Survey administration and participation: a randomized trial in a panel population health survey.

BACKGROUND

Choice of survey administration features in surveys mapping population health may influence the participation and the generalizability of the results. This randomized study aimed to investigate whether three digital letters (denoted single-mode administration) lead to similar participation as two digital and three physical letters mailed with shorter duration between reminders (denoted sequential mixed-mode administration).

METHODS

In total, 9,489 individuals who participated in The Danish Capital Region Health Survey in 2017 were randomized to re-invitation in 2021 (≥ 20 years) by either single-mode with three digital letters (N = 4,745) or sequential mixed-mode survey administration with two digital and then three physical letters (N = 4,744). To investigate the influence of survey administration on representativeness, the two groups were compared regarding sociodemographic characteristics of participants (age, sex, country of origin, education, labor market attachment). Generalized linear models were used to estimate absolute and relative differences between the two administration groups in participation rates (overall and the increase after reminders). It was also investigated whether sociodemographic factors moderated the association between administration group and participation.

RESULTS

At the end of follow-up, the participation rate was significantly higher in the sequential mixed-mode group, which received five letters (78%), than in the single-mode group, which received three letters (61%), primarily due to a greater increase in participation after switching to physical administration and an increase after the two additional reminders in the mixed-mode group. Overall, individuals who decided to participate in the two groups were comparable in all sociodemographic factors, yet older participants appeared to benefit more from switching to physical administration and younger participants from additional reminders.

CONCLUSIONS

Depending on the target population, sequential mixed-mode survey administration and/or multiple reminders could be considered to increase participation; however, it does not necessarily improve the sociodemographic representativeness of participants.