'Spoiling the party. On the willingness to transmit inconvenient ethical info' (AsPredicted #111393)
Author(s) This pre-registration is currently anonymous to enable blind peer-review. It has 3 authors.
Pre-registered on 2022/11/01 - 07:37 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? Are people willing to send others 'inconvenient' information about the consequences of their actions. Do they withhold such information under social pressure?
3) Describe the key dependent variable(s) specifying how they will be measured. - We consider an interaction between a sender and a receiver. The receiver can take an action that is personally profitable, but leads to an unknown monetary impact for a charity (the Red Cross). The sender can send precise information about these consequences to the receiver at a small cost. Using the strategy method, each sender makes decisions to send (yes/no) for three negative impact levels (-2.5, -1.0 and -0.5 pounds), and one positive impact level (+0.5).
- Our first key interest is in the sender's decision to send information about negative consequences (i.e. we ignore sender's decisions for positive consequences). For each sender, we define a "sender-index" that measures how much information they send for different impact levels (0=no information sent for any impact level, 1=information sent only for the worst impact level, 2 =information sent for the worst two impact levels, 3=information sent for all three impact levels).
- Our second outcome variable is the payoff for the Red Cross resulting from the task.
4) How many and which conditions will participants be assigned to? We have 3 between-subject treatments:
1) Baseline: after sending, senders face no further interaction with the recipient
2) Request: before the sender's decision, receivers make a request to senders for information or non-information. These requests thus create two groups of senders, one with a request for information and one with a request for ignorance.
3) Request + Punishment: in addition to making the request, receivers can punish senders by reducing their payoffs by a small amount, after they learn whether information had been sent.
5) Specify exactly which analyses you will conduct to examine the main question/hypothesis. H1: People send inconvenient information about the charity to their partners.
We evaluate the sender-index in the Baseline treatment. We regress the sender-index using ordered probit, on a number of control variables that measure their preferences for information and for the charity, as well as demographics (gender, age, income, identification with the charity, time spent in experiment, number of attempts to get comprehension questions correct, reveal behavior in DWK task).
H2: The sender-index increases with request for information and decreases with request for ignorance.
We use a non-parametric Jonckheere trend test to test the alternative hypothesis that the sender-index follows: Request Info > Baseline > Request Ignorance. In addition, we use parametric ordered probit regressions to examine robustness of this effect, under several control variables (see above).
H3: The possibility of receiver punishment increases the impact of requests on the sender-index.
We use a non-parametric Wilcoxon rank-sum test to test the alternative hypothesis that the sender-index is: Request Info Punishment > Request info and Request Ignorance Punishment < Request Ignorance. In addition, we use parametric ordered logit regressions examine robustness of this effect under several control variables (see above).
H4: Request for information is associated with higher earnings for the charity, and request for ignorance with lower earnings for the charity. These effects are amplified in the punishment treatment.
We will use a chi-square test to compare the distribution of payoffs across all treatments, as well as split by receiver request (three and five groups, respectively). We will also do an OLS regression with the payoffs for the charity as a dependent variable and as explanatory variables a dummy equal to one when a request for information is made, a dummy equal to one in the punishment treatment, and their interaction.
6) Describe exactly how outliers will be defined and handled, and your precise rule(s) for excluding observations. We will exclude subjects who are not able to answer basic understanding questions. We will exclude interactions in which the sender-index cannot be computed because senders switch twice (e.g. no/yes/no or yes/no/yes for the three negative consequences).
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. We will recruit 750 senders (150 in Baseline, 300 in Request, and 300 in Request + Punishment) and 750 receivers. Sampling will stop when we have 750 pairs that completed the main task. We collect twice the amount of observation in the Request and Request + Punishment treatment to be able to analyze the behavior of senders conditional on the request for information or ignorance. The sample size allows us to detect an effect on the sender-index of size d = 0.30 based on one-sided Mann-Whitney Rank Sum Test (under the assumption of normality of the distributions, power = 0.8, and alpha = 0.05).
8) Anything else you would like to pre-register? (e.g., secondary analyses, variables collected for exploratory purposes, unusual analyses planned?) - We will do robustness checks of the results of the sender-index using parametric regressions with different specifications: OLS and Tobit.
- We will do robustness checks with an alternative sender-index, which is 1 if the sender sends information for the worst outcome, and zero otherwise.
- We will investigate the role of beliefs about receiver behavior in sender decisions. Specifically, we will elicit sender's beliefs about receivers behavior towards the Red Cross as well as their expectations of receivers to implement punishment in the Punishment treatments.
- For the receivers, we will correlate their request behavior with their willingness to reveal information in an individual decision making task with hidden information, as well as the identification with the charity, age, income and gender.
- We conducted a pilot session of the Request + Punishment treatment to test the software, the comprehension of the task, the size of the incentives, and to assess the proportion of people requesting information to determine the sample size.
