#104347 | AsPredicted

'To be the partner in crime: Unethical loyalty in an online setting'
(AsPredicted #104347)


Author(s)
This pre-registration is currently anonymous to enable blind peer-review.
It has 3 authors.
Pre-registered on
08/09/2022 08:00 AM (PT)

1) Have any data been collected for this study already?
No, no data have been collected for this study yet.

2) What's the main question being asked or hypothesis being tested in this study?
When witnessing others' unethical behavior, individuals can behave in two ways: covering up the transgressor's behavior and possibly profiting from it personally, or blowing the whistle and revealing the transgressor's dishonesty to others. In this study, we will examine individuals' willingness to cover up others' unethical actions, that is, their tendency to engage in unethical loyalty.
In a recent laboratory study (Thielmann et al., 2021), a high prevalence of unethical loyalty was found. Our study's first goal is to investigate whether individuals are equally willing to engage in unethical loyalty in an online setting.
In Thielmann et al.'s study, unethical loyalty was positively associated with the money offered for unethical loyalty (i.e., the "bribe") and negatively associated with individuals' Honesty-Humility levels. The current study aims to replicate these findings in an online setting. We thus hypothesize that:
1) Unethical loyalty increases with increasing bribes.
2) Unethical loyalty decreases with increasing levels in Honesty-Humility.
Moreover, we will examine the role of perceived justifiability for unethical loyalty. Specifically, we hypothesize that:
3) The negative relation between Honesty-Humility and unethical loyalty is mediated by the perceived justifiability of unethical loyalty. By implication, we also expect a positive relation between the perceived justifiability of unethical loyalty and engaging in unethical loyalty and a negative relation between Honesty-Humility and the perceived justifiability of unethical loyalty.
Finally, we consider two specific reasons for perceiving unethical loyalty justifiable - self-interest and prosociality. We hypothesize:
4) Honesty-Humility is negatively related to perceiving unethical loyalty justifiable because it serves one's own self-interest (4a) and positively related to perceiving unethical loyalty justifiable because it is prosocial (4b).

3) Describe the key dependent variable(s) specifying how they will be measured.
The key dependent variable is unethical loyalty which will be measured using an online version of the Unethical Loyalty Game (Thielmann et al., 2021). A player (Player 2; P2) is defined as unethically loyal if they claim that their interaction partner's (Player 1's; P1's) response in a previous die-rolling task is factually true even though it is not.
Given that we are mainly interested in P2's behavior in reaction to a dishonest P1, we will present the P2s with the decisions of a subset of P1s who lied (no deception will be involved).

Our key independent variables are (a) Honesty-Humility as measured via the 60-item HEXACO Personality Inventory-Revised (HEXACO-60), (2) the bribe offered by P1 (i.e., the relative proportion of the payoff P1 received), (3) the perceived justifiability of unethical loyalty (justifiable yes/no), and (4) the extent to which unethical loyalty is perceived as justifiable due to self-interest and prosociality.

4) How many and which conditions will participants be assigned to?
Participants will be randomly assigned to act as either P1 or P2 (see 3). The design is sequential, that is, the data of P1s will be collected first (T1) in order to present their responses to the P2s at a later point in time (T2). All P2s will be matched with a dishonest P1.

5) Specify exactly which analyses you will conduct to examine the main question/hypothesis.
Research question (prevalence of unethical loyalty):
Descriptive analysis of the observed prevalence; exact binomial test (two-tailed) comparing the current study's prevalence of unethical loyalty with the prevalence of unethical loyalty found by Thielmann et al. (2021) (i.e., 85.8%).

Hypothesis tests:
All non-binary variables (i.e., the bribe, Honesty-Humility, perceived justifiability due to self-interest / prosociality) will be z-standardized using the sample mean and SD.
1) Logistic regression predicting unethical loyalty by the bribe offered (one-tailed test)
2) Logistic regression predicting unethical loyalty by Honesty-Humility (one-tailed test)
3) Single mediator model where the relation between Honesty-Humility and unethical loyalty is mediated by the justifiability judgment (justifiable yes/no). The indirect effect will be tested using the effect size Lachowitz υ, with Monte Carlo confidence intervals
4) Correlation analyses in the subsample of P2 perceiving unethical loyalty as justifiable: Relation between Honesty-Humility and perceived justifiability for the reason of self-interest (4a); relation between Honesty-Humility and perceived justifiability for the reason of prosociality (4b) (both one-tailed)

6) Describe exactly how outliers will be defined and handled, and your precise rule(s) for excluding observations.
All participants completing the study will be considered for inclusion in the analyses. However, we will exclude participants whose responses indicate careless responding. Specifically, P2s will be excluded if they (a) fail the instructed attention check embedded in the HEXACO-60, (b) take less than 2 seconds on average per item of the HEXACO-60, (c) show very low variation (i.e., SD < 0.3) in responses on the HEXACO-60, or (d) provide a wrong answer to the comprehension check for the Unethical Loyalty Game more than three times.

7) How many observations will be collected or what will determine sample size?
No need to justify decision, but be precise about exactly how the number will be determined.

Using G*Power (for analyses 1, 2, 4; Faul et al., 2009) and the R shiny app Mc Power Med (for analysis 3; Schoemann et al., 2017), we ran a priori power analyses (α = .05, 1 – β = .80) for the main hypothesis tests.
1) To detect an OR = 1.6 (small to medium effect) for the effect of the bribe predicting unethical loyalty in a logistic regression (one-tailed test) and expecting a prevalence of 85% for unethical loyalty (based on the findings by Thielmann et al., 2021), the power analysis yielded a required sample size of N_P2 = 222.
2) To detect an OR = 0.6 (small to medium effect) for the effect of Honesty-Humility predicting unethical loyalty in a logistic regression (one-tailed test) and expecting a prevalence of 85% for unethical loyalty (based on the findings by Thielmann et al., 2021), the power analysis yielded a required sample size of N_P2 = 189.
3) The Monte Carlo power analysis for indirect effects with 10,000 replications and 20,000 Monte Carlo draws per replication, 95% CI, assuming the following correlations and standard deviations (r_HH-J = -.20, r_J-UL = .20, r_HH-UL = -.20, SD_HH = 1.00, SD_UL = 0.03, SD_J = 0.03) yielded a required sample size of N_P2 = 333. The assumed correlations are based on the findings by Thielmann et al. (2021) and evidence on the link between justifiability of unethical behavior and the Dark Factor of Personality, which has strong conceptual overlap with Honesty-Humility (Hilbig et al., 2022), while we used more conservative estimates here (e.g., adjusting r = .22 downward to r = .20).
4) To detect a small to medium-sized correlation (r = |.20|) between Honesty-Humility and the perception of justifiability due to self-interest/prosociality in a one-sided test, the power analysis yielded a required sample size of N_unethically loyal P2 = 150. Assuming at least 70% of P2s to perceive unethical loyalty justifiable, the required sample size is N_P2 = 215. The assumed proportion is based on the proportion of unethical loyalty found by Thielmann et al. (2021), expecting a large proportion of unethically loyal individuals to judge unethical loyalty justifiable but using a more conservative estimate (the larger the proportion, the smaller the sample size needed).
Based on these results, we based our final sample size on the a priori power analysis for hypothesis 3 (mediation effect), which yielded the largest required sample size among all tests (i.e., N = 333). Anticipating a certain dropout due to the exclusion criteria specified above, we aim to collect data from N = 350 P2.
We will recruit a smaller sample of P1 (i.e., N = 130) than P2 and randomly match the data of one lying P1 to several P2s (assuming the dishonesty rate among P1s similar to the one found by Thielmann et al. (i.e., 70%), one lying P1's data will be randomly matched to 3-5 P2s, depending on the actual number of dishonest P1).

8) Anything else you would like to pre-register?
(e.g., secondary analyses, variables collected for exploratory purposes, unusual analyses planned?)

Exploratorily, we will investigate the interaction between Honesty-Humility and bribe size in predicting unethical loyalty. The interaction may allow taking a closer look at the decision to behave unethically loyal: When deciding to cover up someone else's transgression, the motives at play may include self-interest and prosociality. Unethical loyalty is both prosocial because it increases P1's outcome and selfish because it increases P2's outcome. Specifically, with an increasing bribe offered by P1 to P2, unethical loyalty becomes less prosocial but more selfish. In turn, individuals low on Honesty-Humility should consider self-interest particularly important for their decision, whereas individuals high on Honesty-Humility should consider prosociality particularly important. Thus, the link between Honesty-Humility and unethical loyalty might be moderated by the (relative) bribe size: the larger the bribe (i.e., the more selfish unethical loyalty gets), the stronger the negative link should be.
In a reanalysis of the study by Thielmann et al. (2021), this interaction was apparent and statistically significant with a small to medium effect size.

Version of AsPredicted Questions: 2.00