How does the Organization of Crowdsourcing Affect Labor Supply, Productivity and Morals

Last registered on June 25, 2021

View Trial History

Pre-Trial

Trial Information

General Information

Title

How does the Organization of Crowdsourcing Affect Labor Supply, Productivity and Morals

RCT ID

AEARCTR-0006242

Initial registration date

June 23, 2021

Initial registration date is when the trial was registered.

It corresponds to when the registration was submitted to the Registry to be reviewed for publication.

First published

June 25, 2021, 1:40 PM EDT

First published corresponds to when the trial was first made public on the Registry after being reviewed.

Locations

Country

Germany

Region

Bremen

Primary Investigator

Name

Lars Hornuf

Affiliation

University of Bremen

Contact Primary Investigator

Other Primary Investigator(s)

PI Name

Kristina Klein

PI Affiliation

University of Bremen

Contact Investigator

Additional Trial Information

Status

In development

Start date

2021-06-28

End date

2022-01-01

Keywords

Behavior, Labor

Additional Keywords

Crowdsourcing, Crowdworking

JEL code(s)

J22, J24, D91

Secondary IDs

DFG HO 5296/3-1

Abstract

The organization of crowdsourcing comes in various forms. The most common organizational forms are crowd labor markets, crowd collaborative communities, and crowd contests. In this paper, we experimentally study labor supply, productivity, and morals under these three organizational forms of crowdsourcing.

External Link(s)

Registration Citation

Citation

Hornuf, Lars and Kristina Klein. 2021. "How does the Organization of Crowdsourcing Affect Labor Supply, Productivity and Morals." AEA RCT Registry. June 25. https://doi.org/10.1257/rct.6242-1.0

Sponsors & Partners

There is information in this trial unavailable to the public. Use the button below to request access.

Request Information

Experimental Details

Interventions

Intervention(s)

There are two interventions:

(1) Organizational form of crowdsourcing

(a) crowd labor market
(b) crowd collaborative community
(c) crowd contest

(2) Degree of overpay

(a) low
(b) high

Intervention Start Date

2021-06-28

Intervention End Date

2021-07-06

Primary Outcomes

Primary Outcomes (end points)

Labor supply
Labor productivity
Reaction to overpay

Primary Outcomes (explanation)

Labor supply is a positive decision by a participant to continue working after the focal crowdsourcing round. On the individual level, this is a dummy for each participant.

Labor productivity is the average time a participant needs to successfully solve a grid during a particular crowdsourcing round. On the individual level, this variable measures the time in seconds needed to successfully complete a grid.

Reaction to overpay is a positive decision by a participant to return the overpaid remuneration to the experimenter in a franked return envelope. On the individual level, this is a dummy for each subject.

Secondary Outcomes

Secondary Outcomes (end points)

We investigate the following second-order effects: the extent of overpay (low = EUR 1.00, high = EUR 5.00) and how it affects participants reaction to overpay.

Secondary Outcomes (explanation)

Experimental Design

We have designed a randomized control trial, in which participants will be randomly assigned to different treatment groups (crowd labor markets, crowd collaborative communities, and crowd contests) and extents of overpay. We use two controlled manipulations: organizational form of crowdsourcing and extent of overpay. The experiment is followed by a questionnaire in order to conduct balancing checks.

Experimental Design Details

For the experimental sample, all participants will be recruited from the BreLAB pool at the University of Bremen. The study is aimed exclusively at German speaking participants. In each round participants will be shown a maximum of 15 so-called grids (= groups) of nine images from which they should identify the robots. To do this, participants click on the respective picture numbers. When participants have fully classified a grid of images, they continue with the next grid. If they have correctly classified all nine images in a grid, participants will earn 0.13 cents for any correctly classified grid in the first round. Participants have 2 minutes for each round. If participants have correctly classified all 15 grids within 2 minutes in the first round, they will earn EUR 1.95. Participants will not earn money for wrongly classified grids. Each round follows the same structure. After the end of the first round, participants can choose to enter the next round and classify images again. The remuneration is then reduced by 11% in each new round so that in the second round a maximum of only EUR 1.80 can be earned in 2 minutes for 15 correctly classified grids. In round 10 it would only be 75 cents per 15 correctly classified grids. Participants can stop classifying images after each round. Participants can participate in a maximum of 10 rounds. Participants receive an additional EUR 5.00 fixed payment for showing up on time.

In the crowd labor markets intervention, participants are supposed to work alone and should not talk to the other participants. In the crowd collaborative community intervention, participants can communicate with others via a chat function. In the crowd contests intervention, only the best performing participant receives a compensation after the focal round, while the other participants receive no compensation. The crowd labor markets, crowd collaborative community intervention, and the crowd contests intervention will be randomly selected with 1/3 probability respectively. If only one participant signed up for a session, the session must be a crowd labor markets session, because no communication and competition can take place.

We investigate the effect of the three crowdsourcing organization interventions on (1) participants’ propensity to participate in the classification task, (2) their productivity during each round, and (3) their willingness to return the overpay. We also investigate the second-order effect of the extent of overpay (EUR 1.00, EUR 5.00) and how it affects participants’ willingness to return the overpay. The experiment is followed by a questionnaire to conduct balancing checks.

Randomization Method

We will randomly allocate sessions (and thus participants) with 1/3 probability to the crowd labor market, crowd collaborative community, and crowd contest treatment. If only one participant signs up for an experimental session, the session must be a crowd labor market session. Treatment status is assigned on the group level.

Randomization Unit

The unit of randomization is the experimental sessions.

Was the treatment clustered?