The Center for Social & Health Innovation (CSHI) is currently conducting an online panel survey to investigate potential social consequences of the COVID19 epidemic. In the first wave 1024 Austrians were interviewed. The second wave was conducted in June 2020. In the blog posts below we show and discuss first results of the study.

CLICK HERE FOR DETAILED SAMPLE INFORMATION & DATA STRUCTURE

#9 All Back to Normal? Behavioral Changes from April to June

10.07.2020

How did Austrian citizens change their behavior since the peak of the Coronavirus epidemic? There has been an immanent concern that citizens would become less and less willing to adhere to restrictive behavioral patterns1 or to support another lockdown in case of a second wave2. Our data indicate that this is not true, at least for less restrictive behaviors. Figure 1 shows that the mean values for “sneezing or coughing into the elbow” has only slightly decreased from April to June. Regular hand washing has also deceased only slightly. However, keeping the 1 to 2 meter distance to others in public has decreased more robustly. In fact, the mean value still remains in acceptable range between “often” and “always”, but looking at the actual observations (black dots) we see that the areas where people reported “sometimes” or “never” have become more populated in June compared to April. 

In Figure 2, we see a more drastic decrease in highly restrictive behaviors, such as avoiding any contact with older people, no or limited contact to friends/family, or that people would leave their homes for urgent matters only. This is not surprising, since more restrictive behaviors are no longer requested and the government has withdrawn most of the behavioral regulations. 

Furthermore, there is reason to believe that the willingness to adhere to restrictive behavioral regulations may have decreased. First of all, some of the policies (e.g., regulations against meeting family members from different households) which have been communicated by the government might have not been consistent with the Austrian constitution3. Furthermore, the skepticism towards government measures has increased from April to June. Figure 3 shows changes in individuals’ perception that the government measures are exaggerated, counterproductive and not well-thought-out. It should be noted that individuals in June are evaluating much looser measures compared to the strict measures people were confronted with in April. Overall, we still see small statistically significant changes, which point towards slightly more skepticism in June compared to April.

In Figure 4, we see the relationship between changes in skepticism towards government measures and change in preventive behaviors (left plot: less restrictive behavior, right plot: more restrictive behavior). To run this analysis, we combined the more and less restrictive behavior items to two mean scales4. We did the same for the three skepticism items. Figure 4 shows negative relationships, indicating that moving towards more skepticism from April to June was associated with less adherence to restrictive behaviors. The relationship is less robust for less restrictive behaviors, indicating that these easily performable actions may be less dependent on overall skepticism towards government measures.

In case of upcoming local clusters or even a second large-scale wave, the question remains whether this current trend marks a state of a general (and maybe continuing) fatigue or a mere intermission, after which citizens are still willing to quickly change and adapt their behavior to situational needs. In Serbia, the announcement of a second lockdown ended in violent protests2. In order to avoid fatigue or frustration effects, the government needs to maintain – or even regain the slight loss of – confidence in their measures. Such confidence may no longer be simply gained by the moment of crisis, but may need to be accumulated by preemptive open and transparent communication. 

 

Notes:

1 https://www.theatlantic.com/health/archive/2020/03/coronavirus-pandemic-herd-immunity-uk-boris-johnson/608065/

https://www.nytimes.com/2020/07/08/world/europe/serbia-protests-coronavirus.html

3 https://www.addendum.org/coronavirus/covid-19-gesetz-verfassungswidrig/

4 A confirmatory factor analysis (CFA) provided an acceptable model fit and supported this two factor solution.

Figure 1. Changes in less restrictive behavioral patterns from April to June. Points with error bars represent mean values and 95% confidence intervals. Small clustered dots show observed data. Behavioral patterns were measured on a 4-point scale (1 = never, 2 = sometimes, 3 = often, 4 = always)

Figure 2. Changes in more restrictive behavioral patterns from April to June. Points with error bars represent mean values and 95% confidence intervals. Small clustered dots show observed data. Behavioral patterns were measured on a 4-point scale (1 = never, 2 = sometimes, 3 = often, 4 = always)

Figure 3. Changes in skepticism towards government measures from April to June. Points with error bars represent mean values and 95% confidence intervals. Small clustered dots show observed data. Behavioral patterns were measured on a 5-point scale (1 = do not agree, 5 = agree)

Figure 4. The relationship between change in skepticism towards government measures (from April to June) and change in preventive behavior (from April to June). Lines shows best fitted lines (based on OLS regressions) and shaded region indicate the 95% confidence intervals. Yellow lines indicate zero behavioral change.

#8 Far-right-voting, conspiracy thinking and vaccination intentions

19.06.2020

In this blog entry, we look at differences and changes in conspiracy thinking across party voters. Furthermore, we look at the relation between conspiracy thinking and vaccination intentions. Conspiracy thinking is a variable composed of  five questions, such as “I think that there are secret organizations that greatly influence political decisions”1. Vaccination intentions were measured by asking respondents whether they intend to get vaccinated, once a vaccine is available. We only use data from participants who responded in the the first (April) and the second wave (June) of our study (N = 632). Since not all opted to report their voting behavior, our final sample size is 497.

The results indicate that voters of the far-right Austrian Freedom Party had significant higher scores in conspiracy thinking in wave 1. Furthermore, when comparing indiviudals with similar conspirary beliefs at wave 1, we find that those who voted for the Freedom Party increased their conspirary thinking more robustly than individuals who voted for other parties (see Figure 2). Finally, our data indicate that higher scores of conspiracy thinking were highly correlated with a decreased probabilty of intended vaccination. For exammple, the probability of reporting to get vaccinated decreases from over 80% in individuals with lowest scores of conspiracy thinking to below 40% for individuals with highest conspiracy thinking (see Figure 3).

One explanation for this findings can be found in Figure 4, which shows the credibility scores for different conspiracy beliefs in June. While the belief that the coronavirus was “bred in a Chinese lab and strategically spread” received the highest creadibility rating, it is closely followed by the believe that the epidemic is “strategically used to introduce mandatory vaccination”. Some 7 percent of the participants believed that this claim is highly credible (scored 5 on the 5-point credibility scale) and another 7 percent believe that it is somewhat credible (scored 4). Also among the top three conspiracy claims is the belief that “the pharma industry is responsible for the coronavirus epidemic”. Some 5 percent believe this is highly credible, and 8 percent believe its somewhat credible.

Taken together, the study underlines the importance to actively counter conspiracy thinking in times of an epidemic. Generally, voters of the right-wing populist Freedom Party seem to be most vulnurable, with a higher likelihood to believe such theories. The reason might be that conspiracy theories often resonate with anti-science components of populist communicaiton2. In governments, these anti-science policy preferences can do significant harm to public health. This was observable when the Freedom Party’s reneged the planned Austrian smoking ban law at the time the party was in charge of the healht ministry. The higher scores of conspiracy thinking among Freedom Party voters may also be explained by the more extreme ideological predispoitions, since more extreme indivudals may be more prone to use even absurd information as long as this information may support their prior beliefs3.

Notes:

1. Bruder, M., Haffke, P., Neave, N., Nouripanah, N., & Imhoff, R. (2013). Measuring individual differences in generic beliefs in conspiracy theories across cultures: Conspiracy Mentality Questionnaire. Frontiers in Psychology, 4, 225.

2. Mede, N. G., & Schäfer, M. S. (2020). Science-related populism: Conceptualizing populist demands toward science. Public Understanding of Science, 0963662520924259.

3. Taber, C. S., Cann, D., & Kucsova, S. (2009). The motivated processing of political arguments. Political Behavior, 31(2), 137-155.

Figure 1. Differences in conspiracy thinking across party voters at wave 1. Mean values with 95 percent confidence intervals. Dots represent single observations.

Figure 2. Changes in conspiracy thinking across party voters from wave 1 to wave 2 (OLS regression controlling for demographics variables and conspiracy thinking scores at wave 1). Average differences with 95 percent confidence intervals.

Figure 3. Relation between conspiracy thinking and the intention to get vaccinated against COVID19 at wave 2 (OLS regression controlling for party voting and demographics variables). Predicted probabilities with 95% confidence intervals.

Figure 4. Conspiracy beliefs at wave 2 (ranked based on mean credibility scores).

#7 Employment Transitions at the Beginning of the COVID19 Epidemic

15.05.2020

In this blog post, we are interested in employment transitions at the beginning of the COVID19 government measures. In particular, we look at group differences by gender, age, financial security, and social “safety” networks. Figure 1 shows the transition flows in the workforce between February and April, indicating an increasing precarisation. The figure shows that 37.5% of our respondents reported a transition to more precarious working conditions: 17.5% of previously full-time employed workers reported a transition into part-time, temporary or marginal employment, and 9.2% of those previously part-time employed switched to marginal- or temporary employment. Finally, 10.8% reported to have lost their jobs.

Employment transitions are characterised by pre-existing gender differences in employment conditions in the Austrian population. Close to 50% of women in Austria are in part-time employment, whereby this only applies for around 10% of the male population1. Not surprisingly, as shown in Figure 2, men were significantly more likely to transition from full-time to part-time employment conditions, whereas women were significantly more likely to transition from part-time- to temporary and minor employment. Although transitions into unemployment was more frequent for men, no statistically significant differences occurred.

Next, we look at age differences. We only included individuals who were 61 or younger, since the average retirement age in Austria revolves around 60.4 years2. To prevent biases in age effects respondents in education or retirement were excluded. As Figure 3 shows, individuals between 36 and 45 years were significantly less likely to transition into unemployment than the youngest (18-25 years) cohort and the second oldest (46-55years) cohort. Although not significantly higher compared to the other age groups, the 36-45 cohort also showed the most frequent descend from full-time to part-time employment. Further significant effects were found for the oldest cohort, which is less likely to transition to unemployment compared to the youngest cohort. However, transitions could be attributable to the pre-existing employment situations among the cohorts before the crisis, for unemployment is more common among younger and older population groups in Austira3. Thus, factors responsible for this distribution may further influence employment transitions. Nevertheless, this is something that should be watched closely, because after all, employers are encouraged to consider social indicators, such as family responsibilities and/or duration of employment (rewarding seniority), when terminating employment contracts. This is also formalised in the Austrian Labor Constitutional Act § 105 Abs. 3 ArbVG. For the remaining age groups and employment conditions no significant differences were identified.

To measure financial security, we asked our respondents if they had “financial reserves they could rely on after job loss”. Figure 4 displays the probability for employment transitions for those with and without such savings. Those who reported to have no savings were significantly more likely to transition to unemployment. However, financial reserves did not play a significant role for transitions within the employed. The branches most affected by the COVID19 measures are tourism (145,1% more unemployed compared to march 2019), construction (94,8%), transportation and logistics (+83,8%), art and entertainment (+49,1%), personnel agency (+34,0%) and system irrelevant parts of the retail sector (34,4%)4. Note that the concerned sectors also tend to have lower income structures5. Additionally, individuals with lower qualifications are most likely to be affected by job losses6. Combined with these numbers, our findings indicate that those who were already financially disadvantaged, e.g. had no possibilities to save money, are most severely affected by the pandemic.

Last, we examined the availability of social networks to cushion financial emergency situations. Respondents were asked if they could receive “financial support from relatives and / or friends in emergency situations?” Figure 4 suggests that especially those who have a weak social network were more likely to become unemployed or to descend from part-time to marginal/temporary employment conditions. We identified two explanations for this observation: First, the unavailability of a social network may lead to a higher likelihood of job loss. This is because social capital often protects from unemployment by lowering information and transition costs, easing worker placement and provide conflict management7. Second, the job loss itself may be experienced as a disruptive event, often related to feelings of shame and psychological distress, accompanied by the belief that this situation is temporary. Such feelings may discourage individuals to expect and seek financial support from others8,9,10.  

Conclusion: A not negligible part of the Austrian population faced employment transitions most likely caused by the COVID19 government measures. Many either experienced a deterioration within the scope of their working hours, or even lost their employment. As shown, the transitions occurred along pre-existing gender and age differences in employment conditions and resulted in increasing precarious working conditions (e.g. part-time, temporary- or unemployment). Our analysis has shown that males were more likely to descend from employment into part-time, marginal, or temporary employment conditions and females from part-time to marginal or temporary occupations. Furthermore, the youngest respondents (18-25yrs.), as well as the 46-55 years old were most likely to face unemployment. Additionally, the already financially disadvantaged (reported to inherit no savings) were most likely to lose their employment. Finally, we also found that respondents, who reported to have no social network to rely on in financial emergencies, are more likely to transition into unemployment.

Notes:

1. http://www.statistik.at/web_de/statistiken/menschen_und_gesellschaft/arbeitsmarkt/arbeitszeit/teilzeitarbeit_teilzeitquote/062882.html

2. Gumprecht, D. (2019): Pensionierungstafeln Bundesländer: Ergebnisse für das Jahr 2018. Statistik Austria, Vienna.

3. https://www.statistik.at/web_de/statistiken/menschen_und_gesellschaft/arbeitsmarkt/arbeitslose_arbeitssuchende/index.html

4. https://www.wifo.ac.at/news/corona-schock_auf_dem_arbeitsmarkt

5. https://www.statistik.at/web_de/statistiken/menschen_und_gesellschaft/soziales/personen-einkommen/allgemeiner_einkommensbericht/index.html

6. https://viecer.univie.ac.at/corona-blog/corona-blog-beitraege/blog09/

7. Freitag, M., Kirchner, A. (2011). Social Capital and Unemployment: A Marco-Quantitative Analysis of the European Regions. Political Studies, 59(2), 389-410. DOI: https://doi.org/10.1111/j.1467-9248.2010.00876.x

8. Atkinson, T., Liem, R., Liem, J.H., (1986). The Social Costs of Unemployment: Implications for Social Support. Journal of Health and Social Behavior, 27(4), 317. DOI: 10.2307/2136947.

9. Wanberg, C.R. (2012). The individual experience of unemployment. Annual review of psychology, 63(1), 369-396. DOI: 10.1146/annurev-psych-120710-100500.

10. Brand, J.E. (2015). The Far-Reaching Impact of Job Loss and Unemployment. Annual Review Sociology, 41(1), S. 359-375. DOI: 10.1146/annurev-soc-071913-043237.

Figure 1. Employment transition between the end of February and early April

Figure 2. Probability values (0.1 equals 10%) for employment transition regarding gender. Points show probability values with 95% confidence intervals.

Figure 3. Probability values (0.1 equals 10%) for employment transition regarding age groups. Points show probability values with 95% confidence intervals.

Figure 4. Probability values (0.1 equals 10%) for employment transition and availability of savings. Points show probability values with 95% confidence intervals.

Figure 5. Probability values (0.1 equals 10%) for employment transition with availability of social networks. Points show probability values with 95% confidence intervals.

#6 Fear, Perceived Susceptibility and Authoritarian Support

05.05.2020

In this blog post, we are interested in citizens’ fear of a COVID19 spread, their susceptibility beliefs and authoritarian support. To measure authoritarian support, we asked our survey participants whether they would support four types of more restrictive government measures. Figure 1 shows how strongly the participants agreed to these four measures. They could indicate their support on a 5-point agreement scale. From a normative democracy perspective, all of these measures may undermine basic democratic principles. Overall, support for these measures was mixed to weak. Most people completely disagreed that the government should govern without the parliament (60%) and only 3.5% completely agreed that the government should do so. Further restricting personal freedom was perceived to be the most acceptable restrictive action. Only 22% completely disagreed with this measure, and even some 16% completely agreed that the government should do so.

In the next step, we were interested in how different groups of individuals varied in terms of their agreement with these measures. We combined the four questions on restrictive measures (see Figure 1) into a single “authoritarian support” scale. Then, we looked at how self-reported fear (“I am afraid that COVID19 will further spread in an uncontrolled way”) and perceived personal susceptibility (“I think it’s very likely that I will get infected”) were associated with authoritarian support. As Figure 2 shows, individuals who had higher levels of fear and felt more susceptible to become infected with COVID19 had higher mean values of authoritarian support than individuals with lower fear or susceptibility scores.

Conclusions: Our data indicate that the more citizens fear a further COVID19 spread and believe that they are susceptible to infection, the more willing they are to support more authoritarian measures. Thus, government representatives are well advised to communicate the risks associated with COVID19 in a responsible manner. Furthermore, the media needs to closely monitor antidemocratic tendencies which may arise in climates of fear. These relations may also be important to explain and inform recent authoritarian shifts in less stable democratic or semi-democratic countries, such as Hungary or Russia.1

Notes:

1. https://www.theguardian.com/world/2020/mar/31/coronavirus-is-a-chance-for-authoritarian-leaders-to-tighten-their-grip

 

Figure 1. Agreement with the four restrictive measures. Different colours represent sample proportions along the 5-point agreement scales.

Figure 2. Mean values of authoritarian support for individuals with different levels of fear of COVID19 spread and perceived personal susceptibility. The two questions were measured on 5-point agreement scales (1 = don’t agree, 5 = agree).

#5 Misinformation & Ideology

27.04.2020

In this blog post, we are interested in what kind of groups are more likely to believe so called “fake news”. The term fake news usually refers to totally made up information and is often used by populist politicians, who try to undermine trust in mainstream media1. We thus speak of misinformation in this post. Figure 1 shows the credibility ratings for different misinforming claims which circulated at the peak of the epidemic (some are still circulating). Of these examples, the claim that the corona virus was created in Chinese lab and was then strategically disseminated by the Chinese was most credible. Only 48% of all survey participants judged this claim as entirely non-credible (marked 1 on the 5-point scale). The second most credible claim was that alcohol sprays would protect from infection, closely followed by the belief that the virus was spread to counter population aging. The least credible claim was that a hot bath would protect from an infection. 81% judged this claim to be entirely non-credible.

Recent research indicates that people on the right end of the ideological spectrum and right-wing populist voters in particular might be more likely to believe in unsubstantiated news claims.2,3 We created a mean index variable composed of the credibility ratings for the six false claims. Figure 2 (left plot) shows the index mean values for individuals across the left-right ideological spectrum.4 The values indicate that credibility scores are lowest in moderate left-wing individuals and highest in extreme right leaning individuals. Interestingly, this pattern remains statistically significant even when adjusting for age, gender, education, knowledge and news use frequency. The reason might thus lie in the socio-psychological profile or the specific media use of right-wing individuals.

Finally, Figure 2 (right plot) also shows variation in credibility across voter groups.5 In line with the ideology finding, voters of the Austrian Freedom Party judged the presented misinformation as most credible. Interestingly, also the left-wing Social democrats had comparably high ratings. However, once controlling for demographic factors and knowledge, the differences between the Social Democrats and other mainstream parties become weak and insignificant, while the Austrian Freedom Party score remains significantly higher compared to all other parties.

Conclusion. Credibility ratings for the presented misinformation are overall low, but were more persistent in some groups. The analysis indicates that right-wing individuals, especially right-wing populist voters, were more likely to believe misinformation. One explanation might be that far right-wing voters tend to be more critical towards mainstream, especially public service, media; and right-wing populist voters tend to use social media heavily for political purposes. In a recent blog post by the Vienna Center for Electoral Research, Eberl et al. (2020) show that citizens who get information on public broadcasting TV have lower rates of believing false claims about the Corona virus epidemic; citizens who get information on social media, by contrast, tend to have higher ratings of believing such claims.6 Far-right voters’ scepticism toward public broadcasting TV (ORF) and their heavy use of social media may thus be two factors which contribute to their susceptibility to misinformation.

Notes:

1. Egelhofer, J.L., Lecheler, S. (2019). Fake news as a two-dimensional phenomenon: a framework and research agenda. Annals of the International Communication Association, 43(2), 97-116.

2. Allcott, H., & Gentzkow, M. (2017). Social media and fake news in the 2016 election. Journal of Economic Perspectives31(2), 211-36.

3. Guess, A., Nagler, J., & Tucker, J. (2019). Less than you think: Prevalence and predictors of fake news dissemination on Facebook. Science Advances5(1), eaau4586.

4. Note: A total of 124 individuals chose not to respond their ideology and were excluded from the analysis.

5. Note: A total of 253 individuals chose not to report their vote choice, did not vote, or voted for other parties. These individuals were excluded from the analysis.

6. Eberl, J.M., Lebernegg, N.S., & Boomgaarden, H.G. et al. (2020, April 25). Alte und Neue Medien: Desinformation in Zeiten der Corona-Krise.

Figure 1. Credibility judgements rated on the 5-point scale (1 = not credible, 5 = credible).

 

Figure 2. Differences in credibility ratings for the presented misinformation across the ideological spectrum and voter groups (mean values with 95% confidence intervals).

#4 Risk groups and self-protective behaviour

20.04.2020

In this blog entry, we are interested in the self-protective behaviour of risk groups. Participants were asked whether they suffer from diabetes, hepatitis B, chronic obstructive pulmonary disease, chronic kidney disease, cancer, hypertension or cardio-vascular problems. They were assigned to the risk group if they marked at least one of these diseases. We also present results for different age groups, for progressing age may also be related to more severe COVID19 outcomes. But before going into how these groups differ, Figure 1 shows how frequently the measured behaviours were performed by our survey participants. The figure shows that more of the people adhere to these new behavioural norms. For example, only around 3% report to never or rarely keep 1-2 meter distance to others; some 7% report to never or rarely have no or try to strongly limit contact with family members or friends; some 8% report to never or rarely go out for urgent matters only; and only 4% report to never or rarely wash their hand properly multiple times per day.

People with chronic diseases and older individuals might be in risk of more severe COVID19 outcomes. But do follow self-protective behaviours more rigorously than others? Figure 2 indicates the probability that individuals respond with “always” (as opposed to “most of the time”, “rarely” or “never”). The results show that the defined risk group does not strongly differ from the non-risk group in terms of their behaviour. They even score significantly lower on the reduction of their social contacts (61%: risk group; vs. 70%: others). They may score low either because they do not feel the need to (i.e. ill-informed) or are not able to reduce their social contacts (the latter e.g. for caring needs). Other differences between risk and non-risk groups are insignificant. Older age groups significantly differ from younger age groups in their more rigorous adherence to keeping distance and limiting social contacts.

Figure 3 indicates the probability that individuals respond to either “always” or “most of the time” (as opposed to “rarely” or “never”). Here we can observe clearer differences between risk and non-risk groups. Risk groups are significantly less likely to respond to “most of the time” or” always” when it comes to distancing behaviour (95%: risk group; vs. 99%: others), going out for urgent matters only (89%: risk group; vs. 94%: others). Differences for handwashing and the reduction of social contacts remain less clear. For age groups, we observe significant upwards trend for all behavioural patterns (but a slightly weaker trend for “going out for urgent matters only”).

Conclusion: Our data indicate that some risk groups may either not be able or sufficiently informed to perform the risk-reducing behaviour. However, the protection of risk groups will become even more important after a step-by-step reopening of the economic and social life. Potential risk groups may thus need to be targeted more effectively, e.g. by using group-specific communication and by creating environments which enable the self-protection of risk groups.

Figure 1. Sample frequencies for different behaviours

Figure 2. Probability values (0.1 equals 10%) for responding “always” to the four behavioural questions. Points show probability values with 95% confidence intervals.

 

Figure 3. Probability values (0.1 equals 10%) for responding “most of the times” or “always” to the four behavioural questions. Points show probability values with 95% confidence intervals.

 

#3 Trust among voter groups

17.04.2020

In this blog post, we are interested in whom party voters trust in times of the Corona epidemic. We asked our participants: When it comes to issues related to the Corona virus epidemic, how much do you trust … a. government officials, b. journalists, c. scientists, and d. policemen and -women (1 = don’t trust, 5 = trust very much). Furthermore, we asked them which party they voted for in the last national election (253 individuals who opted out, did not vote or voted for ‘other parties’ were excluded from the analysis). The results indicate comparably high rates of trust for government officials, scientists and policemen/women, and relatively low levels of trust in journalists. Voters from the right-wing populist Austrian Freedom Party trust least in journalists and government officials. Compared to other party voters, they also have significantly less trust in scientists. However, right-wing populist voters do not differ in terms of their trust in policemen/women from most other parties (except conservatives, who have exceptional high trust in the police force).

Conclusion: The comparably high rate of trust in government officials may be explained by an Austrian ‘rally around the flag’ effect. Right-wing populist voters had lower trust rates for government officials, journalists and scientists, which may be explained by the “anti-elitist” nature of right-wing populist thinking.1 While this is only a cross-sectional snapshot, we will also look into how trust will develop in these voter groups over the next months. We are looking forward to present more details on these developments after the second panel wave this summer.

Notes:

1 Heiss, R., & Matthes, J. (2019). Stuck in a nativist spiral: Content, selection, and effects of right-wing populists’ communication on Facebook. Political Communication.

Figure 1. Trust in government officials, journalists, scientists, and policemen/women across party voter groups. Mean values with 95% confidence intervals shown.

#2 Information Sources, Knowledge, and Misinformation Sharing

16.04.2020

In the course of the Corona epidemic, misinforming content about government measures and treatment have gone viral on social media. Especially the widely used instant messaging service WhatsApp has been identified as a key platform for misinformation.1 In this blog entry, we are interested in the predictors of misinformation sharing behaviour. We asked our survey participants how often it happened that “others have told me that information I have shared a. was exaggerated, b. was not entirely correct, c. was completely made up”. In our sample, only 55 to 60 percent responded that this has never happened to them (scored 1 on the 5-point scale, see Figure 1).

We then combined the three questions into a single “shared misinformation index” and looked at which groups of people reported higher scores on misinformation sharing. Figure 2 indicates that age and education rarely explain any variance. Interestingly, younger people reported slightly higher scores than older citizens, which, however, might be related to their more frequent social media use. More importantly, COVID19 knowledge and policy knowledge are the strongest predictors: The higher COVID19 or policy knowledge, the lower the score for misinformation sharing. Interestingly, shared misinformation scores get slightly lower for citizens with zero knowledge, maybe because these individuals do not engage with the issue at all. We measured COVID19 (e.g. meaning of herd immunity) and policy knowledge (e.g., regulations for public events) with 6 multiple choice quiz questions each and created two summative indices (0 = all questions wrong, 6 = all questions correct).

Finally, we were also interested in how information behaviour is related to the two knowledge scores, which predicted lower levels of misinformation sharing. We asked how frequently the citizens in our panel reported to use social media, internet websites, newspapers, private TV and public broadcasting service TV (PSB) to get information on the Corona epidemic. They could rate the frequency of their media use for each source on a scale from 1 (never) to 5 (often). We ran a regression predicting knowledge scores from these media use patterns and adjusted for age, gender, education, and overall news use frequency. Figure 3 indicates PSB TV use is a consistent positive predictor of knowledge. Social media use is a consistent negative predictor of knowledge. General internet use significantly predicts higher COVID19 knowledge and private TV use is a significant negative predictor of COVID19 knowledge.

Taken together, our data indicate that individuals who rely on PBS TV to get information on the current Corona epidemic are better informed than individuals who do not use PBS, while citizens who mostly rely on social media have less knowledge. As a result, reliance on social media for Corona information may make individuals specifically likely to share misinformation. Our data thus may support arguments which highlight the importance of a strong PBS in times of a crisis, and its relevance in disseminating high quality information to counter the spread of misinformation.

Notes:

1 https://www.sueddeutsche.de/medien/coronavirus-fake-news-whatsapp-1.4858827

2 Chadwick, A., & Vaccari, C. (2019). News sharing on UK social media: Misinformation, disinformation, and correction. Loughborough University.

 

Figure 1. Sample frequencies for sharing misinformation.

Figure 2. Age, education and knowledge as predictors for sharing misinformation. Y-axis represents the shared misinformation index, ranging from 1 to 5. Error bars indicate 95% confidence intervals.

Figure 3. Coefficient plots: If the point-value is negative, then more frequent media use is negatively related to knowledge; if the value is positive, then more frequent use is positively related to knowledge. The coefficients indicate how knowledge scores (on the 0 to 6 knowledge scales) change when the media use score increases by 1 unit (on the 5-point frequency scale). Error lines indicate 95% confidence intervals around the coefficients.

#1 Skepticism towards government measures

15.04.2020

In this first blog entry, we look at the number of people who report to be affected in some way be the Corona epidemic and how affection related to scepticism towards government measures. Around 28 percent of the participants reported to have less income due to the Corona epidemic. Some 24 percent of the participants say that they now have troubles making their regular payments and 10 percent reported to have lost their job due to the epidemic (see Figure 1.). A smaller share had to switch to part-time work. Taken together, there are reasons to assume that the measures which are necessary to contain the Corona epidemic are creating social problems for many people.

But how do those affected rate the restrictive government measures? Participants were asked how sceptical they were towards the government measures. They could rate their scepticism on a scale of 1 to 5. Scepticism was measured with three questions which were combined to a single index (government measures are… a. …exaggerated, b. …counterproductive, and c. …ill-conceived).

The results show that scepticism towards the government measures is still overall low. However, scepticism is higher among those who are economically affected by the epidemic. For example, scepticism is significantly higher for those who lost their job, had to switch to part-time work, have less income, trouble with payments or who are afraid of losing their flat. Unsurprisingly, people who report to have no garden/balcony report higher scepticism. These are often individuals who have smaller flats and who cannot compensate the lack of access to public spaces with private open-air space (see Figure 2.).

Conclusion: Even though scepticism towards government measures is still low, there is reason to believe that the more people become negatively affected by the measures, such as by job loss or problems in paying bills, the more scepticism may grow. Communicating with these vulnerable groups and responding to their needs will be crucial for the further implementation of necessary measures to contain the epidemic.

Figure 1. Frequencies (in sample percent).

Figure 2. Mean values of scepticism towards government measures across people who are more or less affected by the epidemic. Error bars indicate 95% confidence intervals.

Cookies make it easier for us to provide you with our services. With the usage of our services you permit us to use cookies.
Privacy Policy