Gaming or Gambling? On Selection Neglect and Loot Boxes
How to cite
Köster, Mats; Dertwinkel-Kalt, Markus; Cordes, Simon; Werner, Tobias. 2021. "Gaming or Gambling? On Selection Neglect and Loot Boxes." AEA RCT Registry. AEARCTR-0008021. https://doi.org/10.1257/rct.8021-2.0
Download: BibTeX · RIS — for reference managers without a browser translator. Zotero and most others pick up this page's citation metadata directly.
* Primary investigator
Abstract
Nowadays, many successful video games feature "loot boxes" that, just like gambling, offer random rewards to be used in-game. In 2020 alone such loot boxes generated $15 billion of worldwide revenue. There is an increasing public concern, however, that video game developers design loot boxes in a way to make gamers "overpay" for the typically small chance of getting the reward. We single out two features of loot boxes that might result in gamers overestimating the probability of winning the reward and, consequently, paying too much. First, developers typically do not disclose the odds, but provide gamers only with an interval-censored distribution of rewards. Second, in many games there is a public announcement whenever someone wins a reward, which results gamers observing a selected sample of the reward distribution. In a controlled laboratory experiment, we systematically study the effect of either feature, as well as combinations thereof, on the willigness-to-pay for (monetary) lotteries.
Intervention
We study the willingness-to-pay for (and beliefs about) monetary lotteries across four treatments that vary how much information subjects receive on the reward distribution.
Experimental design
In each treatment, subjects state their willingness-to-pay for 10 different monetary lotteries. All lotteries have the following structure: with probability 0.99 - p, a subject gets an amount of x > 0; with probability 0.01, a subject gets an amount of y > x; and with probability p, a subject gets nothing. In each decision, subjects learn the feasible outcomes, x, y and 0, but we vary across four treatments how much information they receive on the corresponding probabilities.
Treatment "FULL:" subjects observe the full reward distribution; that is, they learn the probabilities of receiving of the three possible outcomes.
Treatment "CENSORED:" subjects only learn the probability, 1 - p, of receiving at least an amount of x.
Treatment "FULL-BEST:" subjects observe the full reward distribution and, in addition, the five highest outcomes in a random sample of 250 draws (from the underlying distribution).
Treatment "CENSORED-BEST:" subjects learn the probability, 1 - p, of receiving at least an amount of x, and, in addition, they observe the five highest outcomes in a random sample of 250 draws (from the underlying distribution).
In CENSORED and CENSORED-BEST, we further ask subjects to state their belief about the probability with which the first lottery that they had to price during the experiment pays the maximum amount.
Design details
Cited by
No citation data available for this registration yet.