This is the dataset that underlies study 2 of the paper: 'Why general moral values cannot predict moral behavior in real life' (still in process)

In this study we have measured people's general moral values and people's compliance with the Corona measures during the first 'intelligent lockdown' March-May 2020 through an internet questionnaire.

To measure people's general moral values we used items from the Morality as Cooperation Questionnaire. Compliance with corona behavior was operationalized by using items from a national survey measuring compliance rates with the proclaimed measures conducted by the National Institute of Public Health (RIVM) (www.rivm.nl).

We used 12 items from the MAC-Q judgment section to measure 4 general values (3 items per value): group loyalty, reciprocity, deference, and fairness. The 12 items are called intrinsiek_1-12. Factor analysis revealed that our data by and large reproduces the original structure given by MAC. One item (people have an obligation to help members of their community) did load higher on a different factor (reciprocity) than it originally belongs to (group loyalty). Still, for considerations with regards to content, we decided to keep it as an item of the group loyalty subscale and preserve the original structure. Sumscores were computed of the four subscales and entered as independent variables in the linear regression analyses. The four subscales are named Belang van groepswaarden (group loyalty), Belang van wederkerigheid (reciprocity), Belang van eerbeid (deference), and Belang van eerlijkheid (fairness).

We used 10 items to measure peoples adherence to the Corona measures during the Intelligent lockdown between March-May in The Netherlands. The items are named regels_1-10. In order to bring down the number of models to be estimated, a Principal Component Analysis (PCA) was conducted to summarize the data. This revealed that 9 out of 10 items converged on three distinct components that were logically interpretable: compliance with measures involving personal hygiene, not visiting the most vulnerable groups in society, and general social distancing (Table 9). The one item that did not sufficiently load on any of the three components, asking about coughing and sneezing in the elbow, was left out of the analyses. For each component the sumscore was computed and included in the linear regression analyses as the dependent variable. The constructed variables are named SOM_pershygiene (personal hygiene), SOM_afstandZwakke (not visit the vulnerable), SOM_socialdistancing (General social distancing).

We entered three control variables as independent variables in the regression analysis: gender (gender consists of three categories: male, female, and other. We created two dummy variables with male as ref., to include it in the analysis: Geslacht_dummy_vrouw and Geslacht_dummy_Anders), age (leeftijd) and education (Dummy3_opleiding_MBO4laag). 

We performed a hierarchical linear regression analyses by first only entering the control variables in the first model and adding the moral value variables in the second.
