im Rahmen eines experimentellen Praktikums meines Psychologie-Studiums replizieren wir gerade eine Studie von Shafir et al. (1997) und nutzen zur Ergännzung eine Studie von Mees et al. (2014) zum Thema "Money Illusion". Kern ist dabei ein Fragenkomplex aus drei Fragen, von denen jede als UV einen anderen Kontext stellt, in denen Menschen als AV mit einer Bewertung in realen oder nominalen Werten reagieren.
Situation (nur aus Interesse):
Consider two individuals, Li Li and Wang Lan, who graduated from the same college a year apart. Upon graduation, both took similar jobs with publishing firms. Li Li started with a yearly salary of ¥120,000. During her first year on the job there was no inflation, and in her second year Li Li received a 2% (¥2400) rise in salary. Wang Lan also started with a yearly salary of ¥120,000.
During her first year on the job there was 4% inflation, and in her second year Wang Lan received a 5% (¥6000) rise in salary.
Ergebnisse (Werte in Klammer von Shafir et al., die anderen Mees et al.)
Economic terms (N = 137 (150)):
As they entered their second year on the job, who was doing better in economic terms?
Li Li: 82% (71%) Wang Lan: 18% (29%)
Happiness (N = 138 (69)):
As they entered their second year on the job, who do you think was happier?
Li Li: 39%+ (36%) Wang Lan: 61% (64%)
Job attractiveness (N = 134 (139)):
As they entered their second year on the job, each received a job offer from another firm. Who do you think was more likely to leave her present position for another job?
Li Li: 65% (65%) Wang Lan: 35% (35%)
Wir erheben unsere eigenen Daten in Deutschland und hätten diese mit einem 3x2-Felder-Chi-Quadrat-Test untersucht, ob die Geld-Illusion die Menschen auch hier beeinflußt. Mees et al. weisen aber noch andere Chi-Quadrat-Werte aus:
Economic terms
Shafir et al. : 58.07 (0.000)
Mees et al. : 10.67 (0.001)
Happiness
Shafir et al. : 221.6 (0.000)
Mees et al. : 198.4 (0.000)
Job attractiveness
Shafir et al. : 462.9 (0.000)
Mees et al. : 462.9 (0.000)
Voraussetzungen 1: We use the non-parametric Pearson Chi-square tests with exacttwo-sided p-values (as computed using IBM SPSS Statistics soft-ware, version 20). The tests are based on the actual frequencies (andnot on the fractions), although we have transformed the MF-dataset into n*datasets with the same sample size as the SDT-dataset. Weassume 10% noise in the data, because respondents may just havebeen careless in responding or mistaken rather than being proneto money illusion. A higher Chi-square score indicates in principlemore money illusion, given a certain sample size.
Voraussetzungen 2: Chi-square score and p-values for the test whether the frequencies in the Shafir et al. sample and Mees et al. sample differ from the expected frequencies under the null hypothesis of no-money illusion. A higher Chi-square score indicates stronger money illusion. The scores in columns I and II can be compared, but the scoresfor the various (sub)problems cannot be compared, because the sample size is different for each (sub)problem.
Frage: Wenn Mees et al. die Daten aus jeder Frage einzeln mit einer Allgemeinheit vergleichen, die nicht der Geld-Illusion erliegt, würde das doch über einen Vierfelder-Chi-Quadrat-Test geschehen, bei dem ein Feld dann zwangsläufig mit 0 (Menschen rechen 100% real, 0% nominal) belegt ist und die Voraussetzungen für den Test somit nicht erfüllt werden. Wie kommen die auf diese Chi-Quadrat-Werte?
Vielen Dank schon mal
