1a). Of all submitted bids players bid zero points on M = 14.4, 95% CI [8%; 21%] of all trials. Surprisingly, players reduced their bids over the course of auctions in the PV± and PV+ conditions measured as the difference between the mean first five bids and the mean last five bids ( Fig. 1b and Table 1). Wide confidence intervals of effect estimates ( Table 1) indicate that the strength of reduction was not consistent across players. Indeed, these differences were, at least partly, driven by the initial difference between the bids of

the two players in the PV± and PV+ condition ( Fig. 2). Players adjusted their bids in the direction of the bids of the other player, with stronger adjustments for the player initially bidding more

(slope estimate for interactions <0.5 in Table 2). This resulted in 85% of the participants bidding initially more in the PV+ selleck inhibitor condition also winning the majority of the auctions. In the PV± condition only 52% of the players that initially bid more also won more than half of the auctions. To examine the effects of underlying dynamics on a trial-to-trial basis, we Galunisertib price focused our analysis on the effect of the two previous auction outcomes on player’s propensity to increase or decrease their bids. Player bids show a consistent pattern across all preference levels where players increased their bids when losing and decreased their bids when winning (Table 3). The positive effect on bids was slightly larger when players

first won and then lost with regard to auctions with one particular item. As final player bids did not reflect the preference for an item, we analyzed pre- and post-auction preference statements for the five auction items. A considerable number of players (66.6%) changed their preference ranking. Our main goal was to identify factors from the auction that influence player preference changes, an index for private value change. We Rucaparib research buy found that the initial difference between player bids and the evolution of bids for a particular item affected bid dynamics (see Results on dynamics during the auction). Two additional factors entered the analysis as measures for the degree of competition: sunk costs, i.e. amount points lost in auctions, and the number of wins minus the number of losses. Based on these factors, we constructed a multinomial model where we contrasted auctions with increasing and decreasing preference with auctions without a change. Two patterns emerge from this analysis. First, some model coefficient estimates for increasing and decreasing preference point in the same direction (same sign) with approximately same effect size (Fig. 3 and Table S1). This indicates that these factors influence the probability to change preference in general, i.e. not restricted to increasing or decreasing changes. The most noteworthy of these factors was the difference between the two initial bids between the two players of a pair (ID).