Marriage and cohabitation under uncertainty: the role of narratives of the future during the COVID-19 pandemic

ABSTRACT

This paper addresses the impact of COVID-19-induced uncertainty on union formation intentions in Italy. We acknowledge that decisions made in uncertain conditions rely on personal narratives of the future, that is socially-constructed contingent plans for achieving a personal imaginary. The data come from an on-line survey experiment carried out during the final week of lockdown in Italy on a sample of 1,846 individuals in a romantic relationship (cohabiting or living apart together). Our findings suggest that narratives of the future have a causal effect on marriage intentions: expectations of a long wait before the return to pre-pandemic conditions negatively influence marriage intentions. On the other hand, cohabitation seems more compatible with the uncertainties of today’s world. The present study gives a prominent role to the future in family formation practices, net of more ‘traditional’ factors that have been considered in the literature on family life courses.

KEYWORDS:

• Marriage
• family
• cohabitation
• COVID-19
• uncertainty
• coronavirus

Introduction

In this paper we contribute to the growing literature on the uncertainty/family dynamics nexus by addressing the impact of COVID-19-induced uncertainty on union formation intentions in Italy. We do this through an on-line survey experiment, which allows for an assessment of causation.

Several studies have examined the role of uncertainty on family dynamics by studying how fertility and family formation behaviours differ depending on individuals’ labour market conditions, such as, for instance, being unemployed or holding a fixed-term contract job (Kreyenfeld et al. 2012; Mills and Blossfeld 2013); or – at best – individuals’ perceptions of their current socioeconomic status (Fahlén and Oláh 2018). These operationalizations of uncertainty, however, do not give enough credit to the very nature of uncertainty: uncertainty, after all, primarily means a lack of clarity about future prospects (Beckert 2016). Recent advances in family demography posit that actors’ choices are influenced by the ‘shadow of the future’ (Bernardi et al. 2019), and the Narrative Framework (for more details, see Vignoli et al. 2020a, 2020b) provides the concepts necessary for operationalizing its influence in family dynamics.

When people face an uncertain situation, they take into account past experiences and present status (the ‘shadow of the past’) but also future expectations, which represent what people expect will happen in the future based on the information to hand. The expected duration of the COVID-19 emergency and its socioeconomic consequences may, for instance, influence family plans. However, expectations are not the only source for the influence of the future in the decision-making process: humans may shift from the expected course of action thanks to their imaginative capacity (Dewey 1922 [1957]; Mead 1932 [2002]). The imaginative capacity allows people to envisage a different future from what can be expected and then try to reach it (Beckert 2016). A family imaginary, in particular, is related to the type of family that people wish to have in their future. Family plans are often imbued of imagined futures that may become a conscious source of aspirations for people, and support their motivations even in adverse conditions (Tuckett and Nikolic 2017; Vignoli et al. 2020a). During the difficulties of the COVID-19 emergency, for instance, individuals may decide to cohabit or to get married because they attach importance to union formation; or, on the contrary, a family imaginary may revolve around the desire to remain single by choice, notwithstanding favourable socioeconomic conditions. Imaginaries, indeed, may provide life goals, irrespective of the shadow of the past and (more or less plausible) expectations.

The shadow of the past and the shadow of the future that influence family decisions find their synthesis in the personal narratives of the future, that reflect the contingent plan for reaching the goals set by imaginaries. Although personal narratives of the future represent individual plans, they are deeply influenced by social context. During the COVID-19 emergency, for instance, although the majority of the population was not directly exposed to the SARS-CoV-2 virus and its health consequences, people were exposed to government restrictions and to media-channelled shared narratives of an uncertain future (Vignoli et al. 2020b). These narratives naturally influenced their personal narratives of the future. Widespread uncertainty may have increased individuals’ concerns about the future and, thus, pushed them to reconsider their family plans.

In what follows, we consider the COVID-19 disaster in Italy as an exogenous uncertainty shock that allows for the operationalization of narratives of the future, and for their impact on the family formation decision-making process to be assessed in causal terms.

Marriage and cohabitation in uncertain times

The most prominent theoretical framework for the study of ‘new’ family patterns is the Second Demographic Transition (hereinafter SDT; see Lesthaeghe 2020 for a recent update). This accounts for the diffusion of cohabitation and the decline of marriage by stressing the role of ideational factors such as the rise of post-materialist values and secularization.

The SDT, however, does not see the rise of (economic) uncertainty as a driver for the diffusion of cohabitation. On the contrary, according to the ‘Pattern of Disadvantage’ (hereafter POD) hypothesis (Perelli-Harris and Gerber 2011), among socioeconomically deprived groups – e.g. among lower educated individuals – or during periods of economic crisis, cohabitation may be preferred over marriage because it is temporary and more easily reversible (Mills and Blossfeld 2013). Individuals do not necessarily reject marriage and long-term commitments, but they might decide to postpone a wedding until their future prospects are clearer or until they become more settled in the labour market.

Prior research generally supports the view that poor economic prospects for men and women are associated with a delay in marriage in favour of cohabitation (Sassler and Lichter 2020). In light of this empirical evidence, one might hypothesise that the uncertainty induced by the COVID-19 pandemic would be negatively associated with marriage and positively associated with cohabitation decisions.

Italy, a case study

Italy represents a unique case study both in terms of the diffusion of cohabitation and with respect to the ongoing pandemic. In this country, unmarried cohabitation is far less common than elsewhere in Europe, but the diversity of union patterns is growing, and a slow but continuous process of secularization is at play (Pirani and Vignoli 2016). Thus, the focus on Italy allows us to analyse the link between uncertainty and the family formation process in a society undergoing secularization and revolutionary family changes. Previous evidence showed that labour market uncertainty – i.e. unstable employment or unemployment – favours cohabitation, while employment stability facilitates marriage (Vignoli et al. 2016), in line with the POD hypothesis.

Italy also offers a unique case study as far as the COVID-19 pandemic and its outbreak responses are concerned. The country had the first severe case of COVID-19 pandemic in the Western world, and Italians faced the longest complete lockdown, which started on the 9th of March and ended on the 4th of May 2020. At the time of writing (July 2020), the number of official deaths stands at almost 35,000, with approximately 240,000 individuals having tested positive for the virus. These figures are difficult to compare across countries, not least because of different ways of identifying and recording the exact cause of death. However, the spread of numbers like these through the media – for whom trends in Coronavirus diffusion were a major topic of interest (Baker et al. 2020) – are likely to have shaped individuals’ perception of the emergency.

We hypothesise that this enormous uncertainty shock has affected family formation processes in Italy, and that personal narratives of the future – which are also influenced by the media – are crucial for grasping its effect on marriage and cohabitation decisions. At the same time, individuals whose family imaginary entails a stronger desire to cohabit and/or marry their current partner are meant to keep their higher pre-pandemic union intentions, regardless of COVID-induced uncertainty. To test these hypotheses, we applied an experimental approach to the unique data we collected during the lockdown.

Data, variables and methods

Sampling and data collection

The data come from an on-line survey carried out by the international survey company Lucid, which has a strong academic reputation for its high-quality and rigorous data collection. The survey was carried out between the 25th of April and the 1st of May 2020, that is during the final week of the Italian lockdown. We used quota sampling targeting men and women aged 20–40, regardless of their living arrangements and partnership status. Based on data from the Italian National Institute of Statistics (ISTAT), we had set quotas proportional to: gender, age, and area of residence. Given the heterogeneous impact of COVID-19 across Italian areas, we set quotas for provinces (NUTS-3) in the Northern part of Italy (including the Marche region) and regions (NUTS-2) in the Central and Southern part of the country (including Sicily and Sardinia).

Respondents who provided deliberately fatuous answers had their answers filtered out. We also discarded interviews that were shorter than three minutes – the average duration of the interview in the final sample is approximately eight minutes. The final sample consisted of 4,039 individuals; the analytical sample of this study includes 1,846 individuals who declared themselves to be in a romantic relationship, of not less than three-months duration (i.e. at least since the beginning of 2020) without being married to the partner. Of these 750 were cohabiting, whereas 1,096 were living apart together (LAT). The representativeness of our sample in terms of individuals’ socioeconomic status, operationalized through a measure of net monthly household income, is quite satisfying: the sample median is approximately 1,800€, whereas, according to ISTAT, in 2017 the median net monthly household income in Italy was approximately 2,100€. Considering the negative impact that the lockdown has had on the labour market situation of many Italians, our sample median may be considered in line with the ‘true’ population value.

Experiment description

Self-reported considerations on the role of the media in personal decisions may be biased by a low level of consciousness of these cognitive processes and the desire to show personal control over one’s own decisions. We addressed these issues by using an experimental method that simulates a ‘real’ exposition of the respondents to a new media narrative, and then we measured (possible) changes in union intentions. Respondents were exposed to a mock news bulletin concerning the expected end of the pandemic emergency, according to a task force made up of leading coronavirus experts in Italy. We opted for this treatment because it should have sounded quite realistic for the respondents. In fact, a few days before data collection the Italian Prime Minister had announced a task force of academics and other prominent experts to address the COVID-19 emergency, and to provide guidelines for the return to normality. Respondents were randomly assigned one of five treatments, each presenting a different expected duration before the return to normality: three months, six months, one year, two years, or more than two years.¹ The five scenarios represent the main independent variable of this study.²

After being exposed to the treatment, respondents were asked about their cohabitation – if LAT – and marriage intentions – all selected respondents – in the next three years in light of the expected duration of the emergency. We then compared their post-treatment and pre-treatment intentions. Thus, the dependent variables of this study are two variables measuring the post-treatment change in the intentions to cohabit and/or marry with the current partner. Both the pre- and post-treatment variables concerning union intentions range from 0 (‘not at all’) to 10 (‘definitely yes’), so that our dependent variables range from −10 to 10.

A second, crucial independent variable relates to family imaginaries, which may entail different aspects of the family life-course (e.g. union formation, fertility, or housing). In this study, we focus on the importance respondents attach to starting cohabitation and/or to marrying their current partner (on a scale from 0 to 10) as a proxy of the relevance of union formation in their family imaginary.

One of the advantages of experiments is that they do not require sophisticated modelling strategies because of the random assignment of the respondents to the different scenarios: a comparison of post-treatment changes in cohabitation/marriage intentions would suffice. We do, however, include in our models a set of control variables, mostly to reduce the statistical uncertainty around the estimated treatment effects. In the survey we included several questions related to the ‘shadow of the past’, that is: personality traits (risk aversion); sociodemographic background; the pre- and post-pandemic socioeconomic status of the respondent and his/her partner; and feelings of insecurity about several life domains. However, since we do not analyse union intentions tout-court, but post-treatment change in union intentions, controlling for baseline intentions at interview, these had little impact on our dependent variables.

Methods

The causal impact of the exogenous narrative about the future of the pandemic on cohabitation intentions is assessed by means of an Ordinary Least Squares regression applied to the 1,096 respondents in a LAT relationship at the time of the interview, specified as follows: (1) $\mathrm{\Delta }Cohabitation\mathrm{\_}intention{s}_{t1-t}=Treatment+Imaginary+Cohabitation\mathrm{\_}intention{s}_t+Recall+Xb$(1) where ΔCohabitation_intentions_t1-t represents the difference between post- and pre-treatment cohabitation intentions. Treatment corresponds to a set of dummies representing the randomized scenarios, Imaginary a continuous variable for the importance attached to start cohabiting with the partner. Cohabitation_intentions_t is a continuous variable for the level of pre-treatment cohabitation intentions. Recall is a dummy variable taking the value 1 for those respondents who were not able to recall exactly which type of scenario they have been exposed to (see note 1). In a first step, equation (1) is estimated without control variables, whereas in a second step the model is augmented with Xb, which represents a set of coefficients for additional control variables included in the model (sex, age and age², risk aversion, area of residence, educational attainment, and both partners’ employment condition before and after the pandemic outbreak). To explore possible heterogeneity in treatment effects, the full model is also separately estimated for men and women, and for respondents with and without tertiary education.

The model for the analysis of marriage intentions, including all 1,846 respondents, is specified as follows: (2) $\mathrm{\Delta }Marriage\mathrm{\_}intention{s}_{t1-t}=Treatment+Imaginary+Marriage\mathrm{\_}intention{s}_t+Recall+Cohabiting+Xb$(2) where all variables have the same meaning as in equation (1), though referring to marriage. The only additional variable is Cohabiting, which is a dummy variable distinguishing respondents who are cohabiting form those who are in a LAT relationship. Also equation (2) is estimated in two steps, with and without control variables, and then separately by sex and by the attainment of a university degree.

Results

Table 1 presents the results concerning cohabitation intentions. In Model 1, which includes all respondents in a LAT relationship, we clearly see that the treatment does not have any impact on the outcome variable. On the contrary, respondents attaching more importance to their family imaginary of cohabitation hold higher post-treatment intentions. Considering that the outcome variable is the difference between post- and pre-treatment cohabitation intentions, the effect of pre-treatment intentions is, of course, negative: those with higher intentions are structurally more at risk of reducing said intentions. In Model 2 we include a set of control variables, which affect our estimates only marginally, confirming the exogeneity of the treatment. In fact, among the controls, the only variable with a substantial effect is age: regardless of the scenario they were exposed to, people in their early ‘30s in a LAT relationship are the most likely to hold higher post-treatment cohabitation intentions, which may be interpreted as a straightforward life-cycle effect.³ In Models 3 and 4 we present results for men and women separately. Among men, notwithstanding the high level of uncertainty around our estimates, results point to somewhat higher cohabitation intentions among those exposed to the most pessimistic scenario (more than two years). Among women, coefficients do not show any pattern. When distinguishing between respondents who did not (Model 5) and those who did attain tertiary education (Model 6), results confirm an overall lack of statistical and substantial significance. We do find some positive effects from more pessimistic scenarios among individuals with less than tertiary education, whereas the opposite holds true among those with a tertiary degree. However, treatment effects are not monotonic and are only marginally statistically significant.

Table 1. The effect of different randomized scenarios concerning the end of the pandemic on cohabitation intentions (post-treatment – pre-treatment).

	M1	M2	M3	M4	M5	M6
	Pooled	Controls	Men	Women	No tertiary	Tertiary
Treatment (ref.: 3 months)
6 months	0.091	0.039	0.117	−0.066	0.531*	−0.517
	(0.237)	(0.235)	(0.337)	(0.331)	(0.319)	(0.342)
12 months	0.076	−0.048	0.325	−0.420	0.253	−0.401
	(0.236)	(0.234)	(0.349)	(0.327)	(0.360)	(0.308)
2 years	−0.026	−0.076	0.127	−0.298	0.303	−0.543*
	(0.235)	(0.233)	(0.359)	(0.311)	(0.342)	(0.324)
>2 years	0.305	0.296	0.551*	−0.019	0.446	0.004
	(0.233)	(0.230)	(0.325)	(0.325)	(0.333)	(0.329)
Imaginary (cohabit)	0.364***	0.379***	0.412***	0.347***	0.368***	0.406***
	(0.042)	(0.042)	(0.055)	(0.062)	(0.063)	(0.053)
Cohabit intentionst	−0.507***	−0.553***	−0.558***	−0.538***	−0.542***	−0.581***
	(0.033)	(0.036)	(0.050)	(0.050)	(0.051)	(0.049)
Recall	−0.427**	−0.435**	−0.538*	−0.328	−0.786**	0.062
	(0.213)	(0.213)	(0.289)	(0.314)	(0.314)	(0.278)
Obs.	1096	1096	544	552	577	519
R-squared	0.265	0.304	0.317	0.329	0.303	0.347

Robust standard errors in parentheses. ***p < 0.01, **p < 0.05, *p < 0.1.

Models M2 to M6 control for: sex; age and age²; risk aversion; area of residence; educational attainment; and both partners’ employment condition before and after the pandemic outbreak.

Overall, COVID-induced uncertainty does not seem to play an important role as far as cohabitation intentions are concerned. This may be due to two diametrically-opposed mechanisms. On the one hand, the pandemic may discourage family formation: but, on the other hand, it may favour cohabitation as a living arrangement more compatible with a condition of uncertainty, especially among lesser educated men; something in line with the POD hypothesis.⁴

In Table 2 we show the results obtained applying equation (2) to our data. Model 1, including both respondents in a LAT relationship and those cohabiting, shows that the treatment exerts a monotonic, negative influence on marriage intentions. Negative effects become statistically and substantially significant only for scenarios of at least 12 months before the return to normality. The influence of the family imaginary related to marriage is again positive, and of similar intensity compared to the previous model referring to cohabitation imaginary. The same holds true for the negative effects of pre-treatment marriage intentions. Individuals cohabiting at interview report higher marriage intentions, but the effect vanishes in Model 2, which adds the control variables. The effects of the treatment variable are only slightly modified by the inclusion of the control variables. To provide a more substantial interpretation of the intensity of the treatment, we estimated a multinomial logistic regression on the probability of decreasing, increasing or holding the same marriage intentions before and after the treatment, specified as Model 2. Results (available upon request) showed that those exposed to the worse scenario, compared to their counterparts exposed to the most optimistic scenario, are approximately 10 p.p. more likely to have decreased their pre-treatment marriage intentions, and are approximately 11 p.p. less likely to have increased them.

Table 2. The effect of different randomized scenarios concerning the end of the pandemic on marriage intentions (post-treatment – pre-treatment).

	M1	M2	M3	M4	M5	M6
	Pooled	Controls	Men	Women	No tertiary	Tertiary
Treatment (ref.: 3 months)
6 months	−0.068	−0.080	0.187	−0.283	−0.150	−0.015
	(0.159)	(0.156)	(0.230)	(0.213)	(0.211)	(0.229)
12 months	−0.399**	−0.498***	−0.300	−0.678***	−0.550**	−0.521**
	(0.170)	(0.167)	(0.224)	(0.247)	(0.231)	(0.244)
2 years	−0.628***	−0.608***	−0.578**	−0.656***	−0.787***	−0.519**
	(0.170)	(0.166)	(0.232)	(0.235)	(0.228)	(0.245)
>2 years	−0.694***	−0.711***	−0.501**	−0.879***	−0.933***	−0.523*
	(0.184)	(0.179)	(0.234)	(0.276)	(0.239)	(0.270)
Imaginary (marriage)	0.311***	0.340***	0.294***	0.383***	0.309***	0.390***
	(0.022)	(0.023)	(0.031)	(0.032)	(0.032)	(0.033)
Marriage intentionst	−0.467***	−0.523***	−0.454***	−0.577***	−0.489***	−0.565***
	(0.023)	(0.024)	(0.033)	(0.033)	(0.032)	(0.035)
Recall	0.221	0.160	0.138	0.162	0.003	0.418
	(0.192)	(0.189)	(0.235)	(0.299)	(0.251)	(0.289)
Cohabiting	0.319***	0.051	0.069	−0.029	0.030	0.043
	(0.115)	(0.123)	(0.168)	(0.182)	(0.169)	(0.187)
Obs.	1846	1846	861	985	1035	811
R-squared	0.256	0.305	0.289	0.338	0.288	0.354

Robust standard errors in parentheses. *** p < 0.01, **p < 0.05, *p < 0.1.

Models M2 to M6 control for: sex; age and age²; risk aversion; area of residence; educational attainment; and both partners’ employment condition before and after the pandemic outbreak.

Models 3–6 suggest that, though COVID-induced uncertainty negatively influences marriage intentions regardless of sex and education, the effects are stronger among women and the less-educated.⁵ In fact, in this case the larger samples allowed for the estimation of a model only for women without tertiary education (N = 545): coefficients associated to the different scenarios are −.50*, −.80**, −97***, and −1.30***, suggesting interaction effects between sex and education. In additional models (available upon request), we also implemented separate models by area of residence (Northern, Central and Southern Italy): the patterns of treatment effects came out to be very similar, notwithstanding the stronger impact of the pandemic in Northern regions, the heterogeneous diffusion of cohabitation, and the different importance attached to the ‘traditional’ marriage and its ceremony.

Conclusions

At least two findings clearly emerge from this short paper. First, our results accord with Vignoli et al. (2016), and support the idea that in Italy cohabitation – in contrast to marriage – is more compatible with the uncertainties of today’s world. They favour a POD-like interpretation of family formation practices in the country: young adults faced with blocked opportunities, and especially the lower educated, might prefer cohabitation to marriage due to its lower level of commitment and due to its more uncertain nature. Alternatively, they might decide to postpone marriage until they are more optimistic about their prospects. During the lockdown, millions of suspended employees had to rely on wage guarantee funds, with a substantial reduction of the usual salary, whereas many workers with temporary contracts lost their job. Thus, respondents exposed to the more pessimistic treatment scenarios may have anticipated possible job losses due to government restrictions and shrinking household income, factors often viewed as prerequisites for marriage. This may hold especially true for (low-educated) women, who may have reacted more negatively than men, in terms of marriage intentions. Labour market precariousness is, indeed, often gendered, with women more exposed to employment uncertainties (such as holding a fixed-term contract) than men. This situation has been found to be particularly detrimental for women’s family plans (Alderotti et al. 2020; for Italy, see Vignoli et al. 2016). Given the long-term socioeconomic consequences of the COVID-19 pandemic, we expect that uncertainty will continue to be important in shaping union formation practices in the years to come.

Second, the results show how important narratives of the future are for the family formation decision-making process. Regardless of the exact mechanisms through which our treatment influences marriage intentions, which can only be hypothesized, a simple narrative of rising future uncertainty, in the form of a (mock) news bulletin concerning the future development of the pandemic, showed to be sufficient to negatively influence marriage intentions, over and above respondents’ current socioeconomic status. The experiment demonstrates a causal effect of narratives of the future on marriage intentions: a long expected duration before the return to pre-pandemic conditions negatively influences marriage intentions. However, personal imaginaries play an independent role here: a positive family imaginary related to cohabitation or marriage encourages cohabitation and marriage intentions even in uncertain times. This study is a first attempt to apply empirically the Narrative Framework to the study of family dynamics. Findings suggest a prominent role for the shadow of the future in marriage intentions, net of more ‘traditional’ factors, belonging to the shadow of the past, that have been considered in the previous literature on family life courses.

The paper clearly has limitations. First, the on-line survey took place in a moment in which only a few people were allowed to attend a wedding. Hence, the negative effects on marriage may only be short-term effects due to couples postponing the wedding given the impossibility of organizing a wedding reception. In any event, qualitative research on marriage and cohabitation suggests that the traditional ceremony is often imagined as being quite expensive. This, in fact, is seen by informants as a direct cause of the postponement of marriage until the couple is economically ‘ready’ (Gibson-Davis et al. 2005; Smock et al. 2005; for Italy, see Vignoli et al. 2016). Consequently, given the economic consequences of the COVID-19 emergency and its outbreak responses, and given that ‘even a small ceremony requires considerable financial investment’ (Vignoli et al. 2016: 264), the consequences of the pandemic for marriage intentions are likely to outlast the emergency.

In any event, our study offers clear support for the thesis that uncertainty favours cohabitation, by employing the COVID-19 pandemic as an exogenous uncertainty multiplier. This is in line with a POD-friendly interpretation of family formation practices.

Disclosure statement

No potential conflict of interest was reported by the author(s).

Additional information

Funding

This work was supported by the Italian Ministry of University and Research under the FARE grant ‘Narratives’ (PI: Daniele Vignoli).

Notes on contributors

Raffaele Guetto

Raffaele Guetto, Ph.D. in Sociology and Social Research, is tenure-track Assistant Professor of Demography at the University of Florence (Italy), where he teaches ‘Demography’, ‘Social Demography’, and ‘Longitudinal Data Analysis’. He is a member, among others, of the International Sociological Association and of the Italian Association for Population Studies. He organized and has been part of the Scientific Committee of several international conferences. His research interests include social demography, social stratification and inequality, and immigrant integration. On these topics, he published several articles in international peer-reviewed journals with ISI-Impact Factor, such as the European Sociological Review, Acta Sociologica, European Journal of Population, Journal of Ethnic and Migration Studies, Demographic Research, International Migration Review.

Daniele Vignoli

Daniele Vignoli is Professor of Demography at the University of Florence, where he teaches ‘Demography’, Social Demography’, and ‘Longitudinal Data Analysis’. Currently, he leads an ERC Consolidator Grant on ‘Economic Uncertainty and Fertility in Europe’, a Miur-FARE Grant on ‘Uncertainty and Fertility: Narratives’, and a Miur-Prin Grant on ‘The Great Demographic Recession’. He is Editor-in-Chief of GENUS – Journal of Population Sciences and co-director of FloPS (Florence Population Studies). He is an elected member of the steering committee of the Italian Statistical Society. His research address: social demography; family demography; life course.

Giacomo Bazzani

Giacomo Bazzani, Ph.D., is Research Fellow at University of Florence (Italy), research project ‘Economic Uncertainty and Fertility in Europe’, research grant funded by European Research Council (ERC), P.I. Daniele Vignoli. He teaches ‘General Sociology’. He won the DAAD one-year research grant and was visiting researcher at the Max Planck Institute for the Study of Societies (Cologne) and Trier University in the academic year 2017-2018. He is the author of the book When Money Changes Society. The Case of Sardex Money as Community. Berlin: Springer VS. Recent publications include Vignoli, Daniele, G. Bazzani, R. Guetto, A. Minello, E. Pirani (2020) Uncertainty and Narratives of the Future. A Theoretical Framework for Contemporary Fertility. In Robert Schoen (eds) Analyzing Contemporary Fertility, London: Springer.

Notes

1 We included a check for the validity of the treatment. Respondents have been asked which type of scenario they were exposed to: the percentage of cases that could not recall the exact expected length of the pandemic included in the scenario amounts to 13%, but drops to approximately 5% if we exclude people who confused ‘more than two years’ with ‘two years’. After being exposed to the experiment, respondents were debriefed about the fictitious nature of the information about the evolution of the pandemic they received.

2 In the Appendix we report the full text of the treatment in English and Italian.

3 Full tables are available upon request.

4 The small number of cases does not allow for an analysis of possible heterogeneity of treatment effects using more detailed educational measures, or by combinations of sex and education.

5 A possible interpretation is that the more educated may have underreacted to the ‘sure predictions’ about the evolution of the pandemic because they took them with more reservation. But it is also true that the more educated hold higher levels of trust in science and institutions, which may increase their level of responsiveness to academics’ and experts’ opinions and recommendations.

Appendix

Wording of the treatment (English)

In the next screen we will provide you with up-to-date forecasts concerning the evolution of the Coronavirus pandemic.

Within the last few days there haven’t been substantial variations in the number of contagions, hospitalizations, and deaths. The task force composed by leading experts of the Coronavirus pandemic eventually obtained sure predictions about the future of the pandemic in Italy.

The experts predict that the Coronavirus pandemic emergency will last X before a return to normality.

(5 randomized scenarios for X: 3 months, 6 months, 12 months, 2 years, more than 2 years.)

Wording of the treatment (Italian)

Nella prossima schermata le forniremo le previsioni aggiornate sull'andamento della pandemia di Coronavirus.

Da alcuni giorni non ci sono variazioni rilevanti sul numero di contagi, dei ricoverati e dei decessi. La task force composta dai maggiori esperti sulla pandemia di Coronavirus è finalmente riuscita a ottenere previsioni sicure sul futuro della pandemia in Italia.

Gli esperti prevedono che l’emergenza da pandemia di Coronavirus durerà X prima di un ritorno alla normalità.

(5 scenari randomizzati (valore della X): 3 mesi, 6 mesi, 12 mesi, 2 anni, più di 2 anni.)