The effects of subjective loss of control on risk-taking behavior: the mediating role of anger

Abstract

Based on the Appraisal Tendency Framework on the antecedents and consequences of emotions two experimental studies examined the relationship between externally caused loss of control experiences and risk-taking behavior, as well as the assumed mediation of this relationship by the emotion anger. An experimental paradigm for inducing externally caused and consequently externally attributed loss of control which should lead to experiences of anger was developed and pretested in a Pilot Study. The relationship between loss of control experiences, anger, and risk-taking behavior was investigated using two separate student samples from Germany (N = 84, 54% female) and China (N = 125; 64% female). In line with our hypotheses, results showed that anger mediated the link between subjective loss of control experiences and increasing risk-taking behavior. Multiple group comparisons revealing similar patterns in both samples affirmed the results' cross-cultural generalizability. These results implicate that anger makes people less risk averse in the process of economic decision making.

Keywords: anger; attribution; cross-cultural study; perceived control; risk-taking.

FIGURE 1

One-factor pre–post design of Pilot Experimental Study and Main Experimental Studies 1 and 2. The Pilot Experimental Study tested the experimental paradigm’s adequacy for inducing subjective loss of control. By decreasing the accuracy with which the participants’ prediction was implemented into the computer game, the experimental group’s (EG’s) objectively given control was increasingly reduced during the four rounds of the manipulation section (represented by the darkening grey color). In a one-factor pre–post design the EG and control (CG) group’s subjective control ratings and external attributions following the baseline (t1) and manipulation (t2) sections of the experiment were compared. Main Experimental Studies 1 and 2 applied the same experimental paradigm to investigate the effects of loss of control on anger and risk-taking behavior in Germany and China, respectively.

FIGURE 2

A sample schematic representation of the computer-based problem-solving task. Participants are asked to continuously indicate the assumed next position of the little white circle while its previous positions fade from dark gray to lighter shades of gray.

FIGURE 3

A sample schematic representation of the experimental decreasing accuracy manipulation of displaying the participants’ prediction of the next position. The range of the intervals in which the clicking position was displayed was ±3.5° centered around the actual chosen position (red arrow) in round 5, ±10° in round 6, ±30° in round 7, and ±90° in round 8.

FIGURE 4

Subjective control ratings of the experimental and CG following the baseline (t1) and the manipulation (t2) part of the Pilot Experimental Study. Error bars represent SEM (±1 SE).

FIGURE 5

External attributions of the EG and CG following the baseline (t1) and the manipulation (t2) part of the Pilot Experimental Study. Error bars represent SEM (±1 SE).

FIGURE 6

Internal attributions of the EG and CG following the baseline (t1) and the manipulation (t2) part of the Pilot Experimental Study. Error bars represent SEM (±1 SE).

FIGURE 7

Anger ratings of the EG and CG following the baseline (t1) and the manipulation (t2) part of the Main Experimental Study 1. Error bars represent SEM (±1 SE).

FIGURE 8

Risk-taking behavior of the EG and CG represented by the mean proportion of chosen sectors during the devil’s task following the baseline (t1) and the manipulation (t2) part of the Main Experimental Study 1. Error bars represent SEM (±1 SE).

FIGURE 9

Structural equation model for testing the indirect effect (Hypothesis 3) of Main Experimental Study 1 on a latent variable level. The coefficients represent standardized factor loadings and standardized regression paths in the German sample (N = 84). ^∗p < 0.05, ^∗∗p < 0.01, ^∗∗∗p < 0.001.

FIGURE 10

Structural equation model for testing the indirect effect on a latent variable level in Main Experimental Study 2. This model also served as baseline model (Model 1) for testing the cross-cultural generalizability (Hypothesis 4). The coefficients represent standardized factor loadings and regression paths of the baseline model 1 in the Chinese sample (N = 125). ^∗p < 0.05, ^∗∗p < 0.01.