An online experiment on the role of psychological ownership in long-term project commitment

This study is an online (laboratory) experiment with 4 experimental groups. In the experiment, all participants generate a donation to a charity. The different experimental groups are created based on a 2x2 treatment design: the first dimension varies whether participants have decision power over which charity receives their donation or if the charity is randomly assigned to them. The second treatment dimension varies whether participants receive the money to be donated as windfall gain or have to exert effort to earn it.

2021-02-15

2021-02-28

- Willingness to pay (in terms of number of summation tasks willing to commit to)
- Preference for personally contributing to project maintenance (instead of delegating)

- The willingness to pay is elicited by a Becker-DeGroot-Marschak mechanism and consists of a number between 0 and 30. Part of the donation is at risk of being destroyed unless the participant commits to solve some summation tasks, so the participant needs to state the number of tasks they are at maximum willing to commit to to save the donation.
- The preference for personally contributing to project maintenance (instead of delegating) is a number between 0 and 35 indicating the number of tasks the participant would rather do themselves instead of (costlessly) delegating them, given they are in a situation where 35 tasks have to be completed for sure to save the donation and they need to decide how many to do themselves and how many to delegate.

Psychological ownership score

Psychological ownership is measured on a 4-item scale, each item being measured on a Likert scale from 1 to 7, with one item being scored in reverse (i.e. the score assigned is 8 minus the score selected); the score is constructed as the simple average of all of the 4-point scores and is taken from prior psychology literature.

All participants generate a money amount that is to be donated to a charity. Based on the treatment group they are in, they either get to decide which charity the donation goes to or not and they either had to work for it or not. In each case, the same amount of money is generated. Subsequently, part of the donation is at risk of being destroyed. In order to prevent this from happening, a certain number of real effort tasks need to be solved. Participants need to make 2 decisions in a strategy-method approach: one, they need to state the maximum number of tasks they are willing to solve to prevent the destruction from occurring. Second, in a scenario where the number of tasks is fixed, they need to state how many tasks they want to do themselves and how many they want to delegate.

All participants generate a money amount that is to be donated to a charity. Three charities are available and the participants all receive the same information about their work. Based on the treatment group they are in, they either get to decide which of the three charity the donation goes to or they are randomly allocated to one of the charities (though they get their preferred charity with 90% probability). Moreover, they either had to work for the money amount or they received it as windfall gain. In each case, the same amount of money is generated by each participant. Subsequently, part of the donation is at risk of being destroyed. In order to prevent this from happening, a certain number of real effort tasks need to be solved. Participants need to make 2 decisions in a strategy-method approach: one, they need to state the maximum number of tasks they are willing to solve to prevent the destruction from occurring. This is elicited using a Becker-DeGroot-Marschak mechanism, where participants state the maximum number of tasks they are willing to complete from 0 to 30. A random draw determines the required number of effort tasks. If the stated WTP is equal to or higher, they save their donation by completing the randomly determined number of tasks. Otherwise, they do not perform any tasks and the donation is destroyed. In a second scenario, the number of tasks is fixed at 35 and they can decide how many of these tasks to do themselves and how many to delegate to a third party. The order of these two scenarios is randomized to avoid any internal consistency or anchoring concerns. The hypothesis is that participants will be willing to commit to more tasks by themselves and also prefer to complete more tasks themselves rather than to (costlessly) delegate if they had decision-power and have invested effort.

Randomization is performed on the individual level by assigning the treatment group by order of arrival of the participant. So the first participant to arrive is assigned to group 1, the second to group 2 and so on and starting again with group 1 after the fourth person arrived and was assigned to group 4 (out of 4). This is done in order to ensure balanced group sizes. The order of arrival is considered to be exogenous, first because participants are assigned to groups on a rolling basis, so it is not the case that early arrivals all end up in the same group. Second, participants arrive in a waiting room first, before the experiment is started. The order in which the link is then opened thus also largely depends on internet speed and how quickly they click "start", once the link is activated. Moreover, the treatment groups without decision power are oversampled to allow for the fact that only 90% of participants in those groups are part of the main sample. This is achieved by sorting - for every 18 participants who log in - 4 to each treatment with decision power and 5 to each treatment without decision power.

individual

No

450 individuals

450 individuals

100 in arm 1 (decision power and effort invested)
100 in arm 2 (decision power and no effort invested)
125 in arm 3 (no decision power and effort invested)
125 in arm 4 (no decision power and no effort invested)

The minimum detectable effect size (with 80% power and α= 0.05) for the willingness to pay will be 2.16 tasks in relation to a baseline average of 16.5 tasks with a standard deviation of 5.43 (so an increase in WTP by around 13% or 0.4 standard deviations can be detected). For the number of tasks allocated to oneself, the minimum detectable effect size will be 2.41 tasks from a baseline of 19.5 tasks on average with a standard deviation of 6.05 (so an increase in tasks allocated to oneself of around 12% or 0.4 standard deviations can be detected).