Abstract
This study examined the relationships between employee perceptions of organizational justice and trust and employee antisocial organizational behavior and communication. Participants from organizations representing two geographic regions were surveyed regarding their most recent performance appraisal. Results indicated that perceptions of justice and trust negatively predicted indirect interpersonal aggression, hostility, obstructionism, and deception. When justice and trust were entered simultaneously into the regression model, perceptions of trust predicted hostility, while perceptions of distributive justice predicted deception. Post hoc analyses demonstrated that justice and trust interacted to predict antisocial responses, and trust mediated the relationships between justice and antisocial responses.
Notes
Note: All correlations are statistically significant at p < .001.
Note: Sex: male = 1, female = 2. Region: southwest = 1, mid-Atlantic = 2. Give evaluation: no = 1, yes = 2.
∗p < .10.
Note: Sex: male = 1, female = 2. Region: southwest = 1, mid-Atlantic = 2. Give evaluation: no = 1, yes = 2
∗p < .05; † p = .079.
Participants recruited from the mid-Atlantic were older [(M = 42.31, SD = 12.66 vs. M = 34.68, SD = 11.37), t (142) = − 3.18, p < .05] and earned more [(M = $37,900, SD = $25,300 vs. M = $29,000, SD = $19,400), t (142) = − 2.40, p < .05] than those from the Southwest; however, the mid-Atlantic participants perceived less distributive justice after their performance evaluations than did the Southwestern participants [(M = 6.18, SD = 1.11 vs. M = 5.71, SD = 1.40), t (142) = 2.26, p < .05]. The form of the evaluation also differed according to the participants' geographic region, F (1, 128) = 7.19, p < .05.
For the justice variables, tests of homogeneity (internal consistency) indicated that for distributive and procedural justice, no error calculated between observed and expected correlations was greater than sampling error. For interactional justice, only one error calculated between observed and expected correlations was greater than sampling error (error was .01 beyond sampling error). Tests of heterogeneity (parallelism) indicated that fewer than 3% of errors calculated between observed and expected correlations was greater than sampling error (on average, errors were .13 beyond sampling error).
Tests of homogeneity indicated that for deception and obstructionism no error calculated between observed and expected correlations was greater than sampling error. For indirect interpersonal aggressiveness, only one error calculated between observed and expected correlations was greater than sampling error (error was .01 beyond sampling error). For hostility, less than 10% of the errors calculated between observed and expected correlations was greater than sampling error (on average, errors were .06 beyond sampling error). Tests of heterogeneity indicated that less than 3% of errors calculated between observed and expected correlations was greater than sampling error (on average, errors were .13 beyond sampling error).
Perceptions of all three justice types differed by participant sex, with women perceiving higher levels of justice. Participants responsible for giving feedback perceived higher levels of distributive justice and managerial trust than did participants not responsible for giving feedback. Participant age was negatively correlated with hostility, obstruction, and deception. Participant salary was negatively correlated with perceptions of procedural justice. As mentioned in Note 1, participant perceptions of distributive justice also differed according to the region from which the participants were recruited.
Because the distributions for the predictor and criterion variables were quite skewed and exhibited non-normal kurtosis, the data were normalized using the logarithm transformation. The results for the hypotheses only became stronger after transforming the variables and the results for the post hoc analyses only changed slightly.
Results after transforming the variables are available from the first author.