Facebook-to-Facebook: online communication and economic cooperation

ABSTRACT

Direct face-to-face communication has traditionally been found to be more effective for fostering economic cooperation than any form of indirect, mediated communication. We inquire whether this is still the case since most young adults routinely use texting and online social media to communicate with each other. We find that young adults in our laboratory public goods experiment are just as adept at finding and sustaining cooperative agreements when communicating within a Facebook group and through online chat as they are in person.

KEYWORDS:

• Laboratory experiments
• communication
• public good
• voluntary contribution mechanism

JEL CLASSIFICATION:

• C91
• D83
• H41

Acknowledgements

We gratefully acknowledge the support of a Technology, Innovation and Society grant from the University Research Council of the University of Hawai’i. Lynham acknowledges the support of the National Science Foundation through grant GEO-1211972.

Disclosure statement

No potential conflict of interest was reported by the authors.

Supplemental material

The supplemental material for this article can be accessed here.

Notes

¹ See also Fiedler and Haruvy (2009), Bicchieri and Lev-On (2011), Greiner, Güth, and Zultan (2012), Greiner, Caravella, and Roth (2014).

² The communication length, 10 mins, is as in Brosig, Weimann, and Ockenfels (2003). The language of the experimental instructions explaining the communication part is in line with earlier studies on the effect of communication on VCM play (Isaac and Walker 1988). See online Suplementary Materials A and E for experimental instructions and samples of communication logs.

³ One needs a Facebook (FB) account to join an FB group. As FB does not allow fake accounts, participants’ genuine accounts had to be used. In Session 1 of the FB treatment, three participants did not have FB accounts, which had to be created on the spot with the experimenter’s help. For later sessions, we requested all of our participants, irrespective of treatment, to have an FB account. Given that around 90% of young American adults use FB (see Section I), the subject pool selection bias introduced by this recruitment restriction is minimal.

⁴ Although Bochet et al. (2006) also compared NC, FTF and Chat, our experiment differs in that (1) all our participants played 10 periods of the VCM before communication and (2) only one 10-min communication period was allowed in all our treatments. The former ensures that participants fully understood the game and were better informed during the communication round while the latter ensures that our FTF treatment is comparable to our Chat treatment. In Bochet et al. (2006), FTF communication happened once before the first round of a 10-round VCM while Chat communication happened before the first, fourth and seventh rounds of a 10-round VCM.

⁵ We are grateful to an anonymous referee for this remark. This ignores end-game effects.

⁶ We have fewer groups in the NC treatment since NC serves only as a benchmark. The target number of communication groups was 7 or 8 per treatment. We accidentally conducted one extra session with 2 groups in the FTF treatment, resulting in 10 groups for FTF. We include all data collected in our analysis. The results are not sensitive to exclusion of the extra FTF session.

⁷ In spite of the recruitment request that all participants should have a FB account, there were four individuals who participated in non-FB sessions but reported in the post-experiment questionnaire that they did not have FB accounts.

⁸ See Deaton (1998, 91) on the dangers of using nonlinear regressions when the distribution of the error term is unknown. As a robustness check, we present the results of both linear and nonlinear regression estimations. We do not use nonparametric tests because the groups within sessions are not independent, due to the rematching after Part 1.

⁹ Here is a typical example of an agreement: ‘…Everyone cool with putting in all 10 every time?’ ‘Sounds good to me.’ ‘Same here.’ ‘Yes, sounds like a good idea…’ (FB Group 2). More samples of communication logs are provided in online Supplementary Material E.

¹⁰ Observing frequent defections in the last period of the post-communication VCM game validates our results even further, indicating that our participants were fully aware of the strong incentive to defect, but did not act on it until the very last period.

Additional information

Funding

Lynham acknowledges the support of the National Science Foundation through grant GEO-1211972.