Does work effort for public versus private organizations differ? Evidence from an online work task experiment

Different forms of work task description in an experiment online.

Our main intervention is providing participants with different forms of an ownership cue (public versus private, for-profit) of the hypothetical organization they are asked to work for in a work task.The study instrument is a work task in which respondents are asked to check (and if necessary) correct numerical entries on 6 hypothetical time sheets. In both forms of ownership cue intervention, the participants undertake the following: Typing in the numbers of the times sheets (which is based on a screenshot of handwritten sheets) into a data entry form is the first part. As a second step, they need to check whether calculations displayed on a respective time sheet are accurate and correct them if necessary. Respondent performance in terms of correctly entering numbers, correcting mistakes and the time it take them doing so as a measure of their work performance, which we hypothesize to vary for public vs for-profit organizations. Work task instruments of this kind or commonly employed in applied psychology, behavioral economics or in public management.

2018-10-05

2018-11-30

Measure of quality of work task (please see below explanation)

Primary outcome measure is the number of mistakes made by participants in the work task overall. We expect that this outcome will be affected by the ownership status of the organization people are described as working for.

Secondary outcomes are the time take to complete the work task, the ratio between mistakes and time, and whether they agree to do an additional 7th worksheet (after completing the 6 worksheets), which ought to measure their prosocial work behavior (because they agree to do an additional worksheet for the organization), assuming that this would be higher in public versus for-profit organizations. Measures of this kind are widely used in the literature.

These additional outcomes are further affected by the interventions.

US MTurkers recruited to the study and allocated to different experimental groups

Subjects who enroll as workers on Amazon Mechanical Prime will be recruited. Amazon does not allow any individuals below the age of 18 to work on the Mechanical Turk platform, hence we will only recruit subjects of at least 18 years of age. A total of 600 adult workers will be recruited through the platform. Recruitment will be restricted to MTurkers who are based in the US. This can be done using MTurk’s graphics interface so that the work announcement (also called a HIT [Human Intelligence Task]) will only be accessible for US MTurkers.
Workers receive financial compensation for participation. All US MTurkers are eligible, and the task will close as soon as 600 subjects have entered it. Inclusion criteria are that the target population are all US-based MTurkers. The recruitment of turkers is to work for something labelled as an elderly care home timesheet, with the different labels for the type of organization constituting the variation in intervention. Participants are asked if they utilized tools that might help them in the task (a calculator). Additional measures are nine items that tap the individual’s conscientiousness taken from the Big Five personality traits inventory. People are also asked about their gender, age and what type of organization they do work for (public, private or nonprofit). The final question is a manipulation check, asking participants if they remember whether the organization mentioned on the time sheets is a public or private, for-profit organization. At end of the study, part pants are debriefed and informed about the full purpose of the study.

Subjects will be randomized into one of the study's two arms using Qualtrics’ randomization procedures.

Units are individual workers.

No

600 participants who are MTurk workers, not clustered

600 MTurk workers are units that are observed

300 MTurk workers per treatment arm