Posted in: Conference
Authors: Anna Blajer-Gołębiewska, Dagmara Wach, Maciej Kos
Conference: 10th Nordic Conference on Behavioral and Experimental Economics
Location, dates: Tampere, Finland, 25-26 September 2015
Organizer: University of Tampere; University of Turku; Hanken School of Economics
Introduction: There is a wide array of research on the impact of emotions on decision-making process (Loewenstein, 2000; Rottenstreich, Hsee, 2001; Godek & Murray, 2008; Schul & Mayo, 2003; Nisbett, Peng, Choi, & Norezayan, 2001; Trope & Liberman, 2000; Sloman, 1996; Epstein, 1994, 1991 and others). Success of such studies in psychology and behavioural economics often hinges upon effectively inducing the desired – emotional or analytical – processing style. The aim of this article is to present results of using one such manipulation applied in a large on-line experiment designed to study risk information behaviour. We put forward the hypothesis that a successful manipulation would be reflected in the AIC indicator (analytical/affective induction check).
One previously verified procedure for inducing either emotional or analytical decision-making style was used by Schul & Mayo (2003) who conducted two experiments in future – focus and in the past-focus conditions. Individuals in the future-focus condition were asked about their plans for the future. In the past-focus condition, they were asked about reasons for past action. This study revealed that orienting subjects toward the future reinforces rational mode of thinking, and orienting them toward the past induces the experiential mode of thinking. We have used this manipulation to test our hypothesis.
Methodology: We designed an on-line experiment, in which we primed subjects to use either an analytical or an affective processing style following the procedure developed by Schul & Mayo (2003) and Godek & Murray (2008). We asked subjects the same questions as the ones introduced in Godek & Murray’s experiment, i.e., we asked them to describe their past and future purchasing behaviours by writing five to seven sentences.
We checked whether the manipulation worked by asking:
“How would you describe your decision process when you are planning a purchase in the future?” (analytical treatment condition) / “How would you describe your decision process when you made a purchase in the past?” (affective treatment condition). “I would base my decision on: (How I would feel) 1 2 3 4 5 6 7 8 9 10 (What I would calculate).”
We stored subjects’ answers to this last question in a variable called AIC.
Findings: Exactly 700 subjects participated in the study. Three observations were excluded from our analysis because essays written by these subjects were too short. We modelled the impact of the manipulation on AIC (our outcome variable) using ordered probit regression and controlled for age, sex, income. Our model is statistically significant (Likelihood Ratio Chi-Square statistic = 32.46 [df=5]), p < 0.001). We found a statistically significant relation confirming our hypothesis, i.e., participants with the induced analytical decision making style were likely to achieve higher value in the AIC test. However the size of the effect is surprisingly small (emotional: Coef.= – 0.2133 [-0.367, -0.058], p = 0.007). Mean AIC score among subjects in the affective treatment condition was 6.558 (SD = 2.150) versus 6.967 (SD=1.875) in the analytical treatment condition with higher values corresponding to more analytical decision making.
Furthermore, we found sex to be significant predictor (Coef.= 0.299 [0.141, 0.457], p < 0.001). The income variable is statistically significant, however the size of the effect is very small. The result show that higher income increases the predicted probability of analytical processing style application (Coef.= 0.043 [0.004, 0.083], p < 0.029). Age, as well as economic education, have no impact on analytical or affective processing style.
Conclusion: Although the described procedure to induce analytical or affective processing produces a statistically significant effect, its size appears very small. Therefore it might be prudent to apply it with caution. Investigation of when and why this and similar manipulations work might be a valuable research path for other scholars to explore.
See: conference paper