Skip to main content
SearchLoginLogin or Signup

Media Use and Avoidance Experiences During Social Distancing

Special Collection: Technology in a Time of Social Distancing. Volume 3, Issue 1, Spring 2022. DOI: 10.1037/tmb0000041

Published onJan 04, 2022
Media Use and Avoidance Experiences During Social Distancing
·

Abstract

Media use significantly increased in many countries as shelter-in-place and social distancing measures were enacted in response to the COVID-19 pandemic. Yet, little is known about what specific media were used; the emotional experiences users associated with media during the pandemic; or how media use may have varied as social distancing protocols changed over time. A mixed-methods study analyzed media use reports from students at two U.S. universities, gathered at the immediate onset of social distancing and again 7 months later. We quantitatively coded and analyzed the media channels and content types users reported seeking out and avoiding at each time point, and thematically analyzed the motives and affective experiences reported. Generally, users increased television viewing and computer-mediated interpersonal communication early in the pandemic, and overwhelmingly avoided the news at the onset of social distancing. In terms of affective responses, participants reported mixed experiences with social media, with some platforms associated with positive affect (TikTok, YouTube) while others were generally a source of stress (Facebook, Twitter, news content). Overall, results suggest unique platforms and content types may fulfill different functions based on the emotional needs of users during times of societal stress.

Keywords: media selection, media avoidance, COVID-19, affect, motives, uses and gratifications

Acknowledgements: Stage Two Registered Report for Technology, Mind & Behavior Special Issue on Technology-Mediated Interactions and Their Impact on the Human Mind and Behavior in a Time of Social Distancing, May 2021. The authors wish to particularly thank Alexis Adamopoulos, Melanie McIntyre, Charlie Sznewajs, and Jacob Spiess and for their assistance in data unitization and coding.

Conflict of Interest: We have no known conflict of interest to disclose.

Open Science Disclosures:
The study data and materials are available at https://osf.io/5tr4q/
The survey materials are available at https://osf.io/ktwrn/
The preregistered design is available at https://osf.io/5tr4q

Correspondence concerning this article should be addressed to Sara M. Grady, Department of Communication, Michigan State University, 404 Wilson Road, East Lansing, MI 48824, United States. Email: smgrady@msu.edu

Video summary: Media Use and Avoidance Experiences During Social Distancing


In March 2020, American universities abruptly canceled in-person instruction in response to COVID-19’s rapid spread. More than 14 million college students experienced upheaval as classes moved online and shops, bars, and restaurants closed (Hess, 2020). These sudden changes to daily routines resulted in a surge of at-home media use (Global Web Index, 2020; Perez, 2020; Weissbrot, 2020), particularly social media use (e.g., Facebook, Twitter, TikTok; Dixit et al., 2020; Wiederhold, 2020). Yet, there has been little consideration of self-aware increased media selection and avoidance during social distancing, or the affective experiences associated with this use. Media can serve many functions for users, and emotional responses are linked to both the selection and avoidance of specific genres and media platforms (Rubin, 2002). For example, users select media for entertainment, escape, stress relief, or to proactively engage with a stressor (Papacharissi & Mendelson, 2011). Therefore, the overarching research aim of the current project is to discover what media were self-reported as being used more, used less, and avoided during social distancing (RQ1); to explore the affective experiences associated with this media use (RQ2); and examine any changes in these reports between the onset of social distancing and 7 months later (RQ3 & RQ4). In all cases, we examined both the (a) types of media and (b) media content mentioned across responses.

Emotional Responses to Media

In this article, we take the approach that media use is functional, that is, people use media to satisfy their own psychological needs (Wright, 1960). Research in the tradition of the uses-and-gratifications approach (Katz et al., 1973) identified a set of basic needs users attempt to gratify via media, including diversion and entertainment, as well as information and influence. These needs are shaped by individual predispositions as well as social and sociological pressures (Katz et al., 1973). Over decades, research has established large groups or clusters of uses and gratifications, including social-integrative gratifications (such as feelings of relatedness toward media characters or other media users), cognitive or information-related gratifications (such as surveillance), self-related motives (such as exploring or displaying identity aspects via media use), as well as emotional gratifications (such as escapism; Ruggiero, 2020).

The functional affective drivers of media use are also emphasized in mood management theory (Zillmann, 1988). The theory posits that selective exposure to media content is driven by the motive of mood optimization, with media users striving to end or ameliorate unpleasant states of negative affect and arousal, and to maximize and sustain pleasant states of positive mood and optimal arousal. The central propositions of mood management theory have received substantial empirical support, suggesting that media use is frequently and successfully used for affective self-regulation in the everyday lives of media users (for an overview, see Reinecke, 2017).

In combination, the uses-and-gratifications approach and mood management theory clearly suggest that media use offers a plethora of gratifications and emotional experiences that may help media users to cope with critical life events and daily stressors. Media content may provide feelings of competence and security, by granting access to relevant information or problem-solving strategies; foster feelings of connectedness in situations where media users feel stressed or isolated; and facilitate mood regulation and distraction from stress-inducing events and cognitions. These emotional experiences may be particularly relevant to users during the uncertainty of a global pandemic (Ledbetter et al., 2016; Rochyadi-Reetz et al., 2020). At the same time, individuals may also be avoiding certain media because they are semantically similar to stressors (e.g., news about COVID or fictional depictions of pandemics) or because they induce negative emotions such as fear (e.g., dystopian thrillers or horror films).

Media for Coping

Previous work clearly underlines that media use can be an important resource and a powerful tool in the process of coping with stress (Eden et al., 2020; Wolfers & Schneider, 2020). Entertaining media use, in particular, has been linked to stress recovery both in laboratory research (e.g., Rieger et al., 2014; Reinecke et al., 2011) and in survey studies exploring the everyday lives of media users (e.g., Reinecke, 2009; Reinecke et al., 2014).

The stress-coping potential of media use is based on two main mechanisms. First, media use represents a form of emotion-focused coping, allowing media users to temporarily distract and psychologically detach themselves from sources of stress, which facilitates relaxation and the return to baseline levels of arousal (Reinecke & Eden, 2016). Second, media use provides opportunities for the acquisition and reinforcement of important coping resources, such as mastery experiences (Rieger et al., 2014) or the satisfaction of basic human needs such as competence, autonomy, and relatedness (Johnson et al., 2021; Reinecke et al., 2012). Social media in particular have been linked to psychological well-being via the accumulation of social resources such as social capital or social support (Domahidi, 2018; Meier & Reinecke, 2020).

Besides short-term effects on situational stress coping, media use may also facilitate the long-term development of psychological resilience factors that support the individual in the face of stress and adversity (Reinecke & Rieger, 2021). The coping resources provided by media use should be particularly functional in times of crisis, such as a pandemic. In fact, empirical evidence supports the notion that media use played a prominent role in the stress-coping attempts of media users during the first phase of the COVID-19 pandemic (Eden et al., 2020).

Investigating how media use has changed since the initial stay-at-home orders is also relevant to understanding media use in general, as social distancing, isolation, and periods of stress are not limited to this specific pandemic. There are many populations for whom media use may be significantly impacted by their inability to engage in other activities (e.g., housebound, chronically ill, or elderly people)—and understanding how media is selected or avoided to cope with emotional experiences may be relevant to understanding media’s broader functions. Additionally, the social distancing orders of 2020 are unlikely to be the last disruptive event to occur in our lifetimes. Understanding how young people use media during a major disruptor (such as social distancing event) and over subsequent months is therefore also relevant to understanding and predicting media use during future stay-at-home situations.

The Present Study

The central aim of the present study was to provide a more nuanced understanding of the interplay of individual media use patterns and their resulting affective experiences, mapped over the course of the pandemic and attendant stay-at-home orders. In line with most functional approaches to media use, we assume that users are able to report accurately about their own motivations for, and gratifications from, using media (Katz et al., 1973). The data are open-ended responses to prompts about media from a larger survey project (Eden et al., 2020). We, therefore, relied on a descriptive, exploratory, and longitudinal approach.

In order to examine media use and associated emotional experiences during social distancing, we initially surveyed college students in the Spring of 2020 after students transitioned to online courses (late March–late April). Students detailed the media they were (a) selecting more, (b) selecting less, and (c) avoiding during initial weeks of social distancing. Additionally, students listed media they found (d) hopeful, (e) joyful, (f) socially connective, (g) stressful, (h) depressing, and (i) guilt-inducing. These affective prompts were selected for their known relationship to coping, well-being, and affectively-motivated media use in other literature (Eden et al., 2020). Coding and quantitatively analyzing these responses provides an in-depth snapshot of media use experiences during social distancing among college students. We therefore ask: What patterns of media use (RQ1) and affective experiences associated with media use (RQ2) are reported during social distancing, for both media (a) types and (b) content genres?

We collected a second sample the following Autumn (October–November 2020) to compare patterns of media use at two time points. While initial plans were made in the hopes that a 6-month time lapse would allow for a return to normal routines and media diets, this was not the case. Only a subset of students had returned to in-person courses on campus at this stage. Therefore, we used the opportunity to compare media use at the onset of and then 7 months into the pandemic. Specifically, we were interested in whether media users were reporting relatively consistent media behaviors over the social distancing period, or if there were significant changes between these time points. While history effects and the periods of data collection prevent us from making any causal claims, the two time points provide interesting snapshots of different, and both potentially irregular, periods of media use. We therefore ask: How do patterns of media use (RQ3) and affective experiences (RQ4) vary, for both media (a) types and (b) genres reported between the onset of social distancing and 7 months later?

Method

Participants

Two separate surveys were conducted in Spring and Autumn of 2020. At both time points, the sample was drawn from two American universities (N = 822, M age = 20.23, 72% women, 78.8% White). The surveys examined stress, anxiety, media use, and well-being. Sample sizes were based on a priori power analysis for additional measures and analyses reported elsewhere (Eden et al., 2020). All participants were recruited through university student participant pools and all procedures approved by institutional review boards. The two waves of data (N spring = 425; N autumn = 397) were collected independently on both campuses, such that each wave had different participants. While we acknowledge this design has some limitations when compared with other longitudinal sampling procedures, it was necessity given the constraints of working with student populations during a tumultuous historical period. However, both waves were recruited from the same participant pools and were comparable on a range of demographic variables (see Open Science Framework [OSF] supplement).1

Measures

Nine open-ended questions were used for the current analysis. Three items asked participants about the frequency of their media use “since social distancing began,” and asked participants to list or describe media content they used more, used less, and actively avoided. Six additional items asked participants about media content that made them feel especially hopeful, joyful, connected, stressed, depressed, or guilty. For the second wave of data collection, prompts were revised to ask about media use “since the semester began.” This was done to ensure the two waves of responses were exclusive rather than overlapping time frames: In both cases, responses would pertain to the last few weeks, thereby providing snapshots of media use at two distinct points during the prolonged social distancing period.

Codebook and Coding Procedure

First, two trained coders unitized the responses. Coders noted if media was mentioned in each response (yes/no), how many media were mentioned, and isolated the first three media mentioned into discrete units (see OSF supplement for unitizing codebook). Only the first three media in each response were coded in order to avoid individual-level reporting differences between participants (95% of responses contained three or fewer media units, see OSF supplement for breakdown).

Coders then segmented all responses by a unit of thought related to specific media. A unit of thought was defined by its grammatical structure (punctuation = a thought break) and the media mentioned. Where a grammatical clause included multiple media with additional text (e.g., “I watch Netflix comedies and play video games to be with friends”), these were broken into separate media units. Each unit redacted mentions of other media but retained all other content [the example above becoming (a) “I watch Netflix comedies to be with friends” and (b) “I play video games to be with friends”]. Notably, these units would be coded only for media mentioned and the additional text would not be quantitatively analyzed.

A subset of Wave 1 data (9%) was used to calibrate and test reliability between unitizing coders. Guetzkow’s formula indicated 100% agreement on the number of units identified in the sample (Guetzkow, 1950). However, we also calculated reliability on the two quantitative codes: if media were mentioned, and how many units were present in each response (αmentionedmedia = .93; αunitspresent = .91).2 There were 360 possible responses in the reliability data set which were broken into 324 and 322 final coded units, respectively.3 There were 15 discrepancies between coders, or less than 5% of the units identified. Once consistency had been established, each coder unitized one half of the data set, resulting in 6,465 media units (3,412 in the Spring and 3,053 in the Autumn).

For each wave of data, two new coders (not involved in the unitization process) noted the medium type and content present in each media unit. Media type captured platforms and channels of media use, including broad categories (e.g., TV/film, audio, print media, social media). One media type labeled “interpersonal” was added to encompass channels that play a large role in computer-mediated interpersonal interactions (e.g., FaceTime, texting), and a media type of “Unspecified” was included to capture ambiguous media use (e.g., where users said “more news,” it is unclear if this exposure is via broadcast, print, and/or social channels; see note in Table 2). Eight popular social media channels (Auxier & Anderson, 2021) were also included by name for cross-platform comparisons. Media content was examined using a separate set of 21 codes defining common media genres (drawn from Rentfrow et al., 2011). If a specific title (e.g., a television program name) was mentioned, this was also noted, and Wikipedia’s top genre listing for that title was coded.4 A full copy of the codebook is available at: https://osf.io/5tr4q/.

Coding happened in two waves. For the Wave 1 data (Spring 2020), two coders were trained in codebook implementation with example data to assist in calibration, then furnished with a random subset of original data (10% of the sample) to code independently. A facilitator assessed reliability and convened the coders to debate points of disagreement (<12% of coded units).5 We adopted two metrics of intercoder reliability, Cohen’s κ and Krippendorff’s α (see Lombard et al., 2002). Reliability for media types (ϰ = .86, α = .86) and genres (ϰ = .85, α = .85) were calculated using ReCal (Freelon, 2013) and values were above acceptable bounds (cf. McHugh, 2012). Several weeks later, this coding process was repeated for the Wave 2 data. The same reliability set was used for consistency and the Wave 2 coders were also reliable (<5% of units adjudicated; αmedia = .96 ϰmedia = .96; αcontent = .94, ϰcontent = .94).

Thematic Analysis

An exploratory thematic analysis of the three frequency prompts (media used more, less, and avoided) was conducted to observe patterns in the subjective experiences associated with these choices. We were particularly interested in self-reported changes in media use. We adopted Braun and Clarke’s (2006) approach to inductive thematic analysis (for details, see OSF supplement). In brief, units for a given prompt (e.g., media avoided) were read closely and annotated by one of the authors, noting the emotional, experiential, and motivational aspects of the media use reported in each unit and these informed the development of a set of themes observed across responses. This process was undertaken for each of the three frequency prompts by a different one of the authors. The authors then read and reviewed each other’s units, annotated codes, and thematic interpretations to debate final language and reach consensus on the primary themes observed within each prompt, as well as to discuss common themes and similarities among the prompts. The identified themes were then interpreted within a uses and gratifications framework (see Papacharissi & Mendelson, 2011), exploring why media is used, sought, or avoided (Fahr & Böcking, 2009).

Results

Overview

A total of 6,465 media units were quantitatively coded (see Table 1, for units coded for each prompt). A list of all titles mentioned is provided in an OSF supplement. After examining the frequencies, rare or uncommon codes were collapsed.6 Media types mentioned fewer than 150 times (or 2% of the sample) were collapsed. Content codes were less common over all and thus the cutoff for genre mentions was set at 50.7 Any media type or content code which did not meet the cutoff was collapsed into a larger category based on similarity and face validity (see OSF supplement for frequency tables for initial codes and collapsed groupings). This yielded a final set of five general media types (TV/film, Social media channels, Interpersonal media channels, Other media channels, and Unspecified); seven social media codes (Facebook, Twitter, Instagram, TikTok, Snapchat, YouTube, and General/Other); and five content genres (Comedy, Drama, News/Documentary, Reality/Lifestyle, and High-arousal genres such as horror and thriller). In general, analyses compare the proportions of (a) media types or (b) genres mentioned across similar prompts (e.g., social media platforms mentioned across the frequency prompts) to look at the distribution of mentions among responses. Tables 2 and 3 show frequencies, residuals, and test statistics. For each set of tests, follow-up analyses examined differences between the two time points.

Table 1

Units of Media Reported Per Response

Wave 1

Wave 2

Media Use Prompts

0

1

2

3

4+

0

1

2

3

4+

Frequency prompts

 More

29

173

122

66

35

37

201

103

40

16

 Less

212

169

33

9

2

114

206

57

16

4

 Avoid

241

161

20

3

0

207

153

28

8

1

 Subtotal

482

503

175

78

37

358

560

188

64

21

Affective prompts

 Hopeful

121

210

70

20

3

91

220

67

15

3

 Connected

66

204

91

47

16

60

201

89

39

8

 Joyful

88

213

88

29

7

66

209

88

25

9

 Stressful

104

242

65

10

2

116

210

51

11

9

 Depressing

218

170

33

2

2

206

145

35

11

0

 Guilty

315

100

8

1

0

278

98

15

4

2

 Subtotal

912

1,139

355

109

30

817

1,083

345

105

31

Grand total

1,394

1,642

530

187

67

1,175

1,643

533

169

52

Note. During unitization, coders annotated the number of media participants mentioned in each response. The first three media mentioned were unitized for additional analysis (95% of responses mentioned three or fewer media, see OSF supplement)

What Media Types and Genres Were Approached and Avoided?

To address RQ1a, we examined the types of media reported in response to the frequency prompts (more, less, and avoid) using two different chi-square tests (see Table 2). The first test compared broad categories of media, the second compared social media platforms specifically. A 3 × 5 chi-square revealed that participants reported more frequent television/film viewing and interpersonal media channel use during social distancing (adj. resid. = 11.3 and 3.1, respectively), Pearson’s χ2(8) = 399.51, p < .001, Cramer’s V = .31.

Table 2

Media Types Mentioned Across Response Units

General media types

Social media types

Media Use Prompts

TV/F

Interp.

Social

Unsp.

Other

Total

SocO.

Fb.

Twit.

Inst

Snap

TikT.

YouT.

Total

Frequency prompts

 More

467

37

495

20

49

1,068

32

31

51

81

35

200

65

495

(11.3)

(3.1)

(−6.1)

(−8.9)

(−2.1)

(−1.1)

(−7.4)

(−5.6)

(−3.5)

(−2.7)

(14.4)

(5.2)

 Less

178

11

389

17

46

641

28

81

79

106

55

13

27

389

(−3.0)

(−1.4)

(4.7)

(−4.9)

(2.0)

(−0.2)

(3.8)

(1.9)

(3.6)

(3.6)

(−10.4)

(−1.2)

 Avoid

46

4

247

105

26

428

24

60

67

53

20

21

2

247

(−10.7)

(−2.3)

(2.2)

(16.6)

(0.4)

(1.6)

(4.5)

(4.5)

(0.1)

(−1.0)

(−5.3)

(−4.8)

 Total

691

52

1,131

142

121

2,137

84

172

197

240

110

234

94

1,131

χ2(8) = 399.51, p < .001, Cramer’s V = .31

χ2(12) = 302.48, p < .001, Cramer’s V = .37

Affective prompts

 Hopeful

88

35

568

48

53

792

37

40

83

119

22

189

78

568

(4.5)

(−5.0)

(1.2)

(−2.1)

(1.3)

(−0.6)

(−3.3)

(−2.6)

(−1.0)

(−6.5)

(6.9)

(7.1)

 Connected

10

286

741

9

37

1,083

67

61

115

173

191

115

19

741

(−9.4)

(23.0)

(−1.2)

(−10.0)

(−3.8)

(2.4)

(−2.7)

(−2.3)

(0.7)

(13.6)

(−5.2)

(−5.4)

 Joyful

126

39

711

18

69

963

36

28

82

125

59

282

99

711

(7.7)

(−6.1)

(3.0)

(−7.9)

(2.2)

(−2.4)

(−6.9)

(−5.4)

(−3.6)

(−3.3)

(12.6)

(8.4)

 Stressed

51

22

494

141

48

756

34

115

156

119

38

26

6

494

(−0.7)

(−6.5)

(−3.0)

(12.0)

(0.8)

(−0.2)

(9.6)

(8.3)

(0.9)

(−3.1)

(−10.0)

(−5.5)

 Depressed

31

9

319

97

27

483

18

70

87

92

30

19

3

319

(−0.8)

(−5.8)

(−2.0)

(10.5)

(−0.1)

(−1.1)

(6.6)

(4.3)

(2.9)

(−1.4)

(−7.4)

(−4.5)

 Guilty

13

1

193

30

14

251

23

18

33

52

16

46

5

193

(−1.4)

(−4.9)

(2.5)

(2.4)

(−0.1)

(2.7)

(−0.8)

(−0.5)

(1.5)

(−1.5)

(0.5)

(−2.5)

 Total

319

392

3,026

343

248

4,328

215

332

556

680

356

677

210

3,026

χ2(20) = 951.78, p < .001, Cramer’s V = .23

χ2(30) = 836.21, p < .001, Cramer’s V = .24

Note. This table shows four tests examining the media types mentioned by participants. Adjusted residuals are in parentheses. We examined general media types (left) and social media specifically (right) for a series of prompts about the frequency of media use (top) and affective states associated with media use (bottom). The Social code in the general media types are the same units in the specifically social portion, broken down by channel. Interp. = Interpersonal channels; Unsp. = unspecified type (e.g., channel was ambiguous or unclear). Units where the type was Unspecified were overrepresented in the avoided media responses. Examining a contingency table of Media type × Media genre codes revealed that these unspecified types were overwhelmingly categorized in the news genre (338 of 385 units; see OSF supplement). Coders were unable to tell if respondents naming “news” were referring to a type code of television, radio, or print, etc. SocO. = Other social media; Fb. = Facebook; Twit. = Twitter; Inst. = Instagram; Snap. = Snapchat; TikT. = TikTok; YouT. = YouTube.

A second 3 × 7 chi-square test exclusively examined the distribution of social media mentions (n = 1,131). While TikTok and YouTube were used more during social distancing (adj. resid. = 14.4 and 5.2, respectively), Facebook was often avoided (adj. resid. = 4.5). By contrast, Instagram and Snapchat were used less (adj. resid. = 3.6 for both) but not particularly avoided (adj. resid. = 0.1 and −1.0, respectively), Pearson’s χ2(12) = 302.48, p < .001, Cramer’s V = .37.

To examine this cross-platform difference more closely, we looked at the two most commonly mentioned social platforms, Instagram and TikTok. A Fisher’s Exact Test (More/Less/Avoid × Instagram/TikTok) demonstrated that although these platforms were mentioned more than any other social media, users were reporting significantly different patterns of use and avoidance (p < .001). TikTok was being used more during social distancing (adj. resid. = 11.5) and Instagram was being used less (adj. resid. = 9.7) or being avoided altogether (adj. resid. = 3.9).

To examine RQ1b, we tested the prevalence of different media content across frequency prompts with a 3 × 5 chi-square (more/less/avoid prompts × genres, Table 3). Some genres were disproportionately sought out while others were avoided, Pearson’s χ2(8) = 290.39, p < .001, Cramer’s V = .51. Since social distancing began, comedies and dramas were being watched more (adj. resid. = 7.1 and 6.9, respectively) while the news was extensively avoided (adj. resid. = 16.1).

Table 3

Media Genres Mentioned Across Response Units

Genres

Media Use Prompts

High arousal

Comedy & family

News & Doc

Drama

Reality, sport, life

Total

Frequency prompts

 More

47

96

40

106

37

326

(1.7)

(7.1)

(−14.4)

(6.9)

(2.5)

 Less

17

12

33

14

11

87

(2.2)

(−1.4)

(0.2)

(−1.5)

(1.4)

 Avoid

5

0

133

4

1

143

(−3.8)

(−6.8)

(16.1)

(−6.5)

(−4.0)

 Total

69

108

206

124

49

556

χ2(8) = 290.39, p < .001, Cramer’s V = .51

Affective prompts

 Hopeful

9

27

30

5

26

97

(2.7)

(6.6)

(−10.7)

(2.2)

(6.4)

 Connected

1

3

4

0

3

11

(0.8)

(2.0)

(−2.9)

(−0.5)

(2.0)

 Joyful

7

23

12

2

18

62

(2.9)

(7.7)

(−10.5)

(0.6)

(5.6)

 Stressed

5

0

206

3

0

214

(−1.8)

(−6.2)

(9.5)

(−1.1)

(−6.1)

 Depressed

1

0

126

2

2

131

(−2.3)

(−4.4)

(6.6)

(−0.6)

(−3.6)

 Guilty

0

0

21

0

2

23

(−1.0)

(−1.6)

(1.9)

(−0.7)

(−0.1)

 Total

23

53

399

12

51

538

χ2(20) = 307.64, p < .001, Cramer’s V = .38

Note. Adjusted residuals in parentheses. A single unit could have both a type and genre code (e.g., a unit mentioning the show Friends would be coded as television in Table 2 and comedy here).

Are These Patterns of Approach and Avoidance Consistent Over Time?

Follow-up analyses compared media that was used more, less, and avoided, over time. We examined the number of units and frequency of media mentions across waves, as well as the media types (RQ3a) and content mentioned (RQ3b) in Spring 2020 (t1) versus Autumn 2020 (t2).

First, early in the pandemic, compared to later, people were writing more about their media use overall. A 2 × 4 chi-square (Time × Units coded) revealed that more responses included multiple media mentions at t1 than t2, χ2(3) = 17.63, p < .001, Cramer’s V = .05. A separate 2 × 3 chi-square (Time × Frequency prompts) compared the number of media units that were reported as being used more, less, or avoided. The Spring cohort reported almost twice as many units of media being used more (adj. resid. = 10.8) than their counterparts 7 months later. Similarly, the Autumn cohort listed significantly more units of media being used less (adj. resid. = 8.3), suggesting that general media use went up in the initial stages of the pandemic but may have decreased over time, χ2(2) = 118.93, p < .001, Cramer’s V = .24.

We next examined changes over time within each media type, using individual Time × Prompt chi-squares (RQ3a). This revealed how often each type of media was mentioned across frequency prompts at the two time points, in order to examine if patterns of use and avoidance varied over the social distancing period. TV and film were mentioned more often overall at t1, and early in the pandemic were particularly mentioned as media being used more (adj. resid. = 6.4), χ2(2) = 44.00, p< .001, Cramer’s V = .25. A similar pattern was observed for Interpersonal media (see Table 4). Looking specifically at social media platforms, Instagram was listed more often overall at t2, and was reported as proportionally being used more at t1 (adj. resid. = 5.4) but avoided at t2 (adj. resid. = 2.1), χ2(2) = 29.22, p < .001, Cramer’s V = .35. By contrast, the pattern of TikTok use remained broadly similar at both time points, χ2(2) = 3.15, p = .21, Cramer’s V = .17, see Figure 1 and Table 4 for details.

Figure 1

Social Media Being Used More/Less Frequently at the Onset of Social Distancing (Wave 1) and 7 Months Later (Wave 2)
Note. Social media channels reported as being used more, less, or avoided in Spring versus Autumn 2020. Notably, Facebook, Twitter, and Instagram use change significantly, while TikTok and YouTube are being reported similarly at both time points (see Table 4 for frequencies and tests).

Table 4

Mentions of Each Media Type Distributed Over Frequency Responses and Over Time

Wave

General media types

1

2

Total count

Chi-square test

TV/Film

More

*326

*141

467

Less

*74

*104

178

Avoid

26

20

46

Total

426

265

691

χ2(2) = 44.00, p < .001, Cramer’s V = .25

All other

More

*37

*12

49

Less

32

14

46

Avoid

*6

*20

26

Total

75

46

121

χ2(2) = 21.63, p < .001, Cramer’s V = .42

Social media (all)

More

*311

*184

495

Less

*146

*243

389

Avoid

*90

*157

247

Total

547

584

1,131

χ2(2) = 73.81, p < .001, Cramer’s V = .26

Interpersonal

More

26

11

37

Less

6

5

11

Avoid

2

2

4

Total

34

18

52

χ2(2) = 1.38, p = .50, Cramer’s V = .16

Unspecified

More

16

4

20

Less

10

7

17

Avoid

76

29

105

Total

102

40

142

χ2(2) = 2.10, p = .35, Cramer’s V = .12

All types

More

*716

*352

1,068

Less

*268

*373

641

Avoid

*200

*228

428

Total

1,184

953

2,137

χ2(2) = 119.52, p < .001, Cramer’s V = .24

Social media types

1

2

Total count

Chi-square test

Social other

More

*25

*7

32

Less

10

18

28

Avoid

*7

*17

24

Total

42

42

84

χ2(2) = 16.58, p < .001, Cramer’s V =. 44

Facebook

More

*24

7

31

Less

*31

*50

81

Avoid

31

29

60

Total

86

86

172

χ2(2) = 13.85, p < .001, Cramer’s V = .28

Twitter

More

*35

*16

51

Less

32

47

79

Avoid

*22

*45

67

Total

89

108

197

χ2(2) = 16.14, p < .001, Cramer’s V = .29

Instagram

More

*54

*27

81

Less

*32

*74

106

Avoid

*16

*37

53

Total

102

138

240

χ2(2) = 29.22, p < .001, Cramer’s V = .35

Snapchat

More

*21

*14

35

Less

21

34

55

Avoid

5

15

20

Total

47

63

110

χ2(2) = 7.30, p = .03, Cramer’s V = .26

TikTok

More

114

86

200

Less

6

7

13

Avoid

8

13

21

Total

128

106

234

χ2(2) = 3.15, p = .21, Cramer’s V = .17

YouTube

More

38

27

65

Less

14

13

27

Avoid

1

1

2

Total

53

41

94

χ2(2) = .37, p = .83, Cramer’s V = .06

Note. * denotes adjusted residual > 2 or < −2, expanded table with residuals available on OSF supplement.

Since participants reported less TV/film media at t2 overall, it was not surprising that the number of units that could be coded for a content genre was also significantly reduced (725 genres coded at t1, compared to 369 at t2). To address RQ3b, a 2 × 5 chi-square (Time × Genre) showed that news was overwhelmingly present in the first wave of responses (adj resid. = 5.0) while reality/sport content played a larger role in the t2 responses (adj. resid. = 3.6; χ2(4) = 29.86, p < .001, Cramer’s V = .17). A series of 3 × 2 chi-squares (More/less/avoid × Time) examined each genre individually, to explore which media were approached and avoided across the two time points (see Table 5). Notably, there were more than twice as many mentions of news content at t1 than t2, despite the fact the Autumn data collection spanned a U.S. presidential election season. Whether this difference is the result of decreased vigilance or perceived risk related to COVID-19 news (the t1 mentions were a spike in news awareness perhaps), or whether participants were returning to prior news-consuming behaviors is unclear, but the decrease in news mentions suggests news viewing shifted substantially over the social distancing period.

Table 5

Mentions of Each Genre Distributed Over Frequency Responses and Over Time

Wave

Genres

1

2

Total count

Chi-square test

High arousal

More

33

14

47

Less

14

3

17

Avoid

2

3

5

Total

49

20

69

χ2(2) = 3.41, p = .18, Cramer’s V = .22

Comedy & family

More

58

38

96

Less

4

8

12

Avoid

0

0

0

Total

62

46

108

χ2(1) = 3.20, p = .07, Cramer’s V = .17

News & documentary

More

*38

*2

40

Less

*17

*16

33

Avoid

92

41

133

Total

147

59

206

χ2(2) = 17.61, p < .001, Cramer’s V = .29

Drama

More

*72

*34

106

Less

*5

*9

14

Avoid

*0

*4

4

Total

77

47

124

χ2(2) = 12.22, p < .01, Cramer’s V = .31

Reality, sport, lifestyle

More

24

13

37

Less

8

3

11

Avoid

0

1

1

Total

32

17

49

χ2(2) = 2.15, p = .34, Cramer’s V = .21

All genres

More

225

101

326

Less

*48

*39

87

Avoid

94

49

143

Total

367

189

556

χ2(2) = 5.87, p = .05, Cramer’s V = .10

Note. * denotes an adjusted residual of >2 or <−2, expanded table with all residuals is available on OSF supplement. Test statistics are not directly comparable, for example no comedies were mentioned in the Avoid responses, resulting in a 2 × 1 chi-square test.

Thematic Analysis of Approach and Avoidance

To probe these patterns further, we turned to our qualitative examination of the response text, both in the whole dataset and at each time point. During the quantitative coding process, the coders noted if responses included additional details beyond the type of media, which we assumed would shed light on the users’ motives, reasons, or the outcomes of their reported media use. A total of 527 units (out of 2,137 more/less/avoid units) included this subjective detail (intercoder reliability on this item ϰ = .81, α = .81). Responses about media being used more frequently (n = 216), less frequently (n = 152), or avoided (n = 159) were analyzed independently, and then identified themes were discussed and compared across these prompts. We report on the five major themes observed (managing emotions, surveillance, social connection, social comparison, and external constraints) below. Additional details can be found on OSF supplement.

First, in response to all three prompts, we saw clear evidence of users managing emotions through media use. Distraction, avoiding other tasks, hedonic pleasure, relaxation, and escapism were all mentioned as key factors in increasing media use. For example, M6298 wrote: “I have been watching movies on Netflix more frequently to help me relax.” And M386 wrote: “I’ve been using TikTok more frequently. I find that the comedic relief is helpful to get my mind off things.” Participants also reported avoiding content which they expected to be distressing or overwhelming, such as fictional depictions of catastrophes. Participant L282 wrote: “Ive[sic] been avoiding horror films that purposely stress out viewers or are intended to illustrate catastrophic situations.”

In terms of negative emotions, news (via mass and social media) was the most frequently mentioned source of negative affect. For example, A91: “I have actively avoided watching the news because all it is talking about is COVID-19 and not shedding any positive light on it and I am already struggling to keep up in school so its extra added negativity that I do not need or want.” Participants also voiced anger and frustration at misinformation, polarization, conflict, distrust, and media bias, leading to lessened use or avoidance. For example, A420: “I have avoided political news sources as I do not wish to see this pandemic spun in a way to support or contest political beliefs.” And A229: “[I have avoided] TV news because it’s full of fearmongering and lies.” However, participants did note that the negativity of the news has made them try to balance needs for surveillance with their emotional state. For example, A704: “Since the semester began, I only check the news occasionally on digital platforms that concern Covid-19 statistics. I still make sure I am informed but try my best not to consume too much information that may affect my mental health.”

Next, using media for social connection was a common theme. A sample response comes from M50, “I would say that I have been using Facetime more than usual to talk to my family and friends.” Early in the pandemic, participants also mentioned viewing films or series together with friends or family or using media to connect with people, for example, by following recommendations from friends. For example, M205 “I’ve also been watching a lot of movies with my family” and M91 “I have been watching the show All American non-stop due to many recommendations of friends and family watching it during quarantine.” Similarly, participants noted rewatching films or series for nostalgia, for example M686: “I’ve also been rewatching some of my favorite classic TV shows (like Friends, The Office) because watching something I’m familiar with helps me feel more familiar and comforts me.” Participants also reported using media (particularly YouTube and TikTok) for background noise.

Related to social connection, social comparison also stood out as a strong central theme in the responses. Participants assessed their experience relative to what other people were doing, which they viewed via social media or through mediated depictions of others. A460: “I try to avoid Instagram to not compare myself to others.” Social media use (in particular Instagram, TikTok, Snapchat, and Facebook) was associated with a mix of emotions from envy to fear-of-missing-out (FOMO) and even social judgment of others’ behaviors. For example, A624: “I’ve been trying to stay off instagram because seeing people on vacation makes me upset.” Many participants were bothered by seeing what others were doing online, such as violating lockdown guidelines, and consumed less or avoided social media in response. Participants also reported lessened media use due to being annoyed at limited content available on social media, or limited content they could post themselves. For example, L580: “I haven’t used snap chat as much or posting stories since I haven’t been doing much I feel I don’t have as many things to post about as I did before.”

Finally, participants were very aware of external constraints that had changed their media use patterns. For example, participants noted having more free time, less free time, or more/less need for specific tools (such as course management software or Uber). For example, M780 “Having nothing to do at home has increased my overall media usage” was a very typical response to the “more” prompt. There were also many responses referencing the need to regulate media use during unstructured time. Participants described quitting media, imposing rules on their usage, or changing media habits, typically for mental health reasons. For example, L631 “Although I use … Instagram … less now because I’ve put limits on my phone apps. I did this because I wasn’t focusing on my homework and work enough.” Self-regulation concerns highlighted a need for self-care and self-protection, as well as a heightened awareness of the role of selectivity in self-regulation. For example, A712: “I haven’t actively avoided any media, but I have put more time into media that connects me with world events and less with media that involves pictures or more meaningless material that I have been less interested in.” This self-regulation is responsive to external constraints (awareness of demands on their time and well-being) but also social comparison and emotion (awareness of how unrestricted social media use made them feel in the past).

Thematic Analysis Over Time

Thematic analysis of these in-depth responses was examined both holistically (all together) as well as by survey time point (t1 and t2). Comparing the responses over time, less surveillance and information-seeking motivations were present at t2 than t1. For example, A518: “I don’t watch the news anymore ever. Today is the morning of November 4th, so I am watching election results, but I have not watched any news since coming to school until Election Day.” Additionally, we observed less talk about moving to new environments or the need to connect with coviewers at t2. Greater feelings of being overwhelmed were reported at t1, compared to fatigue at t2. For example, L467 “Since the semester has started I have not watched Netflix shows as often, because I don’t have the time like I used too[sic].” Most prevalent at t2, compared to t1, was an awareness of one’s own media use patterns as well as failure to regulate this usage. Participants reported wanting to limit themselves, and failing to do so; or labeled prior media use or habits as unhealthy. For example, A418: “I have deleted my facebook app to try and limit the time I spend on social media. I do not think that social media is an effective or healthy use of my time, but it is unfortunately what I resort to when I am bored.”

Which Media Types and Genres Were Associated With Specific Affective States?

The final set of analyses examined the media types and genres associated with the specific emotional prompts in the survey (hope, joy, human connection, stress, depression, and guilt). These chi-square analyses would parallel those reported above, but examine responses to the affective prompts, first looking at the (a) media types and (b) genres mentioned across the whole data set (RQ2), and then compared across two points in time (RQ4).

For RQ2a, we examined media types mentioned in all responses to the affective prompts (5 types × 6 prompts) Pearson’s χ2(20) = 951.78, p < .001, Cramer’s V = .23 (see bottom of Table 2). TV/film viewing was disproportionately associated with hopeful and joyful media experiences over the social distancing period (adj. resids. = 4.5 and 7.7, respectively), while interpersonal channels were particularly associated with feelings of connection (adj. resid. = 23.0). Examining social media as a broad category (n = 3,026 units) shows that these media were generally associated with joyful (adj. resid. = 3.0) and guilty (adj. resid. = 2.5) experiences.

We next examined individual social media platforms (7 social channels—Facebook, Twitter, Instagram, Snapchat, TikTok, YouTube, Other— × 6 affective prompts; see Table 2); χ2(5) = 836.21, p < .001, Cramer’s V = .24. TikTok and YouTube were both associated with hopeful and joyful media experiences (adj. resid.TikTok = 6.9 and 12.6, adj. resid.YouTube = 7.1 and 8.4). Snapchat was associated with feelings of connection (adj. resid. = 13.6). By contrast, Facebook—and to a lesser extent Twitter—were more often associated with stressful and depressing media use (adj. resid.Facebook = 9.6 and 6.6; adj. resid.Twitter = 8.3 and 4.3).

Focusing once again on the two most commonly listed social media platforms, Instagram and TikTok, revealed that while Instagram was more likely to be mentioned as stressful or depressing (compared to TikTok), it was also more likely to be mentioned as a source of connection with other people, χ2(5) = 196.17, p < .001, Cramer’s V = .38. A Fisher’s Exact Test collapsed the six affective prompts into two factors: Positive (hopeful/joyful/connected) versus negative (stress/depressed/guilty) states and demonstrated that TikTok use was generally associated with positive affective states (adj. resid.Tikok = 10.6; adj. resid.Insta = −10.6) while, overall, Instagram use was the more likely of the two to be discussed in response to the negatively valanced prompts (p < .001).

For RQ2b, we examined the content genres mentioned in response to the affective prompts using a 5 × 6 chi-square test, χ2(20) = 307.64, p < .001, Cramer’s V = .38. As observed in the thematic analysis, the news was disproportionately associated with negative media experiences (see Table 3, for genre frequencies, residuals, and test statistics).

Are Media Consistently Associated With These Affective States Over Time?

Examining responses to the six affective prompts over time (RQ4) shows the Spring and Autumn cohorts reported their media use in both similar and different ways at these two time points. First, the number of media mentioned in association with these affective states was broadly similar in fall and spring: A 2 × 6 chi-square (Time × Affective prompts) test was nonsignificant, χ2(5) = 7.21, p = .21, Cramer’s V = .04. However, examining frequencies revealed that interpersonal media and unspecified (news) media mentions both went down over time, while social media mentions increased overall (frequencies and residuals in Table 6).

Table 6

Mentions of a Media Type Distributed Over Affective Responses and Over Time

Wave

General media types

1

2

Total count

Chi-square test

Social media (all)

Hopeful

*287

*281

568

Connected

355

386

741

Joyful

339

372

711

Stressed

221

273

494

Depressed

132

187

319

Guilty

*72

*121

193

Total

1,406

1,620

3,026

χ2(5) = 15.23, p < .01, Cramer’s V = .07

Interpersonal

Hopeful

26

9

35

Connected

180

106

286

Joyful

29

10

39

Stressed

*8

*14

22

Depressed

6

3

9

Guilty

0

1

1

Total

249

143

392

χ2(5) = 12.55, p = .03, Cramer’s V = .18

TV/Film

Total

168

151

319

χ2(5) = 3.36, p = .65, Cramer’s V = .10

Unspecified

Total

276

67

343

χ2(5) = 7.50, p = .21, Cramer’s V = .15

All other

Total

129

119

248

χ2(5) = 13.36, p = .02, Cramer’s V = .23

All types combined

Total

2,228

2,100

4,328

χ2(5) = 7.17, p = .21, Cramer’s V = .04

Wave

Social media types

1

2

Total count

Chi-square test

Facebook

Hopeful

30

10

40

Connected

39

22

61

Joyful

17

11

28

Stressed

75

40

115

Depressed

42

28

70

Guilty

13

5

18

Total

216

116

332

χ2(5) = 3.20, p = .67, Cramer’s V = .10

Twitter

Hopeful

*56

*27

83

Connected

64

51

115

Joyful

40

42

82

Stressed

77

79

156

Depressed

43

44

87

Guilty

16

17

33

Total

296

260

556

χ2(5) = 9.43, p = .09, Cramer’s V = .13

Instagram

Hopeful

*70

*49

119

Connected

81

92

173

Joyful

63

62

125

Stressed

*35

*84

119

Depressed

*26

*66

92

Guilty

18

34

52

Total

293

387

680

χ2(5) = 34.57, p < .001, Cramer’s V = .23

TikTok

Hopeful

73

116

189

Connected

41

74

115

Joyful

*128

*154

282

Stressed

*5

*21

26

Depressed

*3

*16

19

Guilty

*11

*35

46

Total

261

416

677

χ2(5) = 18.39, p = .002, Cramer’s V = .17

Snapchat

Total

156

200

356

χ2(5) = 8.03, p = .16, Cramer’s V = .15

YouTube

Total

93

117

210

χ2(5) = 1.10, p = .95, Cramer’s V = .07

Social other

Total

91

124

215

χ2(5) = 1.80, p = .88, Cramer’s V = .09

Note. * denotes adjusted residual >2 or <−2, expanded table with frequencies and residuals for all tests are available in the OSF supplement.

Once again, we used a series of Time × Prompt chi-squares to see how mentions of a particular type of media were distributed across the affective prompts, and if these proportions were similar at both time points (RQ4a). Notably, interpersonal channels of communication were mentioned fewer times at t2 (the Autumn group), and more people experienced these media as stressful compared to t1 (adj. resid. = 2.7), χ2(5) = 12.55, p = .03, Cramer’s V = .18.

Among social media channels, TikTok and Instagram mentions increased and Facebook mentions decreased at t2 (see Figure 2 and Table 6). Examining Instagram mentions among the six affective prompts reveals that channel use was associated with more hope at t1 (adj. resid. = 3.8) but more stress and depression at t2 (adj. resids. = 3.3 and 3.1, respectively), χ2(5) = 34.57, Cramer’s V = .23, p < .001. Similarly, TikTok is overrepresented in joy responses at t1 (adj resid. = 3.1), and increasingly associated with negative affect later in the pandemic, χ2(5) = 18.39, p = .002, Cramer’s V = .17. By contrast, Facebook was mentioned about half as many times in the Autumn (t2) as in the Spring (t1), yet the distribution of mentions across affective prompts was similar at both time points. This suggests that how users felt about Facebook remained broadly similar, as evidenced by a nonsignificant chi-square distribution of mentions across time: χ2(5) = 3.20, p = .67, Cramer’s V = .10. Taken together, this demonstrates that while users reported using less media overall as the pandemic wore on, some types of media were increasingly associated with negative affect (see Table 6).

Figure 2

Social Media Channels That “Made You Feel … [Emotion]” at the Onset of Social Distancing and 7 Months Later
Note. Social media channels associated with various emotional states associated with coping in Spring versus Autumn 2020. Instagram and Twitter in particular are considered more hopeful early on in the pandemic, but increasingly associated with negative affect as social distancing wore on (see Table 6 and OSF supplement, for frequencies and tests).

We also examined how each genre was distributed across affective responses between Spring and Autumn collections, shedding light on how these genres were perceived and experienced by both groups (RQ4b, see Table 7). Overall, the news was mentioned 399 times in the affective prompts—and the vast majority appeared at t1 (n = 293). However, although the total number of mentions decreased over time, the way these mentions were distributed among the affective prompts were broadly similar, suggesting that the ways news made people feel was similar in Spring and Autumn, χ2(5) = 6.71, p = .24, Cramer’s V = .13. Comedies were also mentioned more in the Spring than Autumn and consistently associated with hopeful and joyful media experiences. Several prompts had zero comedy mentions (e.g., no one mentioned comedies in the guilty or depressing prompts at either time point). A similar pattern arose with the Reality genres, where the vast majority of mentions were hopeful and joyful at both time points, and several cells had frequencies of zero. Therefore, cell sizes were insufficient for a formal test comparing these genre over time, but see Table 7 for frequencies and details.

Table 7

Mentions of a Genre Distributed Over Affective Responses and Over Time

Wave

1

2

Total count

Chi-square test

Comedy & family

Hopeful

16

11

27

Connected

3

0

3

Joyful

14

9

23

Stressed

0

0

0

Depressed

0

0

0

Guilty

0

0

0

Total

33

20

53

χ2(2) = 1.94, p = .38, Cramer’s V = .19

News & documentary

Hopeful

20

10

30

Connected

3

1

4

Joyful

10

2

12

Stressed

148

58

206

Depressed

92

34

126

Guilty

*20

*1

21

Total

293

106

399

χ2(5) = 6.71, p = .24, Cramer’s V = .13

High arousal

Total

10

13

23

χ2(4) = 4.63, p = .33, Cramer’s V = .45

Drama

Total

4

8

12

χ2(3) = 4.35, p = .23, Cramer’s V = .60

Reality, sport, lifestyle

Total

18

33

51

χ2(4) = 3.98, p = .41, Cramer’s V = .28

All genres combined

Hopeful

*54

*43

97

Connected

7

4

11

Joyful

*32

*30

62

Stressed

152

62

214

Depressed

93

38

131

Guilty

*20

*3

23

Total

358

180

538

χ2(5) = 18.80, p = .002, Cramer’s V = .19

Note. * denotes adjusted residual >2 or <−2, expanded table with frequencies and residuals for all tests are available in the OSF supplement.

Discussion

The central aim of the present study was to provide a more nuanced understanding of the interplay of individual media use patterns and their resulting affective experiences over the course of the pandemic and its attendant stay-at-home orders. What we discovered, via an analysis of open-ended prompts asking about student media use during the pandemic, was that users were quite aware that they were using media to manage emotional states and needs. While the types of needs and motivations seemed to be consistent across the responses, the use of media to satisfy these needs and motivations changed over time. Both television and social media were used to help cope with negative emotions, to bring about positive emotions, or distract from overwhelming circumstances. Media offering these emotional gratifications were eagerly sought out, especially early in lockdown. Many people increased their television viewing and computer-mediated interpersonal communication use during social distancing, while news exposure was generally avoided. More interestingly, many different social media platforms were mentioned in responses, but the frequency of use varied widely between them. For example, TikTok was being used more while Facebook was being avoided. Taken together, responses about the frequency of use and the affective states associated paint a vivid picture of how students were using media to cope. For example, Facebook was avoided in the frequency prompts, and was associated with stressful and depressing media use in the affective prompts—suggest the channel was eliciting negative emotions and therefore users managed their mood by avoiding Facebook, while by contrast, the positive mood boost TikTok was affording led to increased use.

The media associated with specific affective responses provided additional insights. Some media were highly popular (television and TikTok) or consistently associated with negative affect (news). However, comparing channels, not all social media were created equal. For example, TikTok was generally associated with positive affect while Facebook and Twitter produced many negative emotional experiences, which was highlighted in the thematic analysis to be commonly related to politicization and interpersonal conflict. Likewise, TikTok and Instagram were mentioned at similar rates and have comparable platform interfaces, yet yielded totally different affective reports and mentions over time. This illustrates the complex ways in which social media offerings may function differently for users, despite their surface similarities.

Our thematic analysis uncovered additional details and nuance that the quantitative codes did not. For example, using media for social connection was highlighted in the thematic analysis, as were themes of social comparison and even envy. The negative perceptions of news media manifested in many ways (e.g., misinformation, politicization, social conflict, semantic affinity). Overall, this qualitative examination was largely consistent with the findings of our quantitative analysis, but also provided rich insights into how media was used to cope with external demands and in response to changing environmental forces. It highlighted the intense emotional experiences media could provoke, and how media were being used to manage feelings in difficult times, as both a source of social community and by providing entertaining, positive, and uplifting experiences. It also highlighted how media use is consciously regulated and often in response to shifting external demands.

Examining these reports over time revealed a shift between Spring and Autumn 2020, suggesting media experiences changed in several ways over the course of the pandemic. Qualitative and quantitative analyses suggest that several types of media use increased early on during lockdown and ebbed later in the year, in response to both external constraints and internal needs. Notably, a recent analysis of pandemic porn use (Grubbs et al., 2021) observed similar patterns of dramatic change followed by a readjustment period. However, our data cannot clarify if these shifts were a return to baseline patterns of media use, or additional coping strategies were adopted as users were confronted with a new normal. For example, participants frequently mentioned work and educational routines as reasons for changing their media use in the latter part of the year—yet it is unclear if this matches prepandemic academic terms, or if the hybrid campus culture of late 2020 was uniquely impacting media use. Similarly, over these 7 months, TikTok use went up, but whether this was indicative of the general diffusion of a new platform (young people are usually earlier adopters of new media, including TikTok; Auxier & Anderson, 2021) or the result of unprecedent amounts of time at home with less access to other forms of leisure and diversion, remains uncertain.

Overall, the present findings clearly underline the crucial role of media use as a coping resource in times of crisis (Eden et al., 2020; Wolfers & Schneider, 2020). The patterns of results observed here reveal that media use reflected several different coping strategies during social distancing. For many respondents, media use represented a form of emotion-focused coping (Lazarus & Folkman, 1984). Media that were perceived to increase pandemic stress (particularly news media) or present additional strain (e.g., social comparison in social media) were being used less or even deliberately avoided, whereas media providing stress relief and mood optimization (particularly TV and movies, but also some social media channels such as TikTok) were actively and increasingly selected by many users.

The current study provides in-depth insights into the emotion-regulation strategies of media users, supporting the central tenets of mood management theory (Zillmann, 1988) and providing compelling evidence of the importance of media use for mood optimization particularly in difficult life phases (e.g., Anderson et al., 1996). While media users’ attempts to distract themselves from frustration, adversity, and stress have often been discussed critically as a form of dysfunctional coping (e.g., Meier et al., 2018), the results of the present study support more recent work that views escapist media use as a functional coping strategy (Halfmann & Reinecke, 2021). Self-management and regulation through media may be particularly relevant to stressful situations where individuals have little control over the source of stress (as is the case in a pandemic), and therefore problem-focused strategies of coping are perceived as less viable alternatives.

We did, however, observe some problem-focused coping (Lazarus & Folkman, 1984) in our data as well. Most strikingly, participants reported using media to compensate for feelings of loneliness and social isolation, either by co-using media in the presence of others or by using social media to connect with absent friends or family. This supports previous findings demonstrating that social media in particular can be a significant source of social resources (e.g., Domahidi, 2018; Meier & Reinecke, 2020). Notably, other media use patterns found in our thematic analyses seem to represent forms of meaning-focused coping (Folkman & Moskowitz, 2007), for instance when media users report seeking nostalgic media experiences that connect them to a happier past and help to generate hope and self-assurance.

Besides the potential to support coping processes, our data also clearly reveal central challenges that arise from media use during the pandemic. First, particularly in the context of news watching, many respondents reported a struggle to find a balance between their needs for emotion regulation on the one hand and their information and surveillance motives on the other. In the pandemic, news media represented a source of conflicting experience for many users. While they provided important information on the health threat originating from COVID-19 and potential protection strategies, the high level of negativity associated with the news was simultaneously perceived as a source of anxiety and stress. The challenges facing media users striving to find a balance between conflicting gratifications or between emotion- and problem-focused coping styles has not been addressed systematically, for example in previous work on media use and coping (Wolfers & Schneider, 2020). The data reported here suggest media offerings are perceived to provide many affective experiences, and how and when these perceived uses and gratifications conflict may be an important avenue of inquiry in navigating suboptimal or inconsistent media selections in real-world contexts (Fahr & Böcking, 2009; Gui et al., 2021).

Additionally, many media users reported self-regulatory concerns regarding their media use. Our media-saturated environment provides ubiquitous access to immediate gratifications through media, even in situations where its use may conflict with other goals or responsibilities. This presents constant challenges to the self-control of media users (Hofmann et al., 2017). Our findings suggest that the large amounts of free and unstructured time resulting from social distancing increased these self-control difficulties for many users, creating additional challenges for successful self-regulation. It remains an important task for future research to explore how media users manage and negotiate between the positive self-regulatory opportunities and the potentially dysfunctional self-control risks associated with media use in times of adversity and stressful life events.

Limitations and Future Directions

As all studies, the results reported here are not without limitations. First, these media usage patterns and affective experiences are self-reported. Like uses and gratifications scholarship more broadly, self-reports assume individuals are relatively aware of and can accurately report their media use. Yet, differences between logged media use and self-reported use can vary significantly (Parry et al., 2021; Verbeij et al., 2021). We acknowledge that our data can only accurately portray subjective experiences of media use. However, given our particular interest in how media is subjectively perceived and sought out, the data reported here provide substantial insight into salient instances of media use and how users felt about this use during the social distancing period. Additional behavioral measures would allow future work to assess how more automatic, routine, or subconscious patterns of media selection and avoidance relate to emotional experiences and other gratifications. We also note that these data were collected as part of larger survey reported elsewhere (Eden et al., 2020), which included additional quantitative measures such as stress and anxiety. Subgroup analyses using these measures could further parse the patterns observed here. For example, are people who are more anxious (as indicated by a quantitative validated measure) more likely to report avoidance of news, or associate the news with negative affect, than people who are not experiencing anxiety?

Notably, our study used comparable waves of participants from the sample population rather than a within-subjects design. This likely introduces variance and also limits our ability to make claims about change over time. Similarly, our data do not have a baseline of media use before the pandemic began. These constraints were pragmatic given the population and data collection period in question. However, it is important to note that while we can see differences in reported media use between the two timepoints, this does not specify whether the changes in Autumn 2020 represent a return to normal media use patterns, or if this is an artifact of longer-term coping and acclimatization. Meta-analysis or systematic review of media use, emotion, and coping before, during, and after the COVID-19 pandemic would be instructive as the constraints on daily life—and their impact on media use—continue to shift.

The data are also somewhat limited in their depth and generalizability. Age and socioeconomic status influence preexisting levels and patterns of media use, in particular, social media platform preferences. Using a college student sample, therefore, restricts the generalizability of the patterns observed here. Similarly, data were generated by participants as short essay responses in an online survey. This is necessarily less nuanced than what interviews or focus-group observations could yield, but richer and more complex than what quantitative scale responses could provide. While we acknowledge this approach has trade-offs, it allowed for both quantitative and qualitative examinations of responses—and the scale of the data set (N participants = 822, N mediaunits = 6,465) included more people and more variation in perspectives than an interview or focus-group approach would allow.

Finally, we note that COVID-19 is a specific event and a unique context for social distancing. For many people in the U.S., first-hand experience of a major health epidemic was both novel and sudden. Yet, how people cope with media is likely to be similar across many types of stressful events and during other periods of social distance and isolation. Future work should examine if these patterns of media coping replicate with other major life-stressing events or other contexts where enforced social distancing is prevalent (e.g., in immuno-compromised populations or individuals isolated far from home).

Conclusion

The widespread and sweeping changes to daily life due to the COVID-19 pandemic created novel circumstances, and they present important contexts for scholars to study how people have responded to the disease and its social effects. The present study is part of a research program investigating how anxiety, stress, and coping informed the media repertoires of college students as they faced lockdown, transition to digital learning and socialization, followed by an incremental journey back to normal life. The present report analyzed open-ended accounts of media use from the early days of the outbreak in the U.S. (Spring 2020) as well as a new school year (Autumn 2020) still under the shadow of the global pandemic. Adopting a mixed-methods approach shows quantitatively significant patterns of use among participants, and identifying thematic patterns of personal experience unearthed rich detail that a codebook alone many not have captured.

Knowing how people cope through social crisis—whether global, national, or local in scale—is informative and valuable. Media are critical infrastructure and a key source of access to information and social support, particularly during times of physical distancing. Our findings show that young people made self-aware and calculating, if not always functional or perfectly executed, choices about the media that would allow them to monitor the changes in the world while still protecting their own emotional well-being and sense of self. Our respondents sought out connection, information, and positive emotional experiences from their media channels, balancing those needs against threats from politicization and conflict, social comparison and loneliness, and upsetting negative emotions.

Responses indicate the structural features and constraints of the pandemic, the affordances and gratifications that users perceived in media channels and genres, and the social contexts surrounding use all impacted media choices and experiences in a variety of ways, both positive and negative. Media enabled unprecedented avenues for social interaction during a period of widespread physical distancing, but these mediated social exchanges facilitated everything from bitter partisan conflict to rebuilding relationships through shared media experiences. The COVID-19 pandemic unexpectedly reshaped billions of lives, but people were in many cases able to strategically use media to reshape their own environments and experiences in turn; working to meet their daily needs and protect their emotional and social well-being during an unprecedented global health crisis. As this particular pandemic begins to ebb, understanding how media use is both in response to changes in the social environment, and seized as an opportunity to shape and influence social and environmental demands (Grady, Tamborini, & Eden, 2021), is relevant to explaining media-based coping strategies long after this crisis is behind us.

We were unable to exclude participants from one university site from completing the survey at both time points. However, given the nature of our analyses and the demographic comparisons we conducted, any effect of this difference between sites is likely to be negligible.

This suggests both coders were identifying the same number of units within each response. While reliability coefficients for unitization exist, these rely on a continuous string of data to be segmented (e.g., a single transcript broken into mutually exclusive sections; Hayes & Krippendorff, 2007). Multiple prompts per participant, as was the case here, would yield hundreds of coefficients across the reliability sample. Thus, a visual inspection of unitized data sets was used.

Where participants left a prompt blank or responded negatively, no instances of media use would be unitized, so the number of units can be fewer than the number of participants.

We adopted Wikipedia genres because other common sources (e.g., IMDb) list genres alphabetically rather than in order of relevance.

Five percent of units were discussed for media type code disagreements, primarily due to blanks and manual errors (typing 33 instead of 3 as intended). Speed and accuracy tradeoffs were discussed, and mistakes were repaired in session. Twelve percent of units were discussed for genre code disagreements. The vast majority of these (80%) were due to one coder consistently omitting the news code and applying no code to units naming news content. After discussion, they recoded the reliability set to correct this. These are the reliabilities reported. Once reliability was achieved, the full data set was randomly split and distributed.

We examined media types in two ways. First, by comparing all social media types to other general media categories (such as tv/film). Here, media mentioned infrequently (such as video games, print, web, audio) were collapsed into an Other Media category. Then, we looked at social media channels mentioned by name in isolation.

For genres, there were significantly fewer units overall (n = 1,107) but more possible codes in the codebook, so the cutoff here was set at 50 mentions. Merges were based on content similarities in valence, tone, and theme: children and family were added to comedy; various types of drama were combined; documentaries and news were merged; and several nonfiction types (lifestyle, music, reality) were collapsed together. Details on OSF supplement.

Quoted participants are referred to by the item they were responding to (e.g., M = responding about media used more, L = less, A = avoid) and the response number.

Supplemental Materials

https://doi.org/10.1037/tmb0000041.supp


Received October 1, 2020
Revision received June 1, 2021
Accepted June 22, 2021
Comments
0
comment

No comments here

Why not start the discussion?