Belgian lottery players & the number 19: Another victim of the COVID-19 pandemic?

Number 19: Another Victim of the COVID-19 Pandemic? Patrick Roger, Catherine D’Hondt, Daria Plotkina & Arvid Hoffmann. Journal of Gambling Studies, Juil 19 2022. https://rd.springer.com/article/10.1007/s10899-022-10145-3

Abstract: Conscious selection is the mental process by which lottery players select numbers nonrandomly. In this paper, we show that the number 19, which has been heard, read, seen, and googled countless times since March 2020, has become significantly less popular among Belgian lottery players after the World Health Organization named the disease caused by the coronavirus SARS-CoV-2 “COVID-19”. We argue that the reduced popularity of the number 19 is due to its negative association with the COVID-19 pandemic. Our study triangulates evidence from field data from the Belgian National Lottery and survey data from a nationally representative sample of 500 Belgian individuals. The field data indicate that the number 19 has been played significantly less frequently since March 2020. However, a potential limitation of the field data is that an unknown proportion of players selects numbers randomly through the “Quick Pick” computer system. The survey data do not suffer from this limitation and reinforce our previous findings by showing that priming an increase in the salience of COVID-19 prior to the players’ selection of lottery numbers reduces their preference for the number 19. The effect of priming is concentrated amongst those with high superstitious beliefs, further supporting our explanation for the reduced popularity of the number 19 during the COVID-19 pandemic.

Conclusion

The number 19 has been heard/read/seen/googled countless times since the beginning of the COVID-19 pandemic in March 2020. This study used Belgium as the testing ground for a natural experiment investigating whether the global pandemic has influenced the popularity of the number 19 among lottery players. To obtain a comprehensive understanding of the impact of the COVID-19 pandemic on the popularity of the number 19 in the context of lotteries, we triangulated evidence from field and survey data. Specifically, using Belgian National Lottery data, we analyzed a sample of 836 draws over the period March 2017 to February 2021. This sample included data on both Euromillions and the Lotto and allowed us to define the first three years as the benchmark period and the last year as the COVID period. To accurately measure the popularity of a given number in the lottery data, we built a popularity index that was inferred from the actual proportion of winners among a subset of ranks. The main advantage of this methodology was that it enabled us to identify any potential shift in conscious selection between the two periods.

Relying on nonparametric permutation tests, our univariate results led us to reject the null hypothesis of unchanged popularity of the number 19 between the two periods. Specifically, we found that 19 is the only number with a significant decline in popularity during the COVID period. Next, we ran regression models to control for potentially confounding effects that could impact the proportion of winners at a given draw. These multivariate findings confirmed a significant decline in the popularity of the number 19 since the start of the COVID-19 pandemic.

To complement our results from the field study that used data obtained from the Belgian lottery games, we performed a survey study in the same country. The advantage of this approach was twofold. First, the survey study allowed us to vary the extent to which participants were exposed to information about COVID-19. Second, the survey study allowed us to measure the strength of participants’ superstitious beliefs, which is recognized as an important factor explaining lottery players’ choice of numbers (e.g., Simon (1998), Farrell et al. (2000), Wang et al. (2016), Polin et al. (2021)). The survey data thus allowed us to check for an interaction effect between the strength of players’ superstitious beliefs and the extent to which our experimental manipulation affected their number preferences.

Using ordinal logistic regression to determine the extent to which priming with the salience of COVID-19 impacts the survey participants’ selection of the number 19 in the two consecutive Lotto tickets, we found evidence that COVID-19 priming significantly reduces the probability that players will select the number 19, which fully supports the results from the field study. Importantly, the effect of COVID-19 priming is still significant when controlling for participants’ socio-demographic variables. Of particular interest, the findings reveal that the effect of COVID-19 priming is moderated by participants’ superstitious beliefs, being concentrated amongst individuals with high levels of superstitious beliefs.

Overall, although there could be unobserved variables affecting the results, the results of both studies provide converging evidence that is consistent with our theoretical explanation that an increased salience of the number 19 due to the COVID-19 pandemic is associated with an increase in the availability of negative feelings or memories associated with this number, which is associated with a decline in the popularity of the number 19 in games of chance. The strong consistency between both studies is also evident from the substantial correlation between the overall preferences for specific numbers in the survey data and the field data. Furthermore, building upon the findings from the survey study showing that the difference in choice frequency for the number 19 between primed and unprimed players is mainly driven by the choices of superstitious players, we hypothesize that the actual decline in the popularity of the number 19 during the Covid period as observed in the field study is likewise most likely driven by the preferences of superstitious people, who are more likely to select numbers themselves when playing Euromillions/Lotto (as opposed to using the Quick Pick system to randomly select numbers for them). Finally, based on a textual analysis of the responses provided by the survey participants regarding their associations with the number 19 which only incidentally mentioned COVID-19, we conclude that the COVID-19 pandemic might mainly generate a subconscious aversion to the number 19, which could lead people to select it less frequently when playing games of chance.

Our results have various implications for public policy, raising several questions. First, the existence of conscious selection in lottery games indicates that people can take decisions that are not based on economic objectives. In particular, playing popular numbers leads to paying an overvalued effective price.Footnote20 However, the distribution of numbers actually chosen by lottery players remains unknown to the general public. Researchers have used various methods to estimate this distribution (e.g., Farrell et al. 2000; Roger and Broihanne 2007) or obtained real data for some draws (e.g., Simon 1998; Polin et al. 2021). At a time where open data policies develop in various countries,Footnote21 an important policy question is transparency around games of chance. Should players be able to know, after each draw, the aggregate distribution of choices (like Fig. 7)? Should players have free access to the entire set of combinations chosen by players at the preceding draw (like Fig. 8)? The well-being of players who bet unconsciously on popular numbers (for reasons that are not linked to religious considerations, superstitious beliefs, etc.) would be enhanced by such measures if they change their number selection based on this information.

Second, our results raise another, and even more important, question linked to education about probability theory. Economic decisions under risk and uncertainty require estimations of probability distributions. For at least four decades, it is well-known that people have a distorted understanding of probabilities (Kahneman and Tversky, 1979). This distortion seems to be related to intelligence/cognitive ability (Choi et al., 2022). Moreover, since probability distortion is already present at a young age (Steelandt et al., 2013), learning elementary probability theory through games of chance in the first years of school might be beneficial for improving people’s future decision processes. In extreme cases, understanding what a uniform distribution is and what independent events are could potentially save lives. For example, in 2005, the number 53 did not show up in the Italian lottery for almost two years. As a result, some people had completely unreasonable reactions about this number, falling prey to the gambler’s fallacy (e.g., Suetens et al. 2016). In an article published just after the number 53 was finally drawn,Footnote22 newspaper The Guardian reported that “Four died in 53-related incidents. A woman drowned herself in the sea off Tuscany leaving a note admitting that she had spent her family’s savings on the number. A man from Signa near Florence shot his wife and son before killing himself. A man was arrested in Sicily this week for beating his wife out of frustration at debts incurred by his 53 habit.”

Another insightful illustration of the need for a better understanding of probability theory in the general population is the case of saliva Covid tests for children. The analysis of a number of different tests by Kivelä et al. (2021) shows that the average specificity of these tests is 99%. In France, these tests were used on a systematic basis in schools when the prevalence of the disease was close to 0.5% in children. Bayes’ theorem tells us that getting a positive test means a probability of being really positive of only one third.Footnote23 In other words, two thirds of positive tests were false positive that may have contributed to  potentially unjustified school closures.

To a broader extent, illustrating conscious selection for children would be easy and might also make them develop their critical thinking skills, a crucial skill for real-problem solving that is not included in standard IQ tests (Halpern and Dunn, 2021). For example, following Kahneman and Tversky (1974), a classroom could be divided in two groups, each group being exposed to a draw of a wheel of fortune (numbered from 1 to 10). Then, the two groups are asked a given question (e.g., estimating the weight of a pet shown on a picture). The comparison of the resulting estimates in the two groups would show that these estimates are related to the number drawn on the wheel of fortune, which is completely irrelevant information to the task at hand.

Our results add to the literature on number preferences and on the interaction between emotions, preferences, and decision-making. One limitation of this study was that it does not provide insights into whether the observed change in the popularity of the number 19 is permanent or transitory. There are several arguments supporting the hypothesis that the effect may be transitory. First, the COVID-19 pandemic should eventually come to an end and/or be replaced by another worldwide topic of concern. Chun and Turk-Browne (2007) show that the limited capacity of human memory indicates that attention determines what is encoded by memory. Moreover, Roy et al. (2005) and Kress and Aue (2017) report that biased memories are the result of an optimism bias, making it likely that the negative association between the number 19 and bad memories will weaken over time. At the time of this writing (September 2021), there is also anecdotal evidence that the media and people in everyday life have begun to shorten “COVID-19” to “COVID”. In the future, this change in terminology in the popular press could lead to a (partial) disconnect in the public’s mind between the number 19 and any bad memories or feelings associated with COVID-19. Google searches for the number 19 already provide tangible evidence of such an evolution (https://trends.google.com). For example, Google searches for the number 19 halved from May to October 2021, after a sharp increase in March 2020 (followed by a decrease until June 2020 and a period of stability for almost one year, until April 2021).

This study paves the way for further research on the impact of the COVID-19 pandemic on number preferences, emotions, and associated decision-making. First, future research should address the aforementioned question whether the decline in the popularity of the number 19 is temporary or permanent. Second, it would also be of interest to investigate whether other games of chance (e.g., sport betting) or other types of decision-making involving a selection of numbers have been similarly affected by the COVID-19 pandemic. Finally, research using data from other countries that were impacted by the COVID pandemic in different respects than Belgium (e.g., because they are more geographically isolated or have less developed health systems) would provide additional insights into the interaction between emotions and decision-making.