Do not bet on your favourite football team: the influence of fan identity-based biases and sport context knowledge on game prediction accuracy

ABSTRACT

Research question: While the influence of sport context knowledge, such as home advantage and team ranking, on prediction accuracy has been discussed in the previous literature, the role of identity-based biases, such as fans’ level of team involvement and the selection of their favourite team, in betting behaviour remain unclear. The main purpose of this study is to develop an understanding of how sport fans’ biases and sport context knowledge influence the accuracy of sport game predictions.

Research methods: A smartphone application enabled us to collect real soccer game predictions and results. A total of 529 football fans participated in 53,943 predictions of 2353 professional football games within a mobile smartphone application. Chi-square tests and logistic regressions were used to analyse the data.

Results and findings: Chi-square test results indicate that individuals overestimate their favourite team to win, as well as they split their predictions into dichotomous outcomes by overestimating wins and losses and underestimating draws. Logistic regression analyses indicate that identity-based biases negatively influence prediction accuracy, whereas individuals’ sport context knowledge positively contributes to prediction accuracy.

Implications: The study contributes to our understanding of the Psychological Continuum Model, individual biases, social identity theory and the psychological concept of splitting. Findings have implications for organizations who need to understand fans’ sport gambling behaviour and sport fans who seek to optimize their game prediction accuracy to improve their bets and fantasy team selections.

KEYWORDS:

• Sports gambling
• social identity
• identity-based bias
• team involvement
• result prediction

Disclosure statement

No potential conflict of interest was reported by the authors.

Notes

1. Although participants’ gender, age, and league levels of predicted football matches were measured in the study, the variables were excluded from the analysis. Providing personal information during the on-boarding of the app was voluntary. Thus, less than 50% of completed the information for age, and less than 50% completed the information for gender. Additionally, international (country-level) games were included in the sample where comparison of league levels is unavailable. Consequently, including these control variables would have resulted in a strongly reduced sample size and we proceeded with the analysis of the whole sample without accounting for these control variables.

2. In the current logistic regression analysis, the previous year’s team rankings (Odds ratio = 1.781, p < .001) had a positive relationship with prediction accuracy. We tested an additional logistic regression analysis utilizing team rankings from the end of the season holding all other variables constant. The result of the new analysis were consistent with the current analysis regarding significance and direction of all predictors, with increased odds for team ranking (Odds ratio = 2.085, p < .001). However, we proceeded with using the ranking prior to the 2014–2015 season because these represent the information participants may have had at the time of predicting.

Additional information

Funding

The authors are grateful for financial support through the Young Scholar Award received from the Fox School of Business, Temple University.