USING GAMIFIED TASKS IN PSYCHOLOGICAL ASSESSMENT: AN ANALYSIS THROUGH NEURAL NETWORKS IN ADOLESCENT RISK TAKING

Gamification is a promising approach in assessment as game-based tools offer engaging experience for participants, short testing times, elicit low test-taking anxiety, are less prone to social desirability bias, and can be more predictive with respect to non-gamified measures. In this study, we compared the performance of a computerized task measuring risky decision making (Cups Task, Levin & Hart, 2003) - with respect to a gamified version of the same task - in predicting risk taking in the financial-gambling domain, measured with the DOSPERT scale (Barkley-Levenson et al., 2013). Both the tasks were composed by 54 trials, but the gamified task contained engaging audio and visual feedback and a character to drive. To analyze the two tasks, we used the Neural Networks (NNs), that can fit almost any nonlinear phenomenon and can identify patterns in subsets of the input variables. Participants were 195 adolescents (57% males, mean age = 17.13). Results showed that the gamified version of the task had a better performance in predicting financial-gambling risk-taking with respect to the original task. However, the percentage of cases correctly identified by the NN was relatively low even with the game. Indications for future research in psychological assessment are discussed.