Abstract
Facebook use continues to be nearly ubiquitous among older adolescent college students (Buffardi & Campbell, 2008; Lewis, Kaufman, & Christakis, 2008; Sachdev et al., 2012). Older adolescents have the highest use of Facebook and typically report more than 100 minutes on Facebook each day (Junco, 2011). Previous work has considered the implications of college students’ investment in Facebook; the experience of Facebook has been described positively as leading to a “flow state” (Mauri, Cipresso, Balgera, Villamira, & Riva, 2012). However, Facebook use has also been criticized as a potential distraction leading to lower grades (Kirschner & Karpinski, 2010), which may increase stress among students. While anecdotal stories and occasional reports suggest links between Facebook and stress or even depression (O’Keeffe, Clarke-Pearson, & Council on Communications and Media, 2011), little empirical research has investigated these links.
Stress is a particularly important implication of Facebook use to consider. Stress is common among college students; approximately one third report that stress negatively affects their academic performance (Boynton Health Services, 2009; Olivardia, Pope, & Hudson, 2000). Stress is also associated with negative health consequences, such as increased risks of headache or worsened depression symptoms (Björling, 2009; Yang et al., 2010). It is possible that Facebook use could worsen stress, as students fail to complete necessary tasks or engage in self-care due to prolonged Facebook use. It is also possible that Facebook use could mitigate stress by providing a way to visualize one’s social network and seek peer support.
While previous work has evaluated associations between social media such as Facebook and self-reported stress (D’Amato et al., 2012), assessments of Facebook use and biological evidence of stress are lacking. An understanding of how older adolescents use Facebook during times of stress would also provide a better understanding of this relationship. Thus, the purpose of this pilot study was to evaluate associations between Facebook use and stress through two approaches: biologic stress response measurement and patterns in Facebook use evaluation.
Method
Setting
This mixed-methods study took place at a large state university in the Midwest. Institutional review board approval was granted by the relevant institution, including permission to incorporate deception as part of the research design.
Participants
Participants included undergraduate students between the ages of 18 and 23 years. Participants were recruited using flyers posted in public campus areas describing a study about college students and stress. Interested participants were given an appointment at the study site and instructed not to eat, drink, smoke, or exercise for 2 hours prior to the appointment. Participants on the appointment list were randomized into control and Facebook groups by random number allocation. Participants who completed all study procedures received an incentive payment of $50.
Trier Social Stress Test
To use a validated procedure to assess stress, participants in this study completed a modified Trier Social Stress Test (TSST; Peterson & Seligman, 1984). The goal of the TSST protocol is to systematically induce a stress response to measure differences in biologic reaction. Meta-analysis suggests that the TSST is the most useful and appropriate standardized protocol for studies of biologic stress response (Peterson & Seligman, 1984). The protocol involves asking a study participant to prepare to deliver a speech, typically as though applying for a job, and providing the participant time alone to prepare while taking measurements. This protocol has been shown to induce changes in the concentration of salivary cortisol as well as heart rate (Weisz, McCarty, & Valeri, 2006).
The first phase of the study began on arrival at the study site. Participants were provided a consent form as well as information about how to take a saliva sample. Participants were asked to read each document thoroughly, which provided participants a 10- to 20-minute window of rest, the “anticipation period” prior to the first (Time 1) cortisol sample and pulse measurement (Weisz et al., 2006).
TSST Protocol for Facebook Group
Facebook group participants were individually escorted to a room by a research assistant and told that the study involved preparing a 10-minute biographical speech as though the participant were applying for a job. The speech’s format and content was at the participant’s discretion. Participants were told that they had approximately 20 minutes to prepare the speech before delivering it to a panel of physicians and that no personal materials could be used. Participants were informed that they would be video-recorded during preparation time to ensure compliance with study procedures. Participants were instructed that they could use Facebook for preparation and were seated at a table with a computer.
The research assistant then moved the participant’s belongings to the far corner of the room and checked all video equipment. After 5 to 8 minutes, the second (Time 2) salivary cortisol and pulse measurement were taken and the research assistant left. The participant was alone under video surveillance for 10 minutes.
After this time the research assistant entered the room and took the third (Time 3) salivary cortisol and pulse measurement. The research assistant then informed the participant that he/she did not have to give the speech. During debriefing, participants were informed about the goals of the study. They were asked to maintain confidentiality regarding the study purpose and signed a confidentiality agreement.
TSST Protocol for Control Group
Control group participants were provided the same instructions for the speech, their belongings were removed, and they were informed that they would be videotaped. Control participants were told that no materials could be used for preparation, and they were seated at an empty table. Participants then remained in the room under video surveillance for 10 minutes. After that time, salivary cortisol and pulse were measured and participants were debriefed.
Biologic Response Variables
Biologic response data included salivary cortisol samples and pulse measurements from the radial pulse, all taken at three time points during the TSST. Time 1 reflected baseline stress state, Time 2 reflected initial stressor, and Time 3 reflected the time period spent preparing for the speech.
Video Recording Procedure
Participants were recorded using free-standing video recorders to ensure fidelity to study protocol. For participants in the Facebook group, Facebook use was also recorded using screen capture video. These videos were viewed to evaluate Facebook use during the study protocol.
Video Evaluation Measures
Through an iterative process, a codebook was developed to assess videos for Facebook use actions. Videos were initially viewed to develop a list of common Facebook actions; this list yielded 13 common actions. Examples of common actions included reviewing photographs and browsing the NewsFeed. These actions were then reviewed and categorized by two investigators. Categories included purposeful (tool-based) actions, which included any actions that likely contributed to preparation for the speech and nonpurposeful (distraction) actions, which included any actions that were not associated with the task.
All videos were then evaluated by one of two trained evaluators for these 13 common actions and counts of each of these actions were recorded. A 20% subset of videos were viewed by both evaluators to assess reliability; interrater agreement for categorization of actions was 96%. After assessing total counts of tool and distraction actions, videos were categorized primarily into the category in which the majority of actions were categorized.
Analysis
The primary outcome measures were changes in pulse and salivary cortisol levels from baseline (Time 1) to final assessment (Time 3), evaluated using a paired t test. Given that previous work has illustrated gender differences in stress responses (Peabody, Luck, Glassman, Dresselhaus, & Lee, 2000), analysis of variance was conducted to evaluate differences in the primary outcome measures between control and Facebook arms and gender.
Participants’ video and pulse categorization were compared using Fisher’s exact test. Pulse counts were categorized as above or below the average for the study population (mean = 71 bpm).
Results
A total of 41 participants completed the study; 14 (36%) were male and 36 (88%) were Caucasian, mean age was 19.6 ± 1.4 years, 19 participants were in the Facebook group (Table 1).
Descriptive Data From College Student Participants.
Note. No significant differences were found between experimental and control groups for these variables.
Biologic Response Measurements
Cortisol levels in the Facebook group rose during the study period, while in the control group, cortisol levels declined (Figure 1A). Pulse values within the Facebook group increased between Time 1 and 2, and then plateaued by Time 3 (p = .001). Within the control group, average pulse increased between Time 1 and 2, and plateaued (Figure 1B). Comparing Time 1 and Time 3, there were no significant differences in cortisol or pulse change between groups.
(A) Cortisol levels among participants in Facebook (experimental) and control groups measured at three time points during study. (B) Pulse levels among participants in Facebook (experimental) and control groups measured at three time points during study.
ANOVA results testing interaction between gender and Facebook/control group illustrated no significant interaction for cortisol levels (F = 0.14, p = .7). However, evidence of gender and Facebook/control was present for pulse (F = 4.1, p = .04). Females in both Facebook and control groups showed similar increases in pulse through the experiment (mean increases = 5.1 ± 9.1 and 5.8 ± 14.3 bpm, respectively). Males in the control group demonstrated an average pulse reduction of 3.5 ± 0.2 bpm between Time 1 and Time 3, while males in the Facebook group had an average pulse increase of 10 ± 8.6 bpm, p = .01.
Facebook Use
Examples of the types of Facebook activities that participants engaged in and their categorization are included in Table 2. Overall, 58.8% of videos were identified as primarily tool based.
Categories and examples of Tool-based and Distracted use Facebook actions evaluated from screen capture videos.
Participants who demonstrated primarily distracted use demonstrated higher than average pulse rates at Time 3 compared with those who use was primarily tool based (mean 85.1 ± 10.3 bpm vs. 73.6 ± 11.2 bpm, p = .03). Among male participants, 66.6% were categorized as primarily distracted use while 45.5% of females were categorized as distracted use, though this comparison was not significant (p = .3).
Discussion
In this pilot evaluation investigating Facebook use patterns and biological stress response during a stressful event, we found no significant changes in cortisol or pulse from Time 1 to Time 3 for either study group. However, males in the Facebook use group exhibited significant differences in pulse compared to males in the control group.
Findings suggest that males who used Facebook to prepare for the stressful task had increased stress compared with those who did not use Facebook. In a previous study, men’s coping strategies in a stressful situation were commonly problem-solving based, while women’s were more likely to be emotional-based (Mundt, 2011). Facebook may have provided inadequate problem-solving tools to male participants. The social nature of Facebook may have provided access to emotional coping strategies for female participants as they viewed photographs and messages from friends and family. Previous work has shown that females use Facebook more often and more frequently for peer communication compared with males (Allen & Badcock, 2003). Thus, females may have felt comfortable in using Facebook during preparation during the stressful event compared with male peers, but not enough to derive significant lowering of stress.
In addition, we applied a codebook to Facebook videos to identify two distinct patterns of Facebook use under stress: tool based and distracted use. Participants who used Facebook as a tool had lower biologic signs of stress compared with those who engaged in distracting activities. It is possible that participants who used Facebook in a primarily tool-based manner were more frequent Facebook users. However, it is also possible that these participants were generally more organized and able to tune out potential distractions within Facebook. It is also worth considering that participants who used Facebook as a distraction may have been aware that they were not preparing for the speech, and felt more stressed as the preparation time ran out.
Limitations to this pilot study include the small sample size and single university. Control and Facebook groups were not evenly balanced as two scheduled control group participants did not keep their appointments after randomization. When investigating Facebook use data, we were limited to investigating the Facebook use group participants (N = 19). This small sample size limited our ability to conduct quantitative analyses regarding gender or biologic response data related to Facebook use. Furthermore, we did not collect baseline information regarding how participants typically use Facebook, which may have affected their use during the experiment. Finally, stress is a multifactorial condition, and other factors may have contributed to individual participant’s stress levels during the experiment.
Despite these limitations, this study represents the first work to evaluate the biologic impact of Facebook use during an acutely stressful event. Findings suggest that some college students may find Facebook use a detriment to performance under stress. We also found evidence that some college students can use Facebook to reduce stress in particular settings, such as prior to a real job interview or other situations of induced stress. Findings also suggest that including biologic measures is a feasible approach toward understanding these relationships. One population of interest for future study may be females with social anxiety, who may derive benefits from using Facebook to visualize their social support network before undertaking a stressful task.
Footnotes
Acknowledgements
The authors would like to thank Natalie Goniu, Lindsay Gordon, Hope Villiard, Alina Arseniev, and Bradley Kerr for their assistance with this project.
Declaration of Conflicting Interests
The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.
Funding
The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: The work described was supported by funding from the University of Wisconsin Graduate School.