Comparing responses in repeated cross-sectional and panel studies: Results across eight weeks during the first COVID-19 lockdown in Denmark.

Abstract

It is often important to study people's thoughts, feelings, and behaviors over time. To this end, researchers have relied on repeated cross-sectional (RCS) studies, in which different people from the same population participate on different measurement occasions. Also, researchers have relied on panel studies, in which the same group of people participate on different measurement occasions. However, few studies have directly tested whether participants' responses in RCS studies were similar to those found in panel studies. To address this gap, we compared the responses to 33 items, 28 of which were further grouped into four aggregates (Affections, Worries, States, Health concerns), over 8 weeks during the first COVID-19 lockdown in Denmark in a RCS study (overall N = 5,993, per measurement occasion 616 ≤ n ≤ 964) with the responses in a largely equivalent panel study (993 ≤ n ≤ 2,546 across measurement occasions). The study participants were randomly drawn from the same quota-representative participant pool and responded to the same items on the same measurement occasions. Results indicated a few differences between the study samples on the first measurement occasion (i.e., selection effects between studies). Further, we found statistical support for different trajectories in 21 aggregates/items. However, visual inspection of the trajectories suggested subtle differences between the studies at large. The results thus raise awareness that the trajectories of people's thoughts, feelings, and behaviors can differ between survey methods, especially when only a few measurement occasions are considered. Nevertheless, such differences might not be substantial over time. Public Significance StatementAnalyses comparing responses to 37 variables over 8 weeks during the first COVID-19 lockdown in Denmark in a repeated cross-sectional (RCS) study with the responses to the same variables in a panel study showed different trajectories in 21 variables. However, in most cases, the differences were subtle over the 8-week study period. The results suggest being more cautious in generalizing trajectories of variables based on one study design and only a few measurement occasions.

Age Differences in COVID-19 Preventive Behavior

“…However, on the population level (over 20 measurement points), the changes in reported behavior still provided important insights. Furthermore, in repeated cross-sectional surveys with multiple measurement occasions, the trajectories of responses were comparable to those from panel surveys (given that participants were sampled from the same population; see Zettler et al, 2021). With the observation of the two focal preventive behaviors, the resulting changes in behaviors may thus serve as the best evidence available for informing future pandemic policymaking alongside further behavioral data, such as cell phone mobility data.…”

Vaccination policy reactance: Predictors, consequences, and countermeasures

“…Since we were interested in effects and changes on the population level, we decided to assess the variables of interest as part of an existing serial cross-sectional survey, the COVID-19 Snapshot Monitoring ( Betsch et al, 2020 ). Recent evidence suggests that trajectories in population-level perceptions and behaviors during the pandemic as assessed via serial cross-sectional and panel data are largely comparable ( Zettler et al, 2021 ). Our data was collected before approval and supply of the first vaccines on four timepoints between April 14 and October 27, 2020 in Germany.…”

Measuring parents’ readiness to vaccinate themselves and their children against COVID-19

“…Participants were recruited from the Danish COSMO branch [36,37]. Participants were invited via the official Danish email system (eboks); invitations were sampled randomly from a larger sample obtained from Statistics Denmark, country-representative for the Danish adult population with respect to age and sex.…”