The Behavioral Effects of Social Media

How to cite

Guo, Xiongfei; Mcnamara, Trent; Mosquera, Roberto; Odunowo, Mofioluwasademi; Petrie, Ragan. 2019. "The Behavioral Effects of Social Media." AEA RCT Registry. AEARCTR-0003952. https://doi.org/10.1257/rct.3952-1.1

Download: BibTeX · RIS — for reference managers without a browser translator. Zotero and most others pick up this page's citation metadata directly.

Xiongfei Guo* , Trent Mcnamara , Roberto Mosquera , Mofioluwasademi Odunowo , Ragan Petrie

* Primary investigator

Registry ID
AEARCTR-0003952
DOI
https://doi.org/10.1257/rct.3952-1.1
First registered
Last updated
Status
Completed
Trial dates
2017-04-10 – 2018-07-15
Countries
United States of America (Texas)
JEL codes
C93, D91, D83, I31
Registry record
http://www.socialscienceregistry.org/trials/3952

Abstract

Social media permeates many aspects of our lives, including how we connect with others, where we get our news and how we spend our time. Yet, we know little about the economic effects for users. In 2017, we ran a large field experiment with over 1,765 individuals to document the value of Facebook to users and its causal effect on news consumption and awareness, well-being and daily activities. Participants reveal how much they value one week of Facebook usage and are then randomly assigned to a validated Facebook restriction or normal use. Those who are off Facebook for a week reduce news consumption, are less likely to recognize politically-skewed news stories, report being less depressed and engage in healthier activities. One week of Facebook is worth $67, and this increases by 19.6% after experiencing a Facebook restriction (33% for women), reflecting information loss or that using Facebook may be addictive.

Intervention

We elicit the value of Facebook using an incentive compatible mechanism. Qualified participants are randomly assigned to a validated Facebook restriction or normal use.

Experimental design

In order to analyze the effects of social media on news awareness, consumption behavior, time allocation and subjective wellbeing the first step is to recover how much people value social media. It is not enough to just call for volunteers for an experiment that closes access to Facebook for a week, since the people that will likely volunteer are those that value Facebook the least. Using a selected sample of volunteers to estimate the effects of not accessing Facebook on the proposed outcomes will yield results that are not generalizable, not even to the population of college students. Thus, it is important to understand the mechanism behind this selection process.

In this sense, the first step of this protocol is to implement a recruitment mechanism that allows to recover the distribution of the value of Facebook in the potential population. Then, we can use this distribution to understand the selection process into the experiment and potentially we could use the valuation distribution to weight the estimates to obtain more externally valid results.

Recruitment mechanism
Participants for this study will be selected from the population of Texas A&M students (graduate and undergraduate) in the College Station campus.

In order to enroll the participants for this study we will apply the following procedure designed to recover the distribution of the value Facebook has for college students:
1. Send a massive email to Texas A&M students in the College Station campus indicating the general characteristics of the experiment. The email includes a link that directs the reader to a Qualtrics survey web page where we will select participants using the BDM mechanism.
2. The first screen in Qualtrics will collect basic demographic information (gender, age in years, major, zip code/country of birth) and will explain briefly the main ideas of the experiment.

3. In the second screen, subjects are explained the functioning of the selection mechanism.

4. The third screen displays only a field to capture the student’s valuation of her time on Facebook, with a description.

5. The forth screen presents the random counter offer. For the case where the counter offer is greater or equal than the valuation, the screen will indicate that the subject has been preselected to participate, that we are going to randomly select the final participants who will be paid the counter offer at the end of the experiment, will provide information on the date and location of the experimental session, will indicate that email reminders will be sent before the experiment, and it will request for the participant’s email address. For the case where the counter offer is less than the valuation, the screen will thank the participant for filling his valuation and basic data.

Experimental Sessions
Treatment

All participants will fill a survey collecting information about the outcomes before the interventions. Half of the participants will be randomly chosen not to access their Facebook (both the news feed and Facebook messenger) for a period of 1 week.

For the treated group, we will implement the following procedure to log them off Facebook:
1. Before the session, we will create a Facebook account for the experiment.
2. Provide participants access to laptop computer.
3. Ask the participants to become Facebook friends with the experiment’s account.
4. Guide each participant through the procedure of logging off from every device where his/her accounts are active using Facebook’s controls.
5. The experimental account will be permanently closed at the end of the experiment.

Design details

Randomization method
Randomly assigned ID numbers, randomization is done by a computer
Randomization unit
Individual
Clusters
0
Observations
For the value of Facebook 2,000, for randomization 170
Arms
85 in treatment, 85 in control
Primary outcomes
Value of Facebook, news awareness, news access through different media, news consumption, core subjective wellbeing measures, daily activities, and time use

Papers from this trial

  1. Mosquera, R., Odunowo, M., McNamara, T., Guo, X., & Petrie, R. (2020). The economic effects of Facebook. Experimental Economics, 23, 575-602. https://link.springer.com/article/10.1007/s10683-019-09625-y

Cited by

No citation data available for this registration yet.

Keywords