Presentation of results
Soon, the latest study results will be presented here.
Presentation of results of the pre-study | Last updated: August 24th, 2020 (N=260)
Here, some of the results of the preliminary study are presented. These include socio-demographic data on the participants, statistical analyses and interpretations. Among other things, the preliminary study was used to gain a first impression of the population’s preferences regarding measures and consequences of the corona pandemic and to adapt and further develop the questionnaire accordingly.
Socio-demographic data of respondents
N= 260 respondents took part in the survey of the preliminary study. Inclusion criteria for participation were a minimum age of 18 years and good to very good German language skills. This is not a representative sample. The majority of participants are students who were easier and faster to recruit for this study (convenience sample).
Table 1 summarizes the sociodemographic data of the participants included in the analysis. The participants have an average age of 32 years (Min: 18 years; Max: 74 years; Median: 26 years). The most frequent value in the sample is 22 years (modal value). The largest part of the sample consists of women (65%). Men are represented with a share of 33.8%. Regarding marital status, 36.5% of the participants are single and 35.4% are in a stable relationship but not married. A slightly smaller part of the participants (22.7%) is married or in a registered partnership. 3.1% are divorced or separated. Most participants in the survey are students with 47.7% and full-time employees with 20.4%. Of the remaining participants, 11.2% are part-time employees, 6.1% are self-employed and 2.4% are retired. With regard to the highest educational level, the majority of the participants (40%) graduated from a college or university, and 35.4% (35.4%) have a high school diploma or comparable qualification. 10.8% have a vocational school diploma or the advanced technical college entrance qualification, 6.9% have a diploma from a technical college or vocational school and 5% have a secondary school diploma. The monthly net income is 63.3% of the participants under 2000 € and 24.3% over 2000 €. The remaining participants (12.3%) did not provide any information regarding their monthly net income. More than half of the participants (56.9%) come from Mecklenburg-Vorpommern.
SD= standard deviation
Note: Due to missing data, the sum of the percentages in individual categories does not add up to 100.
Table 1: Socio-demographic data
Preference weights for characteristics of pandemic measures
The preference data were estimated by means of logistic regression (conditional logit model). Figure 1 shows coefficients for each level of a property (attribute) in the 95% confidence interval. The coefficients represent the relative importance for each level within an attribute. A positive coefficient means a positive preference for that level. The larger the coefficient, the greater the preference for that level within an attribute. Conversely, a negative coefficient indicates a negative preference for that level. Positive coefficients indicate an increased influence on the choice decision, negative coefficients a reduced influence on the choice decision. The vertical distances (or differences) between the levels within an attribute indicate the preference weight for a change from one level to another. If two considered levels are close together, the change from one level to the other is of little importance. If the levels are far apart, the change is of great importance. For example, for the attribute exit restrictions, the first two levels (1 month state border and 1 month states and state borders) are approximately at the same level. There is no measurable significant distance between the two levels. However, the distance to the third level 1 month curfew is clearly visible. This means that the participants accept a 1-month closing of the state border and a 1-month closing of the federal states and state borders to the same extent. A 1-month strict curfew, however, is rejected. The difference between the first and last level of an attribute shows the importance of this attribute compared to the other attributes. For example, the level difference for the attribute individual risk of infection (1.7 units) is greater than the level difference for the attribute decline of GDP (1.1 units). Thus, for the participants, the risk of infection was more important in their choice than the decline of GDP.
Figure 1: Coefficients of the most important attributes in the regression model in the 95% confidence interval (extract)
The level coefficients of the remaining attributes contact restrictions, closure of facilities and personal data are close to zero with some overlapping standard errors. Therefore only a trend can be reported here. As an example, the level coefficients of the attribute facility closure are shown in Figure 2. The coefficients for the closure of 1 – 3 months have a positive sign with slightly decreasing preference. This can be interpreted in such a way that a closure of up to 3 months during a pandemic seems acceptable. But the longer the closures last, the lower the preference. A threshold value for the acceptance of closures seems to be between 3 and 4 months. From 4 months on, the signs of the coefficients are negative, indicating a negative preference.
Figure 2: Level coefficients of the attribute Closure of facilities
Relative Attribute Importance
The graph shows the relative importance of the different attributes of the choice scenarios. The relative importance shows how much a attribute contributes to the overall assessment of a scenario. The greater the importance of an attribute, the greater the influence of the attribute on the participants’ choice. The importance was normalized on a scale of 10. The attribute excess mortality takes first place in the ranking of attribute importance. It is closely followed by the decline of individual income in second place. The risk of infection is ranked 3rd and exit restrictions 4th.
Figure 2: Relative importance of all attributes
Latent class analysis for modeling preference heterogeneity
The pandemic affects the entire population. It can be assumed that there are a large number of groups in the population with different preferences regarding coping with and accepting the consequences of a pandemic. The groups (or classes) may be a cluster of people with different experiences, preferences, characteristics, behavioral patterns or even political attitudes. With the help of latent class analysis, the respondents were divided into different classes according to their choices. Two classes with different preferences could be identified. The following figure 4 compares the relative importance of the attributes for the two classes. In the case of class 1, excess mortality had the greatest impact on the choice decisions. In contrast, the risk of infection was the most important decision criterion for the participants assigned to class 2. Common to both classes is that the decline of individual income had the second largest impact on the choice decisions.
Figure 4: Relative attribute importance in class comparison
Table 2 shows the socio-demographic data of participants in both classes. Class 1 with N= 149 participants is larger than class 2 with N= 111. Both classes show a similar age structure. Class 1 has an average age of 31 years (Min: 18; Max: 72; Median: 25; Modal: 21). Class 2 has an average age of 33 years (Min: 19, Max: 74; Median: 28; Modal: 22). In both classes there are more women (Class 1: 65.8%; Class 2: 64%) than men (Class 1: 33.6%; Class 2: 34.2%). Regarding marital status, there are about the same number of unmarried participants in both classes (Class 1: 36.2%; Class 2: 36.9%). In class 1, 36.9% and in class 2, 33.3% are in a fixed relationship. Both classes consist mainly of students (Class 1: 49.2%; Class 2: 45.7%). Full-time employed are 22% in Class 1 and 18.1% in Class 2. 13% more people in Class 2 are part-time employees than in Class 1 (9.9%) and slightly more self-employed (6.5%) than in Class 1 (5.8%). The proportion of retired participants is low in both classes. With regard to the highest educational level, there are more participants with a college or university degree in Class 1 (Class 1: 46.3%; Class 2: 31.5%) and with a high school diploma or comparable qualification (Class 1: 38.3%; Class 2: 31.5%). In class 2, more participants have acquired a vocational school diploma or an advanced technical college entrance qualification (class 1: 8.1%; class 2: 14.4%). 4.7% of the participants in class 1 and 9.9% in class 2 have a technical college or vocational college diploma. The share of secondary school qualifications is lower in class 1 (2%) than in class 2 (9%). The monthly net income is 48.3% of the participants in class 1 and 42.3% in class 2 under 1000€. In class 1 22.9% of the participants have a net income between 1000€ and 3000€ per month and in class 2 34.2% of the participants have a net income between 1000€ and 3000€ per month. The percentage of participants with a monthly income of more than 3000 € is higher in class 1 with 17.5% than in class 2 with 9.9%. In both classes more than half of the participants come from Mecklenburg-Vorpommern (Class 1: 51.7%, Class 2: 64%). In class 1 there are more participants from the South (class 1: 13.8%; class 2: 5.4%) and West (class 1: 10.7%; class 2: 7.2%) of Germany.
Table 2: Sociodemographic data of the classes
In the survey, participants were also asked to assess and evaluate measures and key figures of the coronavirus pandemic. Below is a comparative overview of the different views of the participants of both classes.
- Greater fear that family members or relatives will be infected.
- More or less satisfied with the crisis management of the federal government in the current coronavirus crisis.
- Larger part of the respondents estimate that it will take up to 12 months to get the coronavirus under control.
- The excess mortality is estimated to be higher (mostly 10%) by the subjects in class 1 than by subjects in class 2.
- The collapse of GDP for 2020 is mostly estimated at 8%.
- A majority (48.3%) considers the measures of exit restriction, closure of facilities (43%) and transfer of personal data on (32.9%) to be promising.
- The coronavirus is considered to be more dangerous for the participants themselves, family members, the economic situation of a country and for the health care system.
- Larger proportion of positively tested persons among acquaintances and friends.
- The class contains more people who do not belong to a risk group.
- A larger proportion of participants estimate that it will take 12 months or more before the coronavirus is brought under control.
- The excess mortality is mostly estimated to be below 5%.
- The transmission of personal data is considered rather unpromising (18.9%).
- The coronavirus is considered to be more dangerous for the own financial situation, the basic civil rights and data protection.
Overall, excess mortality is more important than the individual risk of infection. It can therefore be concluded that the health of the general public or the prevention of general mortality is given higher priority among study participants than their own individual risk of infection. This corresponds to the assumption that during the pandemic most people remained at home in self-imposed quarantine to protect not primarily themselves but others, e.g. people with pre-existing conditions and other risk groups. Individual responsibility for the health of others is more important than the risk of infection. In contrast, the focus is on reducing individual income rather than preventing a decline in the country’s overall economic performance (GDP). Data protection, contact restrictions and the closure of facilities take a back seat to expected economic and health losses. However, a strict curfew is more likely to be rejected than the closure of state borders or the closure of federal states and national borders.
In the extended analysis, however, differences between individual classes become clear. While one class is more concerned about excess mortality, the other class is more concerned about the individual risk of infection and its consequences. The difference in preferences is also clear when it comes to the critical issue of personal data.
The results of the study could help decision-makers in government and the health care system to make difficult decisions about the measures to be taken and the consequences of a pandemic in the interest of the population and also of population groups.
When analyzing and interpreting the results, it should be noted that the total number of study participants is not a sample representative of the population.