we have not provided information on proxy responses in the published outputs. To further support users’ understanding of the likely accuracy or reliability of this data, we would ideally like to provide this information to all users in the microdata. Since we possess the survey paradata required, adding this variable to the derived fields remains a potential target for future improvement. However, feedback from users has not indicated this to be a high priority.