Social Media Use Early in the Pandemic Predicted Later Social Wellbeing and Mental Health in a National Online Sample of Adults in the United States

Abstract

Social media use increased early in the Covid-19 pandemic, but little information is available about its impact. The present study examined associations of frequency of use of different social media and the motives for use with subsequent social wellbeing and mental health. Data were gathered on a nationwide sample of 843 Americans during the first wave of lockdowns and infections in mid-April 2020, and again five weeks later. Participants were adults ages 20 to 88 years old (M=39.3 years old) recruited from Amazon Mechanical Turk (MTurk). Controlling for age and gender, greater frequency of Facebook and video chat apps use predicted higher levels social support, but also higher levels of cumulative Covid-related stress appraisals and posttraumatic stress symptoms. Greater use of video chat apps also predicted less loneliness. Greater use of both Instagram and Snapchat predicted more anxiety and cumulative Covid-related stress appraisals. Greater use of Instagram also predicted higher levels of posttraumatic stress symptoms. Motives for use (e.g., connect with others, waste time/avoid responsibility, online video gaming with others) also differentially predicted social wellbeing and mental health. Results indicate that greater social media use early in the pandemic was often associated with more distress and lower levels of social wellbeing but effects varied depending on types, frequency, and motivations for use. Overall, the study revealed that social media use related to social wellbeing and mental health in complex ways.

Introduction

As the COVID-19 pandemic spread across the U.S. in late March, 2020, the Center for Disease Control and Prevention issued guidelines for social distancing, restricting people’s usual social interactions.¹ While keeping people physically safe, social distancing had adverse mental health effects and social isolation.² Mental health concerns including loneliness increased during the pandemic.^2,3,4, The inability to interact safely in person made social media a primary way to connect with others. Early studies from China and the United States ^5,6 indicated that many people increased their social media use during the pandemic. In the present study, we aimed to determine how social media use early in the pandemic related to subsequent social wellbeing and mental health.

Social support theory suggests that social support buffers the impact that a stressful life event has on health.⁷ Supportive actions may lead to better coping while perception of available support may decrease the amount of threat one feels from a stressful event.⁷ However, less is known about whether social media can serve the role of social support. Research findings regarding social media’s general effects on social support are mixed.⁸ A systematic review of 70 studies concluded that social media may provide greater feelings of social support.⁸ For example, a nationally representative survey of American adults demonstrated that routine use of social media related to increased social wellbeing;⁸ another showed that daily use of Facebook associated with less loneliness.¹⁰ Research prior to the pandemic suggested that the use of different types of social media and the motives for their use may differentially relate to subsequent feelings of loneliness.¹⁰ In another study, using social media for fun, for online community, and for providing and receiving social support were associated with overcoming loneliness among older adults.¹¹ One study demonstrated that using social media to stay connected with friends related to higher quality of life but using it for dating or to alleviate boredom related to lower quality of life.¹²

On the other hand, a systematic review before the pandemic demonstrated that social media use carries potential risks such as higher levels of depression and anxiety but can also provide benefits including lower levels of distress.⁸ Early in the pandemic, social media may mitigate or exacerbate the negative experience of restricted in-person socializing. One study demonstrated increased average amount of time spent on social media related to mental distress among U.S. adults early in the pandemic.⁶ Another study demonstrated that longer time spent on social media predicted higher depression and suicide ideation among adolescents.¹³ Similarly, a study among adults in China indicated that frequent use of social media during the pandemic positively associated with anxiety and depression,⁵ perhaps due to misinformation which can create confusion and distress.⁵ A study of German adults also showed using social media related to higher anxiety and depression.¹⁴ Further, a study of Israeli adults suggested that one’s perception of media may impact mental health differently.¹⁵ Despite these findings, several studies have indicated that technology may buffer loneliness and isolation that can exacerbate mental health problems by connecting people to physically distant loved ones during Covid-19.^16,17

Thus, evidence is mixed regarding how frequency of and motivations for social media use relate to social wellbeing and mental health during the pandemic. Given that few studies of social media use during Covid-19 have been based on U.S. samples, this exploratory study offers insight into how social media use was associated with social wellbeing and mental health in a U.S. sample early in the pandemic. Specifically, we posed these specific research questions: Early in the pandemic/lockdown in the U.S.,

1. How frequently did people use each type of social media at T1 in the past 30 days?

2. What were people’s motives for using social media at T1?

3. How did frequency of different types of social media use and motives for that use at T1 relate to social wellbeing and mental health at T2?

Methods

Participants

The study and all study materials were approved by the University of Connecticut Institutional Review Board (X20–0057) as an exempt protocol. All participants consented to participate in the study. MTurk is an online platform where people sign up as workers and complete various tasks. Eligible participants were aged 18 or older, residing in the U.S., and able to read English. We attempted to oversample young adults given our research team’s concurrent research projects on coping among 18–25 year-olds. Despite this effort, we struggled to engage and retain younger participants.

We used best practices to screen out suspicious or poor-quality responses as recommended for MTurk data (e.g., removal of inattentive cases and responses originating outside valid locations in the U.S., ensuring unique human responders as opposed to computerized bot responses., fast respondent screening based on average response time) through the use of time to completion, Captcha, and GPS coordinate confirmation.^{18, 19} At baseline (Time 1; T1), we received 1,578 unique responses, and at follow-up (Time 2; T2), the sample included 843 participants of the participants that responded at T1. Data collected through MTurk has been shown to be of high quality, replicable, and valid compared with other frequently used platforms.^20,21 MTurk workers have been shown to be more diverse than typical student or online forum samples and fairly representative of larger populations, including the U.S.^{20, 21}In some research, MTurk respondents reported slightly higher levels of depression,²² but in other studies, the mental health (as assessed by the DASS-21, also used in the present study) of MTurk workers was similar to the general U.S. population.^{23, 24}

Our sample identified as 44.6% (n=359) male, 55.6% (n=469) as female, and 1.8% (n=15) as non-binary, transgender, or provided another descriptor. Participants were an average of 39.3 years old (SD = 14.1, range = 20–88) and primarily White (n = 685, 81.3%). The sample included 12.5 % (n=105) African-Americans, 12.3% (n=104) Asian/Asian-Americans, 5.0% (n=42) Native Hawaiian/Other Pacific Islanders, and 6.3 % (n=53) American Indian/Alaskan Natives. Regarding ethnicity, 7.0% (n=59) identified as Hispanic/Latinx. Most participants identified as straight/heterosexual (n = 753, 89.3%); 3.9% (n= 33) identified as gay or lesbian, 4.6% (n = 39) as bisexual and 2.1% (n = 18) preferred to self-describe or not say. Most participants were married (n = 338, 40.1%) or single (n = 295, 35.0%); others were cohabiting with a significant other they were not married to (n = 119, 14.1%), divorced (n = 69, 8.2%), widowed (n = 14, 1.7%), or separated (n = 8, 0.9%). Nearly one-fourth (n =197; 23.4%) reported being a caregiver in their home. Locations were reported across the U.S. (n = 188, 18.5% in the Northeast; n = 199, 19.6% in the Midwest; n = 244, 30.0% in the West; and n = 383, 37.7% in the South) (see Table 1). At the 30-day follow-up (T2), most (n = 582, 69.0%) participants reported that their state had active shelter-in-place guidelines.

Table 1:

Demographics

Age (years), M (SD)	38.9 (13.5)
Gender, N (%)
Male	453 (44.6%)
Female	547(53.9%
Non-binary/third gender	3 (0.3%)
Transgender	4 (0.4%)
Prefer to self-describe	4 (0.4%)
Prefer not to Say	4 (0.4%)
Race, N (%)
Black/African American	122 (12.0%)
Asian/Asian American	121 (11.9%)
Native Hawaiian/other Pacific Islander	50 (4.9%)
American Indian/Alaska Native	70 (6.9%)
White	836 (82.4 %)
Ethnicity, N(%)
Hispanic	86 (8.5%)
Non-Hispanic	929 (91.5%)
Sexual orientation, N (%)
Straight/heterosexual	895 (88.2%)
Gay or lesbian	40 (3.9%)
Bisexual	64 (6.3%)
Prefer to self-describe	7 (0.7%)
Prefer not to say	9 (0.9%)
Geographic state, N (%)
West	244 (24.1%)
Midwest	199 (19.6%)
South	383 (37.7%)
Northeast	188 (18.5%)
Marital status, N (%)
Married	407 (40.1%)
Single	359 (35.4%)
Divorced	73 (7.2%)
Separated	14 (1.4%)
Widowed	18 (1.8%)
Living with but not married	144 (14.2%)
Caregiver status, N(%)
Yes	219 (21.6%)
No	796 (78.4%)

Data Collection

Participants voluntarily signed up for the study on the MTurk homepage and provided informed consent prior to screening and completing baseline (T1) questionnaires. The project was advertised as an anonymous, longitudinal study of the impact of Covid-19 on daily life, providing participants with $2 for completing the T1 survey and $3 for subsequent surveys. The MTurk platform provides compensation directly to participants in U.S. dollars or Amazon.com gift cards.

Data presented here are drawn from the T1 survey, administered April 8–25, 2020 (approximately 3 weeks after the president declared a national emergency²⁵, and widespread shelter-in-place recommendations were first issued in the U.S.) and the T2 survey, administered May 15–29, when many areas of the U.S. had begun to implement phased reopening plans.²⁶ Each survey took approximately 20 minutes to complete.

Measures

Demographics

At T1, participants reported on their location by state, financial security, status as primary caregiver for a dependent in their home, partner status, gender, sexual orientation, race, ethnicity, and age. For descriptive purposes, states were categorized into four distinct regions based on divisions used in the U.S. Census (2020).

Social Media

Measures of social media use were created specifically for the present study. At T1, participants were presented with a list of social media sites (Instagram, Facebook, Snapchat, and video chat apps (e.g. Zoom, Facetime, etc.) and asked whether they had ever used each, answered as “yes” or “no”. For each endorsed as ever having used, they were asked how often they used or visited that site in the past 30 days, rated from “several times a day” (4), “about once a day” (3), “a few times a week” (2), “every few weeks” (1), and “less often” (0). Participants were asked to indicate motives for using social media sites in the past week (mental health appointment, substance use support groups, dinner date/happy hour, exercise with others, online video gaming with others, connect with friends, waste time/avoid responsibility, and social anxiety/discomfort), by endorsing each motive with “yes” or “no.”

Social Wellbeing

At T2, participants completed the 4-item appraisal subscale of the Interpersonal Support Evaluation List-12 (ISEL-12)²⁷ to assess perceived availability of supportive others. The ISEL-12 has demonstrated strong psychometric properties in a wide range of samples. We selected the appraisal subscale as the most appropriate type of social support to assess given social distancing and barriers to physical interaction and travel early in the pandemic. Items on the ISEL-12 are rated from 0 (“definitely false”) to 3 (“definitely true”) and summed to create a total score. Cronbach’s alpha in the present sample was .88.

Loneliness

At T2, participants also completed a Three-Item Loneliness Scale²⁸ as a measure of frequency of perceived loneliness. The Three-Item Loneliness Scale is a shortened version of one of the most widespread scales of loneliness: the Revised UCLA Loneliness Scale (R-UCLA)²⁹ and has shown good psychometric properties in a study that examined participants with major depressive disorder and social anxiety disorder³⁰; the scale has demonstrated good construct validity and other psychometrics.²⁸ Cronbach’s alpha in the present sample was .91. Each item is rated from 0 (“hardly ever”) to 3 (“often”).

Mental Health

Mental health was assessed at T2 using five different outcomes. The 21-item version of the Depression, Anxiety, and Stress Scales (DASS-21)³¹ assessed the extent to which depression, anxiety and stress were experienced in the past week, rated from 0 (“Did not apply to me at all”) to 3 (“Applied to me very much or most of the time”). Summed scores are multiplied by 2 to create separate seven-item subscales. The depression subscale taps into depressive symptoms such as feeling “downhearted and blue”, the anxiety subscale covers mostly somatic symptoms of anxiety (e.g., shortness of breath, trembling), and the stress subscale captures primarily stress reactivity and arousal (e.g., ‘touchy’, ‘agitated’ and ‘difficult to relax’).The DASS-21 demonstrated good psychometric properties in samples of Chinese exposed to Covid-19³² and MTurk workers.³³ Cronbach’s alphas in the present sample at T2 were .94 for depression, .91 for anxiety, and .91 for stress. At T2, participants indicated whether they had experienced a set of Covid-19 stressors in the past week (“yes” or “no”) and how stressful they found each from 0 (“not stressful at all”) to 4 (“extremely stressful”), producing an overall perceived stress score.³⁴ Finally, participants completed the Impact of Events scale-Revised (IES-R),³⁵ 17 items that measure symptoms of posttraumatic stress, in this case, specific to Covid-19. Items are rated from 0 (“not at all”) to 4 (“extremely”) and summed. IES-R has demonstrated validity in measuring trauma-related distress.^35,36, Present sample Cronbach’s alpha was .91.

Analytic Approach

Before addressing our research questions, we described participants’ lifetime social media use and how social media use and wellbeing varied by demographic characteristics. To address Research Question 1, we characterized frequency of use of each social media by means. To address Research Question 2, we reported frequency of motives for social media. To address Research Question 3a: we conducted bivariate correlations between Frequency of Use of each social medium over the Past Month with Social Wellbeing and Mental Health. To address Research Question 3b, we conducted a series of independent t-tests comparing participants who responded “yes” to using social media sites for a specific motive to those who responded “no” on all seven of our social and mental health outcomes.

Results

Descriptive Results

Most participants reported some form of lifetime social media use: 88.96% (n=741) reported having ever used Facebook, with mean age of 39.08 years; 74.43% (n=620) reported having ever used video chat app, with mean age of 37.17 years; 69.15% (n=576) reported having ever used Instagram, with mean age of 36.67 years; 44.12% (n=368) reported having ever used Snapchat, with mean age of 32.16 years.

The frequency of social media use varied by demographic characteristics; age was significantly associated with less use of Instagram, Snapchat, and video chat apps. Women used Facebook more often. Those identifying as white used less Instagram and video chat apps, while those identifying as Hispanic/Latinx used Snapchat more often.

Social wellbeing and mental health also varied by demographics. Age was negatively associated with social support, loneliness, depression, anxiety, stress, Covid-related stress appraisals, and posttraumatic stress symptoms. Women reported more stress and Covid-related stress appraisals. Neither race nor ethnicity was associated with any of the social wellbeing or mental health variables.

Research Question 1: Frequency of use.

Participants’ frequency of social media use differed across platforms during the past 30 days, early in the pandemic. Participants used Snapchat most often, on average between “about once a day” and “several times a day” (M=4.64), followed by video chat apps, used on average between “a few times a week” and “about once a day” (M=3.89), Instagram, used on average, between “a few times a week” and “about once a day” (M= 3.58), and Facebook, used between “every few weeks” and “a few times a week” (M=2.55).

Research Question 2: Motives for use.

Participants reported using social media for a variety of motives: 74.67% (n=622) of participants used social media to “connect with friends”; 71.70% (n=514) used social media to “waste time/avoid responsibility” ; 37.45 % (n=312) used social media for “online video gaming with others”; 35.77% (n=298) used social media because of “social anxiety/discomfort.”; 11.16% (n=93) used social media for “dinner date/happy hour”; 9.12% (n=76) used social media to “exercise with others”; 4.92% (n = 41) used social media for a “mental health appointment”; and 2.4% (n=20) used social media for “substance use support groups.”

Research Question 3a: Associations of Social Media Frequency of Use Over the Past Month with Social Wellbeing and Mental Health.

Bivariate correlational analyses indicated that more frequent use of Instagram was not associated with social support, loneliness, depression, anxiety, or stress, but was correlated with higher cumulative Covid-related stress appraisals and posttraumatic stress symptoms. Using Facebook more frequently was correlated with more social support and less loneliness, but more cumulative Covid-related stress appraisal; it was not correlated with depression, anxiety, stress, or posttraumatic stress symptoms. Greater frequency of use of Snapchat was associated with higher levels of cumulative Covid-related stress appraisals and posttraumatic stress symptoms but not correlated with social support, loneliness, depression, anxiety, or stress. Using video chat apps more frequently was significantly correlated with more social support and less depression, but also with more cumulative Covid-related stress appraisals and posttraumatic stress symptoms; it was not significantly correlated with loneliness, anxiety, or stress (see Table 3). A series of hierarchical multiple linear regressions were conducted, testing frequency of use of each of the four types of social media on each of our seven outcomes, controlling for gender and age. Results indicated that after controlling for age and gender, greater use of Facebook at T1 predicted higher levels of T2 social support (b=.14, p<.05), but also predicted higher levels of cumulative Covid-related stress appraisals (b=.13, p<.05) and higher levels of posttraumatic stress symptoms (b=.07, p<.05). Greater use of video chat apps predicted higher levels of T2 social support (b=.23, p<.05), lower T2 loneliness (b=−.07, p<.05) and (marginally) less depression (b=−.06 p=.08) but also higher levels of cumulative T2 Covid-related stress appraisals (−20, p<.05) and higher levels of posttraumatic stress symptoms (b=.09, p<.05). More frequent use of Instagram predicted more anxiety (b=.09, p<.05), higher levels of cumulative Covid-related stress appraisals (b=.14, p<.05), and higher levels of posttraumatic stress symptoms (b=.10, p<.05). More frequent past month use of Snapchat predicted more anxiety (b=.09, p<.05) and higher levels of cumulative Covid-related stress appraisals (b=.12, p<.05). No specific social media platform significantly predicted T2 stress levels.

Table 3:

Bivariate Correlations Between Frequency of Use and Social Wellbeing and Mental Health Outcome

	Frequency of use M (SD)	Correlations (r) between frequency of use and social wellbeing and mental health outcome
		Social Support	Loneli ness	Depress-ion	Anxiety	Stress	Covid-related Stress Appraisals	Postttraumatic Stress Symptoms
Instagram	3.58 (2.03)	−.00	.06	.05	− .07	− .05	.16**	.15**
Facebook	2.55 (1.78)	.17**	−.08*	−.07	.03	−.03	.08*	.05
Snapchat	4.64 (1.84)	.04	.07	−.03	.04	−.06	.19**	.14**
Video chat apps	3.89 (1.63)	.14**	−.07	−.08*	.07	−.02	.17**	.11**

^**Correlation is significant at the .01 level (2-tailed).

^*Correlation is significant at the .05 level (2-tailed).

Research Question 3b: Motives for Use and Social Wellbeing and Mental Health.

A series of independent t-tests compared participants who responded “yes” to using social media sites for a specific motive to those who responded “no.” Participants who used social media sites for a “mental health appointment” in the past week reported higher depression t(831)=6.42, p<.05 anxiety t(42.36)=6.57, p<.05, stress t(831)=−5.94, p<.05, cumulative Covid-related stress appraisals t(819)=2.34, p<.05, and posttraumatic stress symptoms t(792)=−5.73, p<.05, and loneliness t(828)=3.34,p<.05, but did not differ on perceived social support t(831)= −1.8, p>.05. Those who had used social media sites for “substance use support groups” in the past week reported higher anxiety t(19.38)=4.78,p<.05, stress t(831)=4.08,p<.05, posttraumatic stress symptoms t(792)=6.37, p<.05, depression t(19.55)=2.87, p<.05 and cumulative Covid-related stress appraisals t(819)=3.06, but did not differ on perceived social support t(831)=−1.74, p>.05 or loneliness t(828)=1.47, p>.05. Those who had used social media sites for “dinner date/happy hour” in the past week reported higher cumulative Covid-related stress appraisals t(819)=4.27, p<.05, anxiety t(112.92)=2.23, p<.05, stress t(831)=2.85, p<.05, but also higher perceived social support t(139.28)=3.80, p<.05 and slightly lower loneliness t(136.228)=−2.21,p<.05; there were no differences on depression t(831)=2.10, p>.05 or posttraumatic stress symptoms t(792)=1,44, p>.05. Those who had used social media sites to “exercise (with others)” in the past week reported slightly higher cumulative Covid-related stress appraisals, and posttraumatic stress symptoms (2.18<t<3.37) but also higher perceived social support; they did not differ on loneliness t(828)=−1.34,p>.05, depression t(831)=.24, p>.05, anxiety t(86.68)=1.54, p>,05, or stress t(831)=.37,p>.05. Those who had used social media sites for “online video gaming with others” in the past week reported slightly higher depression, anxiety, stress, cumulative Covid-related stress appraisals, and posttraumatic stress symptoms (2.17<t<3.74), but did not differ on perceived social support t(831)=1.64, p>.05 or loneliness t(707.40)=−.76, p>.05. Those who had used social media sites to “connect with friends” in the past week reported higher perceived social support t(306.66)=5.35, p<.05 and cumulative Covid-related stress appraisals t(819)=4.03, p<.05, but did not differ on loneliness t(332.46)=−1.54.p>.05, depression t(328.63)=−1.44, p>.05, anxiety t(831)=−.13, p>.05, stress t(831)=.64,p>.05, or posttraumatic stress symptoms t(792)=1.27, p>.05. Those who had used social media sites to “waste time/ avoid responsibility” in the past week reported higher loneliness t(753.35)=7.21, p<.05, depression t(807.21)=8.36, p<.05, stress t(747.49)=8.46,p<.05, cumulative Covid-related stress appraisals t(819)=4.10, p<.05, and posttraumatic stress symptoms t(685.57)=6.48, p<.05; they also reported slightly higher anxiety t(755.12)=3.94, p<.05 and slightly lower perceived social support t(831)=−2.62, p<.05. Those who had used social media because of “social anxiety/ discomfort” in the past week reported higher loneliness t(558.71)=7.18, p<.05, depression t(498.26)=10.86, p<.05, anxiety t(438.91)=9.77, p<.05, stress t(497.73)=13.08, p<.05, cumulative Covid-related stress appraisals t(519.69)=9.59, p<.05, and posttraumatic stress symptoms t(456.17)=11.13, and lower perceived social support t(831)=−3.44, p<.05 (see Table 4).

Table 4:

Results of t-tests Comparing T1 Motives for Use and T2 Social Wellbeing and Mental Health

	Social Support		Loneliness		Depression		Anxiety		Stress		Covid-related Stress Appraisals		Posttraumatic Stress Symptoms
Motive for use	Yes	No	Yes	No	Yes	No	Yes	No	Yes	No	Yes	No	Yes	No
Mental health appointment	8.20	9.12	5.85	4.82**	18.88	8.48**	14.98	4.79**	18.15	9.42	33.70	27.25*	30.95	16.31**
Substance use Support Group	7.85	9.09	5.50	4.85	17.40	8.78**	17.50	4.99**	18.20	9.65**	39.37	27.29**	40.33	16.51**
Dinner date/ happy hour	9.98	8.95**	4.53	4.91*	9.20	8.96	7.08	5.06*	12.45	9.53**	34.61	26.66**	19.38	16.76
Exercise (with others)	9.70	9.00*	4.59	4.90	9.26	8.96	6.76	5.14	10.24	9.82	34.99	26.83**	24.35	16.35**
Online videogaming with others	9.29	8.92	4.80	4.90	10.01	8.38*	6.66	4.46**	10.78	9.30*	30.23	25.96**	18.99	15.88**
Connect with friends	9.44	7.96**	4.80	5.05	8.67	9.93	5.27	5.35	9.97	9.50	28.93	23.43**	17.47	15.81
Waste time/ avoid responsibility	8.84	9.43**	5.23	4.29**	11.12	5.56**	6.09	4.00**	11.86	6.61**	29.45	24.47**	19.83	12.56**
Social anxiety/ Discomfort	8.56	9.34**	5.52	4.51**	14.19	6.09**	9.00	3.22**	15.34	6.80**	35.14	23.27**	25.57	12.32**

^**p<.01

^*p<.05

Discussion

This study aimed to gain a better understanding of how Americans’ social media use early in the pandemic related to their subsequent social wellbeing and mental health 30 days later. In general, we found that the frequency of use of specific types of social media and the motivations for use related in complex ways to social wellbeing and mental health symptoms. Early studies from China⁵ and the U.S.⁶ indicated that many people increased their social media use during the pandemic. Although we did not measure their use prior to the pandemic, our participants reported frequent use of a variety of social media early in the pandemic. Among the types of social media examined, Snapchat was used most frequently, followed by video chat apps, Instagram, and Facebook. The high levels of social media use among MTurk workers¹² makes this an ideal group for examining social media use on social wellbeing and mental health.

Frequency of social media use related to different levels of indicators of social wellbeing and mental health for some of the platforms. More frequent use of Facebook use predicted higher levels of perceived social support, but also higher levels of cumulative Covid-related stress appraisals and posttraumatic stress symptoms. Higher frequency of use of video chat apps predicted higher levels of perceived social support, cumulative appraised Covid-related stress and posttraumatic stress symptoms and lower levels of loneliness. More frequent use of these two types of social media appeared to fulfill their purpose of connecting people and fostering social support virtually across age and gender. These findings are consistent with previous research suggesting that social media may relate to lower levels of distress and greater feeling of social support⁸ and social wellbeing.⁹ At the same time, use of Facebook and video chat apps predicted higher levels of cumulative Covid-related stress appraisals and PTSD symptoms. This finding is consistent with results of another study among U.S. adults which found an indirect effect of traditional and social media use on Covid-19 exposure and higher levels of stress and depression.³⁷ Perhaps those experiencing more pandemic-related distress turned to social media more often, and while the use of social media may have provided a sense of support, this use did not mitigate feelings of pandemic-related distress. It is also possible that those to who turned to social media more frequently discussed or viewed content regarding Covid-19, which could have contributed to increased distress.³⁸ Higher frequency of Instagram and Snapchat use predicted higher levels of anxiety and cumulative Covid-related stress appraisals even after controlling for age and gender; Instagram use also predicted higher levels of posttraumatic stress symptoms. These two types of social media were unrelated to feelings of social wellbeing, suggesting that they do not provide a sense of social connection. This finding is consistent with a previous study that demonstrated a direct link between social media use and distress.⁸ Collectively, the findings across these four types of social media are consistent with past research demonstrating inconsistent findings regarding relationships among social media use, social wellbeing, and mental health.⁸

Further, we found that different motives for using social media related differently to subsequent social wellbeing and mental health. Using social media to receive help (mental health appointments/substance use support group) was not positively related to mental health, but rather to greater depression, anxiety, cumulative Covid-related stress appraisals, posttraumatic stress symptoms, and loneliness. Additionally, using social media for substance abuse support related to more stress. Since we did not ask participants about their mental health history, this relationship may reflect individuals’ mental health status prior to the pandemic. However, it could also reflect dissatisfaction with online help seeking. Additional analyses that examine patterns in help-seeking during the early months of COVID-19 can be found separately.³⁹ These findings underscore the role of depression and anxiety symptoms for those seeking professional services in the first year of the pandemic.

Using social media to interact with others was generally positive: Using social media for dinner date/happy hour, exercise (with others), and to connect with friends all related to more feelings of social support. At the same time, these motives were associated with higher cumulative Covid-related stress appraisals, perhaps because those who felt higher levels of stress about the virus restricted in-person interactions more, rendering social media one of the only platforms for social interaction. Using social media for dinner date/happy hour also related to less loneliness but to more anxiety and stress, consistent with a previous study that demonstrated that people who use online dating platforms reported higher levels of depression, anxiety and distress and lower quality of life.^12,40 Further, using social media to exercise (with others) predicted higher levels of posttraumatic stress symptoms. This finding is somewhat inconsistent with a study that showed that the strategies implemented to mitigate the spread of Covid-19 adversely impact physical activity, and that those who decrease their physical activity also experience higher levels of stress and anxiety.⁴¹ It is also possible that exercising with others online is less effective than exercising with others in person.

While online gaming can be a platform for social interaction, in our sample, it predicted more depression, anxiety, and stress, which is consistent with past research that demonstrated that problematic video gaming may predict poor psychological functioning.⁴² Using social media for negative reasons (e.g., to waste time/ avoid responsibility or because of social anxiety/ discomfort) predicted less perceived social support and more loneliness, depression, anxiety, stress, cumulative Covid-related stress appraisals, and posttraumatic stress symptoms. This finding suggests that when a person is already in a negative emotional state, social media use may not improve social wellbeing or mental health. These findings are consistent with past research that indicated how social media may be beneficial or harmful depending on the motivation for use.⁹

This study had several limitations. Although social wellbeing and mental health were measured at a later time point, demonstrating temporal order, results remain correlational and do not demonstrate change in distress. Measure of social wellbeing, loneliness, and mental health were collected in May, 2020, when the country was beginning to reopen and relax social distancing measures ²⁶ Thus, it is possible the T2 data does not accurately depict people’s thoughts and feelings during social distancing. Nonetheless, the data was collected at the beginning of reopening, and the at T2, 69.0% participants reported that their state had active shelter-in-place guidelines, so it is likely that many were still experiencing social isolation. The social media measure was created for this present study and has not been validated, which is a limitation. Further, the measurement of use frequency was suboptimal; the focus on “days per week” did not allow us to gauge the number of hours per day, which may be a better indicator of intensity. On the other hand, frequency was assessed during the past month, coinciding with the widespread stay at home mandates implemented during the early stages of the Covid-19 pandemic in the US. This allowed us to examine the correlates of social media use during this critical period. We included only a limited number of social media types; including a broader range of types could reveal a broader picture. Further, several motives were reported by a small number of participants, possibly limiting generalizability. Another limitation is that anhedonia was not measured, although we did assess depression more broadly. Additionally, further examining motivation for social media use would have shed more light on why some social media use related to social wellbeing and mental health positively while others did not. One study among Chinese college students demonstrated that the average number of hours spent accessing Covid-19-related information related to higher levels of depression and anxiety.³⁸ It may be that participants accessing or discussing information regarding Covid-19 through social media contributed to decreased social wellbeing and mental health. On the other hand, if social media use served the purpose of distracting from Covid-19, it may have led to positive outcomes. This notion is only speculative as we did not gather this information, which is another limitation of this study. Finally, while MTurk workers have been demonstrated to be a reasonable national sample^{20, 21} they may report higher depressive symptoms compared to the general population²¹ which may also limit generalizability to the general U.S. population.

Despite these limitations, these results reveal interesting relationships between social media and subsequent social wellbeing and mental health. Some form of social media may relate to more positive outcomes compared to others. Social media sites where people generally tend to directly interact with others through text or video, such as Facebook and video chat apps, appear to relate more positively with social wellbeing by serving as a platform to provide and receive social support. Social media may help increase feelings of social support and decrease loneliness by providing a platform for interaction, which is consistent with studies on COVID-19 thus far that indicated that technology may help decrease the feelings of loneliness and isolation.^16,17 On the other hand, image-driven social media sites such as Snapchat and Instagram, where interactions may be less direct and intimate, related to higher levels of anxiety and no indicators of social wellbeing. These findings are consistent with a study from China early in the pandemic showing higher frequency of social media use related to higher levels of distress, including anxiety and depression.⁵

Conclusions

This study found that social media use early in the Covid-19 pandemic related to social wellbeing and mental health in complex ways. The specific type, frequency, and motive for social media use predicted different positive and negative changes in social wellbeing and mental health. Future research may explore why certain social media sites relate to more positive outcomes, and how social media use could be improved to affect wellbeing more favorably. In addition, future research may investigate how the relationship among social media use, social wellbeing, and mental health may change over time as people become increasingly accustomed to different social media platforms. Specifically, it will be important for future research to identify how using social media during highly stressful times can promote social connection and mental health in an increasingly virtual world.

Table 2:

Correlations Between Age, Gender, Race, Ethnicity and Social Media Use, Social Wellbeing, and Mental Health Outcome

	Age	Gender	Race	Ethnicity
Instagram	−.31**	.06	−.19**	−.06
Facebook	.03	.08*	−.02	.05
Snapchat	−.44**	.06	−.05	.10**
Video chat apps	−.24**	.04	−.10**	.05
Social Support	−.24**	.02	−.03	−.05
Loneliness	−.14**	.04	−.06	−.03
Depression	−.25**	.03	−.03	−.02
Anxiety	−.27**	.03	−.03	−.05
Stress	−.28**	.08*	.03	−.01
Covid-related Stress Appraisals	−.22**	.11**	.03	−.02
Postttraumatic Stress Symptoms	−.24**	.02	−.03	−.05

Note: Gender: Men=1, Women=2; Race: non-White= 1, White=2

Ethnicity: Hispanic/Latinx=1, Not Hispanic/Latinx=2

^**Correlation is significant at the 0.01 level (2-tailed)

^*Correlation is significant at the 0.05 level (2-tailed)