Valuation of Personal Data: Experimental Analysis

We conduct three experimental studies in a laboratory environment. In all studies, we elicit subjects' monetary valuation of their personal data, by asking them for their willingness-to-accept the disclosure of this data in the experimental session. In Study 1, we vary the type of personal data which is collected and disclosed. In Study 2, we compare different mechanisms for eliciting subjects' monetary valuation. In Study 3, subjects face different degrees of uncertainty regarding their collected personal data.

2019-08-21

2021-03-31

Subjects’ (incentivized) monetary valuation of their personal data, i.e., their willingness-to-accept the disclosure of their data to other participants in the laboratory.

Subjects’ (hypothetical) willingness-to-pay (WTP) for their personal data, i.e., the maximum price that they would be willing to pay to prevent disclosure of their data to other participants in the laboratory.

Experiments are run in an experimental laboratory at the School of Business, Economics and Information Systems at the University of Passau. Treatments are randomized at the session level. Participants will be recruited from the student subject pool of the University of Passau. Each subject participates in only one of the three studies and in only one treatment (between-subject design). In all three studies, subjects are fully informed about the timeline of the experiment and the consequences of their actions. In the experiment, personal data is only disclosed if subjects give their explicit consent in accordance with the European General Data Protection Regulation. Subjects receive a monetary compensation for disclosing their data in addition to a participation fee. We record subjects' monetary valuation of their personal data (revealed willingness-to-accept, hypothetical willingness-to-pay), the data that is collected and put under threat of disclosure in the laboratory, subjects' answers to an ex-post questionnaire and to several manipulation checks, as well as background statistics and variables related to the data valuation procedure.

In Study 1, we vary the data type for which subjects' monetary valuation is elicited. In three treatments, we collect data on the subjects' opinion on controversial topics (DT1), subjects' performance in a logic test (DT2) and subjects' sensitive, personal statements (DT3). Here, we use the Becker–DeGroot–Marschak (BDM) mechanism to measure subjects' willingness-to-accept (WTA). Subjects state their WTA after the data is collected and without any additional feedback. If possible within time and budget constraints, we plan to conduct an additional treatment (DT4), where we collect subjects' personal information from a popular online service, which they actively use, and where we measure their WTA for this type of "field data".

In Study 2, we compare different elicitation mechanisms for personal data and run two additional treatments. Specifically, we compare subjects' revealed WTA in a reverse Vickrey Auction (EM2) and subjects' hypothetical WTA in a survey (EM3) to subjects' valuation in a BDM mechanism, for which we re-use the experimental outcomes from treatment DT1. In all treatments of study 2, we collect data on subjects' opinion on controversial topics.

In Study 3, we vary the information uncertainty that subjects face when valuing their personal data. We use the BDM mechanism to elicit subjects' WTA and collect data on subjects' performance in a logic test. Thus, we re-use treatment DT2 as a benchmark and run two additional treatments: In the first treatment subjects receive additional feedback on their test score and their relative performance in the logic test compared to other participants in the same sessions (IU2). In the second treatment, subjects receive less information and state their WTA before they answer the questions of the logic test (IU3).

Participants will be recruited from the subject pool of the University of Passau using ORSEE (Greiner, 2015) and they are paid a participation fee of 12 Euro each. One participant in each session can additionally earn up to 50€ for data disclosure in the laboratory. As the experiment is conducted in German, participants must be proficient in the German language. A session is expected to last about 60 minutes. The experiment is programmed using the experimental software oTree (Chen, Schonger, and Wickens, 2016). Participants are fully informed about the potential data disclosure in the laboratory and have to sign consent forms at the beginning of the experiment. Additionally, each participant may exit the experiment at any time. The general experimental procedure is as follows: After signing the consent forms, participants have to answer several comprehension questions. Next, personal data about the participants is collected and their monetary valuation (i.e., WTA) is elicited. Then, participants have to fill out a questionnaire with questions on the experimental procedure, privacy attitudes and demographics. At the end of each session, one participant is selected (randomly in the case of the BDM treatments) and the experimental outcome is implemented. In the case of data disclosure, the selected participant has to come to the front and say her name which is verified by means of her student card. Additionally, a photo of the selected participant is taken. Her data, her name and photo are then disclosed in the laboratory by displaying them on the participants’ screens.

Randomization by computer in office

Experimental sessions

No

Observations on the subject level are assumed to be independent, because subjects decide only once on their data valuation and decide without interacting with other participants in the experimental session. Thus, the number of clusters equals the number of observations.

We schedule data collection aiming at 75 observations per treatment. This corresponds to five experimental sessions each, with 15 participants in a single session. Thus, we aim for a total of 500-600 individual participants across the three studies.

DT1: 75 participants, DT2: 75 participants, DT3: 75 participants, EM2: 75 participants, EM3: 75 participants, IU2: 75 participants, IU3: 75 participants