Coping with the COVID-19 Pandemic by Using Media: Extending the Coping Goodness-of-Fit Hypothesis to Media Use

ABSTRACT

During the COVID-19 pandemic, people had to cope with stress in two different ways: engaging in problem-focused behaviors (e.g. seeking information) and regulating their emotions. In particular, during lockdowns, media played an important role in these coping processes. This short-term longitudinal study during the first stage of the COVID-19 pandemic in April 2020 in Germany (N = 348) tested how coping strategies related to two categories of coping effectiveness outcomes, situation-addressing outcomes (preventive behaviors and knowledge) and mental health outcomes (stress and perceived coping efficacy). We moreover tested which kind of media use is related to effective coping derived from a combination of the coping goodness-of-fit hypothesis and mood management theory. The study demonstrated that information seeking during the uncertain pandemic circumstances might have been a double-edged sword: Increased use of the coping strategy information seeking was related to increased stress but also to enhanced knowledge and better adherence to behavioral guidelines, however only on the cross-sectional level. Additionally, our results showed that using positive media content might have helped media users to reduce stress and to adhere to behavioral guidelines. Thus, a combination of seeking information at predefined, limited time points and using positive media content might have been a fruitful way of using media for coping during this situation of crisis.

KEYWORDS:

• Stress
• coping using media
• mood management
• goodness-of-fit hypothesis
• COVID-19

In Spring 2020, contact restrictions to contain the spread of the COVID-19 virus were introduced across European countries. Many people practiced social distancing, most service providers and schools were closed, and no vaccination was available (Bundesregierung, 2020; Jungblut, 2020). This situation was stressful: In addition to worries about their health and a growing economic crisis, people faced a situation in which their ways of coping with stress, such as going to the gym or meeting friends, were limited (van Bavel et al., 2020).

Consequently, media-based coping strategies gained importance (Nabi et al., 2022). While media use for coping has received growing attention in recent years, systematic assessments of them and their integration with stress and coping theories remain scarce (Wolfers & Schneider, 2021). Thus, the COVID-19 lockdowns provided a unique context to study how media can be used for successful coping with crises.

In crisis situations, individuals need to find ways to cope with stress for two main purposes (World Health Organization, 2020): First, people need to cope to maintain their mental health (Xia et al., 2021). Studies reported increased stress and mental health difficulties during the pandemic (WHO, 2022), suggesting people faced difficulties adjusting emotionally to the situation. Second, individuals need to cope in ways which address the crisis situation. During the COVID-19 pandemic, individuals needed to prevent infection (Betsch, 2020). This necessitated adherence to behavioral guidelines and staying informed about new scientific results to adjust preventive behaviors when required (Betsch, 2020).

Thus, a need to investigate two kinds of coping effectiveness outcomes in tandem emerged: protecting one’s mental health, such as minimizing stress, and addressing the situation by, for example, displaying preventive behavior. Although the notion that both mental health-related and situation-addressing outcomes are necessary parts of effective coping has been discussed repeatedly (Folkman & Moskowitz, 2004; Tennen & Affleck, 2002), few studies examined both coping effectiveness outcomes together (Folkman & Moskowitz, 2004; Glanz et al., 2015). Examining one type of effectiveness outcome separately could result in coping behaviors being recommended that are beneficial for addressing the crisis but detrimental to mental health, or vice versa. A joint study allows determining which coping behaviors are positively related to both types of effectiveness outcomes.

Building on the goodness-of-fit hypothesis and mood management theory, the current one-week longitudinal study aims at investigating how employing a set of coping strategies was related to situation-addressing coping effectiveness (i.e., preventive behaviors and perceived knowledge about COVID-19) and mental health coping effectiveness (i.e., stress and perceived coping efficacy, PCE). This allowed examining the effects certain coping strategies have on both outcomes in times of crisis. Furthermore, we investigated how the use of media content for coping affected effectiveness for both sets of outcomes, answering the call for an integration of media and coping theories (Wolfers & Schneider, 2021).

Stress and Coping

Stress can be defined as a transaction between the person and the environment in which the person perceives the demands placed on them as exceeding their resources (Lazarus & Folkman, 1984). When stressed, individuals cope to resolve the stressful transaction. Coping strategies are thereby often divided into problem-focused coping strategies, where individuals aim at solving the stress-evoking problem, and emotion-focused coping strategies, where individuals try to regulate the emotions triggered by the experienced stress (Lazarus & Folkman, 1984). Media can be understood as coping tools which can be used to implement emotion-focused and problem-focused strategies (Wolfers & Schneider, 2021).

In this study, we focus on self-distraction as emotion-focused strategy and information seeking and planning/active coping as problem-focused strategies. Self-distraction includes coping behaviors aimed at taking “one’s mind off the stressor” (Carver, 1997, p. 95). As its focus points away from the problem, self-distraction is classified as emotion-focused strategy (Finkelstein-Fox & Park, 2019). Information-seeking represents coping behaviors aimed at finding out more about the stress-evoking problem. Active coping includes “taking steps to try to remove or circumvent the stressor” (Carver et al., 1989, p. 268), and planning includes thinking about which of these steps are best and feasible to implement (Carver et al., 1989). As these three strategies are focused on addressing the stress-evoking problem, they are usually seen as problem-focused strategies (Finkelstein-Fox & Park, 2019). Because planning and active coping are part of the same process (thinking about and implementing steps to address the stressor), we combined these strategies (see Meyer, 2001).

Assessing how coping behaviors affect coping effectiveness is one central aim of coping research. Definitions of coping effectiveness depend on the research aim (Folkman & Moskowitz, 2004) and can be represented as two general categories: Outcomes addressing the effectiveness of coping for an individual’s well-being and emotional state (mental health outcomes), most prominently the reduction of stress and negative emotions, and outcomes involving an effective change in the situation itself (situation-addressing outcomes), such as solving the stress-evoking problem or adjusting behaviors (Tennen & Affleck, 2002).

As situation-addressing outcomes, we chose preventive behaviors that helped avoid infection and staying informed about new knowledge (Betsch, 2020). For mental health outcomes, we examine perceived stress and the general feeling that one is coping well with the situation (perceived coping efficacy; Betsch, 2020; Tennen & Affleck, 2002). In the following literature review, we show how using emotion-focused and problem-focused coping strategies could relate to these coping effectiveness outcomes before focusing on an integration of media characteristics in these processes.

The Goodness-of-Fit Hypothesis and Coping Effectiveness

Lazarus and Folkman (1984) emphasized that effective coping depends on the fit between the situation and coping. A central situational characteristic is controllability, the degree to which individuals can change the conditions of a stressful situation (Wolfers et al., 2023). The coping goodness-of-fit hypothesis states that problem-focused coping is effective in situations with high controllability (Conway & Terry, 1992; Person & Frazier, 2024). Conversely, if individuals cannot change stress-evoking circumstances, trying to alter them should be dysfunctional, leading to ineffective coping. Therefore, emotion-focused coping is considered more effective for less controllable circumstances (Park et al., 2004).

While some findings support the goodness-of-fit hypothesis (e.g., Vitaliano et al., 1990), not all studies find all predicted relations (e.g., Finkelstein-Fox & Park, 2019; Park et al., 2004; Person & Frazier, 2024). Finkelstein-Fox and Park (2019) conclude in their review that more longitudinal studies are required to test the hypothesis. In research on the goodness-of-fit hypothesis, the dependent variables are usually stress (English & Zhang, 2020; Person & Frazier, 2024), coping efficacy (Conway & Terry, 1992), or other measures of well-being (Conway & Terry, 1992). Thus, we assume that the goodness-of-fit hypothesis is especially applicable to mental health outcomes.

The circumstances in Germany in Spring 2020 can be considered a situation of low individual control. The uncertainty about the new virus and its consequences was still high, and strict contact restrictions were implemented, limiting individuals’ personal freedom. Therefore, the goodness-of-fit hypothesis could be tested by comparing the effectiveness of different coping strategies in a situation where controllability is constantly low.

H1:

The more emotion-focused strategies (i.e., distraction) are employed, the a) less stress and b) the higher perceived coping efficacy (PCE) participants report.

H2:

The more problem-focused strategies (i.e., active coping/planning, information seeking) are employed, the a) more stress and b) the lower the PCE participants report.

For situation-addressing outcomes, the applicability of the predictions of the goodness-of-fit hypothesis is less clear than for mental health outcomes. The goodness-of-fit hypothesis assumes that the effectiveness of problem-focused compared to emotion-focused coping depends on the individuals’ control of the situation. However, regardless of the controllability of the situation, implementing only emotion-focused coping strategies should not lead to improved situation-addressing outcomes as emotion-focused strategies do not aim at changing situational circumstances. Indeed, research on disasters and health threats, which often focus on situation-addressing outcomes, suggests that problem-focused strategies are in general more effective (Kievik & Gutteling, 2011).

For instance, experimental data revealed positive relationships between information seeking and the intention to engage in preventive behaviors regarding flood risks (Kievik & Gutteling, 2011). Similarly, survey studies about Zika-virus risk prevention found that information seeking on social media was positively related to preventive behavior intentions (Lee et al., 2020) and perceived knowledge about the virus (Hubner & Hovick, 2020). A study on the COVID-19 pandemic in Mexico also confirmed a positive relationship between active coping and preventive behavior (Javier & Roberto, 2022). Therefore, we assume that problem-focused coping is associated with higher situation-addressing coping effectiveness.

H3:

The more problem-focused strategies (i.e., active coping, planning, information seeking) are employed, a) the higher participants‘ perceived knowledge and b) the more people show preventive behaviors.

Associations between situation-addressing coping effectiveness and emotion-focused coping are less clear. Emotion-focused coping could substitute efforts for problem-focused coping strategies, yielding less knowledge and less preventive behaviors. However, Watson (2018) proposed that individuals who employ more emotion-focused strategies are less stressed and therefore better able to also use problem-focused strategies. Recent findings on coping with COVID-19 did not reveal a significant relationship between emotion-focused coping and preventive behaviors (Javier & Roberto, 2022). Due to the uncertain relationship between emotion-focused coping and situation-addressing outcomes, we ask:

RQ1:

Is using more emotion-focused strategies (i.e., distraction) related to more perceived knowledge or preventive behaviors?

Extending the Goodness-of-Fit Hypothesis by Media Characteristics

The restrictions during COVID-19 lockdowns emphasized the role of media in coping processes. In a recent literature review on coping using media, Wolfers and Schneider (2021) suggested conceptualizing media as coping tools, defined “as instruments through which (a) a coping goal can be achieved and (b) a coping behavior can be performed” (Wolfers & Schneider, 2021). Accordingly, a combination of coping strategies and coping tools is chosen in a stressful situation. The fit of this combination to the situation determines coping effectiveness. Thus, not only do the strategies have to fit the situational circumstances but they should be combined with suitable coping tools to be effective. For categorizing media as coping tools with different characteristics, the factors proposed in mood management theory (MMT, Zillmann, 1988) can be of particular value (Wolfers et al., 2020).

MMT assumes that stressed individuals select media stimuli that are unrelated to the stressful encounter (low semantic affinity), positive (high hedonic valence), calming (low excitatory potential), and absorbing (high absorption potential) to regulate their mood toward a positive, hedonic homeostasis (Zillmann, 1988). Stevens and Dillman Carpentier (2017) related MMT to emotion-focused coping as MMT focuses on stress relief and mood regulation by regulating moods and emotions and not on solving problems, implying that Zillmann’s assumptions concerning media use for stress regulation might particularly apply to emotion-focused coping.

To reduce complexity, we focus on the qualities of hedonic valence and semantic affinity. Both qualities have been used in research assessing diverse forms of media content spanning news, social media posts, and movies (e.g., Kim & Oliver, 2011; Wolfers et al., 2023). As our focus in terms of media content is broad, we decided to focus on hedonic valence and semantic affinity for a first test of the combination of MMT and the goodness-of-fit hypothesis.

Studies on MMT found mixed results concerning the stress-reducing effects of media (see Reinecke, 2017 for a review). Supporting the stress-reducing effect of semantically different media content, Krakowiak and Tsay-Vogel (2018) found that participants confronted with their vices reported more relaxation and psychological detachment when they were exposed to a story featuring a good media character compared to a bad media character. For valence, Rieger et al. (2017) found that positive media content was related to increased recovery from a strain-inducing task compared to a control condition. This effect, however, was not different from the effect of exposure to negative media content.

As most research on MMT is experimental, differences between different motivations for using media have not been identified. However, Stevens and Dillman Carpentier (2017) showed that MMT assumptions only hold for individuals with an emotion-focused coping tendency. This suggests that the coping strategy with which an individual chooses media content might explain previous, mixed findings in MMT research.

In the present study, we asked about the coping strategy for using media and presume that MMT’s assumptions hold when media are used for emotion-focused coping. Following Wolfers and Schneider’s (2021) reasoning that coping strategies must be combined with a fitting set of coping tools to be effective, we propose that individuals who chose media content for emotion-focused coping that is both unrelated to the current situation and positive cope more effectively. This should hold for mental health coping effectiveness outcomes, as MMT focuses on affective outcomes.

H4:

The associations between emotion-focused coping and stress and PCE (postulated in 220 H1a and b) are stronger for those that use media content (a) nonrelated to the current situation (semantic affinity). (b) that has a positive valence (hedonic valence).

Studies suggest that MMT’s assumptions might not hold if media are used for problem-focused coping. For example, students who experienced a high threat due to an economic crisis appreciated reading a story about a victim who could not find a job more than reading about a survivor or outperformer who found work (Kim & Tsay-Vogel, 2016). This contradicts MMT’s assumption that positively valenced media content is more effective in reducing stress. Similarly, Nabi et al. (2006) found that individuals who experienced greater regret after cheating on a relational partner chose cheating-related TV programming. Watching experience-related stories lowered participants’ regret, contradicting MMT’s assumption that using semantically different media content reduces negative emotions more successfully. Active engagement with the experience is described as one explanation of these effects (Kim & Tsay-Vogel, 2016; Nabi et al., 2006), which can be considered a form of problem-focused coping (Stevens & Dillman Carpentier, 2017). Taken together, for problem-focused coping, stressor-related and negative media content might be associated with increased coping effectiveness.

Conversely, the literature on the effects of viewing television programming about a crisis suggests the opposite (for an overview see Pfefferbaum et al., 2014). Dougall et al. (2005), for example, found that watching more television programming related to the anthrax attacks in 2001 increased anthrax-related distress. Similar results were reported for social media content related to Hurricane Sandy (Goodwin et al., 2013). These studies do not differentiate between reasons for media use or between different media characteristics. Thus, it remains unclear if using media with specific characteristics improves the fit of problem-focused coping to an uncontrollable situation. We ask:

RQ2:

Do certain characteristics of used media (hedonic valence, semantic affinity) moderate the relationships between problem-focused strategies and stress and PCE postulated in H2?

MMT focuses on regulating mood and not on acquiring knowledge and displaying preventive behaviors. Thus, these outcomes play only a minor role in research on MMT. We therefore explore how semantic affinity and valence strengthen or weaken the relationships between problem-focused or emotion-focused coping and situation-addressing outcomes.

RQ3:

Do media characteristics moderate the relationships between problem-focused or emotion-focused coping and perceived knowledge and preventive behaviors?

Materials and Methods

Participants and Procedure

We conducted a two-wave panel survey from April 2–3 (Wave 1, W1) to April 9–11 (Wave 2, W2) 2020. Participants were recruited via the website Prolific. All individuals registered on Prolific who currently resided in Germany were eligible. We aimed at recruiting 300 participants but planned to expand this number if the subsample used in the media-use analyses were below 250. As this was necessary, W1 was completed by 362 participants who received 2.50 £ as compensation. We excluded 14 participants for failing attention checks or not consenting to data usage. We invited 348 participants to W2 1 week later. Of these, 323 took part (93% response rate) representing the final sample for the longitudinal analyses. Participants received 2.13 £ for completing W2. Participants in the final sample were young (M_AGE = 28.48, SD = 8.74, 18–67). The majority was male (61%,) and well-educated (56% finished at least a Bachelor’s degree). Of all participants, 41% were students, 33% worked full-time, 15% part-time, and 7% were unemployed (4% other). None of the participants reported having tested positive for COVID-19. Twelve percent indicated belonging to a risk group for suffering more severe consequences from COVID-19 (e.g., having a preexisting medical condition). When including participants who did not take part in W2, demographics differ only marginally (M_AGE = 28.25, SD = 8.58, 18–67; 61% male).

Questionnaires were administered in German. After providing informed consent, participants answered questions about their media use, stress, emotions, and coping behaviors. For participants who employed the coping strategies distraction, active coping/planning, and information seeking at least sometimes, respectively, we asked about the media they used (if any) for the respective strategy to assess H4, RQ2, and RQ3. The samples for these analyses therefore contain only data from participants who answered the questions in W1. For the distraction subsample, 234 participants remained (M_AGE = 27.78, SD = 7.86, 18–67; 59% male). For the active coping/planning sample 281 participants remained (M_AGE = 28.22, SD = 8.44, 18–67; 63% male). For the information-seeking subsample, 311 participants remained (M_AGE = 28.30, SD = 8.53, 18–67; 62% male). The data, questionnaires, and R code can be found at OSF: https://osf.io/t7uxr/.¹ The survey was approved by the ethics committee of the Leibniz-Institut für Wissensmedien Tübingen.

Measures

For descriptives of the mean indices of all measures see Table 1.

Table 1. Means and Standard Deviations of the Measured Variables.

Measured variables		Wave 1				Wave 2					Correlations wave 1^3
		n	M	SD		n	M	SD		1		2	3	4	5	6	7	8	9	10	11	12
	Coping strategies
1	Information Seeking	348	3.68	0.91		323	3.19	0.96
2	Distraction	348	3.33	0.95		323	3.42	0.86		.12
3	Active Coping/Planning	348	2.98	0.82		323	2.78	0.80		.32		.31
	Coping efficacy outcomes
4	Stress	348	38.65	24.45		323	35.94	23.96		.28		.20	.24
5	PCE^1	348	3.71	0.83		323	3.75	0.78		−.07		−.14	.03	−.60
6	Perceived knowledge	348	4.77	0.94		323	4.73	0.91		.34		−.01	.21	.10	.05
7	Preventive behaviors	348	4.49	0.53		323	4.48	0.49		.25		.16	.22	.11	.11	.13
	Media characteristics^2
	Distraction
8	Hedonic valence	234	4.10	0.80		236	4.13	0.68		−.09		.05	−.13	−.14	.24	−.05	.18
9	Semantic Affinity	234	2.49	1.37		236	2.36	1.27		.14		−.03	.14	−.01	.09	.08	−.05	−.27
	Active Coping/Planning
10	Hedonic Valence	282	3.71	0.89		238	3.80	0.81		−.09		.04	.13	−.15	.26	.02	.14	.15	.07
11	Semantic Affinity	281	4.11	1.05		237	3.97	1.05		.26		.08	−.06	−.01	.09	.07	.09	.23	.06	−.02
	Information seeking
12	Hedonic Valence	311	3.64	0.94		264	3.47	0.93		.01		−.10	.09	−.22	.27	.11	.07	.12	.02	.35	−.06
13	Semantic Affinity	311	4.35	0.94		263	4.38	0.92		.06		.06	−.03	−.03	.07	−.09	.20	.16	−.04	.12	.21	−.16

Note. ¹PCE = Perceived Coping Efficacy. ²Only answered by participants who used respective strategy at respective wave at least sometimes. ³Correlations between variables 1–7 based on full sample, correlations between media characteristics and variables 1–7 based on respective subsample, correlations between media characteristics across different strategies based on subsample of participants who used all three strategies at least sometimes, N = 178.

Stress

Stress was measured by asking participants how much stress they experienced in the past week and yesterday on a scale from 1 (no stress) to 100 (extreme stress) adapted from Nabi et al. (2022). Both items correlated highly (r_w1 = 0.82, r_w2 = 0.77). We chose this direct way of measuring stress because it is efficient, readily understood by a wide range of participants, and correlates with longer scales (Nabi et al., 2022). Further, many longer scales also measure perceived control over the situation (e.g., Cohen et al., 1983), which should be considered a factor distinct from stress for testing the coping-fit hypothesis.

Perceived Coping Efficacy

For PCE, participants rated statements such as “I feel I cope well with the current situation“on a scale from 1 (does not apply at all) to 5 (fully applies). Existing PCE scales focused on specific contexts (e.g., pain). We adapted items from Cummings et al. (1994), Prati et al. (2011), and Slepian and Moulton-Tetlock (2019) to form a five-item scale. Internal consistency was excellent (ω_{u, w1} = .90, ω_{u, w2} = .89).

Perceived Knowledge

For perceived knowledge, participants rated three items concerning their knowledge of COVID-19, how to stop the spread of the virus, and treatment options on a scale from 1 (very poor knowledge) to 7 (very good knowledge) as in Utz et al. (2022). Internal consistency was acceptable (ω_{u, w1} = .69, ω_{u, w2} = .72).

Preventive Behavior

Preventive behaviors were assessed by asking participants how they adapted their behavior in response to the crisis (Utz et al., 2022). To reduce social desirability, two filler items were included. In line with public recommendations, subjects were surveyed about acting according to “sneeze and cough”-rules, washing hands more often, socially distancing from other people, staying at home if sick, minimizing visits to older family members, avoiding touching face with unwashed hands, and minimizing meeting people on a scale from 1 (does not apply at all) to 5 (applies completely). Based on Utz et al. (2022), two reversed questions were excluded. Internal consistency was acceptable (ω_{u, w1} = .74, ω_{u, w2} = .74).

Coping Strategies

Coping strategies were assessed using the Brief COPE (Carver, 1997), adapted from the translation of Knoll et al. (2005). Two items for information seeking were added. Participants rated how often they applied 12 strategies to cope with stress evoked through the COVID-19 crisis in the past week on a scale ranging from 1 (never) to 5 (constantly). We focus on the strategies of distraction (e.g., “I did something to think about it less”), information seeking (e.g., “I looked for information about the current situation”), and active coping/planning (e.g., “I took action to try to make the situation better”). The Brief COPE measures each strategy with two items. Items for distraction (r_w1 = .48, r_w2 = .40) and information seeking (r_w1 = .70, r_w2 = 67) correlated moderately to strongly. For the combined coping strategies of active coping and planning, the four items showed good internal consistency (ω_{u, w1} = .76, ω_{u, w2} = .77).

Perceived Media Characteristics

Participants who reported using the strategies of distraction, planning/active coping, and information seeking at least sometimes were asked if they employed the respective strategy by using media. Participants who reported at least some media use for a coping strategy were then asked about the characteristics of the media they had used for a particular strategy. They indicated the hedonic valence of the media content they used for a respective coping strategy on a scale from 1 (positive) to 5 (negative) and the semantic affinity on a scale from 1 (related to the current situation) to 5 (unrelated to the current situation). Values were recoded such that higher values mean more positive media or greater semantic affinity, respectively.

Data Analysis

As a pre-analyses step, we tested the measurement model of all variables using a structural equation model (SEM) with a maximum likelihood estimator with robust standard errors and Satorra–Bentler scaled test statistic (MLM). To ensure measurement invariance across time, we included data from both waves and set factor loadings for same items equal over time. The measurement model fits the data well (χ ² [1077] = 1311.37, p < .000; CFI = 0.96; RMSEA = .03, SRMR = .06, see OSF).

To test our hypotheses, we employ two different sets of analyses. First, testing H1–3 and RQ1, we ran a two-wave cross-lagged panel SEM using an MLM estimator with our full sample including the coping strategies information seeking, active coping/planning, and distraction, as well as stress, PCE, perceived knowledge, and preventive behavior in both waves. Stress was standardized as it was measured on a 100-point scale. We constrained factor loadings of same items over time and only allowed residual correlations for same items across waves. Only complete cases were used.

To test the assumptions including the media characteristics (H4, RQ2, RQ3), data of participants who used a strategy at least sometimes at W1 were used (Ns = 215–299). We, thus, ran separate regression models for each coping strategy once cross-sectionally with W1 data and once longitudinally with the dependent variable measured at W2. Linear regressions and mean indices were used as the sample size was reduced for these models and modeling approaches for interaction effects in SEMs are still limited. For these analyses, the control models included age, gender, risk group membership, household size, number of children, quarantine status, coping strategies (information seeking, active coping/planning, distraction), and media characteristics (hedonic valence, semantic affinity) for the respective strategy. For longitudinal models, the respective dependent variable at W1 was also included in the control model. In the second step, interactions between coping strategies and media characteristics were introduced.

As we tested a large number of effects, we conducted a Bonferroni-Holm correction of p-values in addition to using the standard significance level of p < .05 including the p-values for the direct effects of the coping strategies from the SEM and the direct effects of the media characteristics and the moderation effects from the regression analyses. Ordering our p-values according to the Bonferroni-Holm (BH) method led to a rejection of all effects with p < .0005. We indicate for all effects with p < .05 if they also hold considering this stricter significance value using “BH: n.s.” for non-significant and “BH: sig.” for significant effects (for the calculation, see OSF).

Results

Descriptive Results

The most frequently used coping strategy at W1 was information seeking. At W2, distraction was applied the most, followed by information seeking. The use of information seeking and active coping/planning decreased from W1 to W2, while distraction increased (see Table 1).

Participants who used distraction as a coping strategy reported using mostly positive media and media that were unrelated to the situation (see Table 1). Participants who used media for active coping/planning used positive and situation-related content. Lastly, those who used media for information seeking used more positive content and overwhelmingly situation-related content. Table 1 shows the correlations between all variables at W1.

Results of the Cross-Lagged Panel Model

Figure 1 shows the SEM results (for full results see OSF). The model fits the data well (χ ² [1083] = 1314.04, p < .001; CFI = 0.96; RMSEA = .03, SRMR = .06). H1 posited that participants who employ more emotion-focused coping (i.e., distraction) would report less stress and higher PCE. Contradicting H1, using distraction as a coping strategy positively correlated with stress (b = 0.20, SE = 0.06, β = .28, p < .001, BH: sig.) and negatively correlated with PCE (b = −0.13, SE = 0.04, β = −.24, p = .001, BH: n.s.) at W1. Longitudinally, using distraction as a coping strategy predicted neither stress (b = −0.02, SE = 0.07, β = −.01, p = .815) nor PCE (b = −0.05, SE = 0.07, β = −.05, p = .486). Hence, H1 was not supported.

Figure 1. Structural Equation Model Including Coping and Coping Effectiveness Variables.

N = 312. Only significant paths are shown, and indicators are omitted. Paths not included in the hypotheses/research questions are shown in gray.

H2 predicted that participants who employ more problem-focused coping strategies (i.e., active coping/planning and information seeking) would report more stress and lower PCE. Indeed, both active coping/planning (b = 0.22, SE = 0.05, β = .36, p < .001, BH: sig.) and information seeking (b = 0.28, SE = 0.05, β = .36, p < .001, BH: sig.) were significantly positively related to stress at W1. However, neither information seeking (b = −0.07, SE = 0.04, β = .112, p = .105) nor active coping/planning (b = 0.01, SE = 0.04, β = .02, p = .766) was significantly correlated with PCE at W1. Longitudinally, more frequent use of information seeking as a coping strategy predicted neither stress (b = 0.02, SE = 0.07, β = .02, p = .782) nor PCE (b = −0.01, SE = 0.07, β = −.01, p = .894). Active coping/planning was not a significant predictor of stress (b = 0.10, SE = 0.95, β = .07, p = .309) or PCE (b = 0.01, SE = 0.08, β = .01, p = .921) longitudinally. Thus, H2 was supported only at the cross-sectional level with stress as the dependent variable.

H3 predicted that participants who employ more problem-focused coping strategies (i.e., active coping/planning and information seeking) would report more perceived knowledge and preventive behaviors. Participants who employed more information seeking at W1 also reported more perceived knowledge (b = 0.34, SE = 0.06, β = .46, p < .001, BH: sig.), which was also the case for active coping/planning (b = 0.16, SE = 0.05, β = 0.25, p = .001, BH: n.s.). Both information seeking (b = 0.12, SE = 0.03, β = .33, p < .001, BH: sig.) and active coping/planning (b = 0.10, SE = 0.02, β = .30, p < .001, BH: sig.) were significantly related to increased preventive behaviors cross-sectionally. However, longitudinally, active coping/planning predicted neither perceived knowledge (b = −0.07, SE = 0.09, β = −.06, p = .424) nor preventive behaviors (b = −0.04, SE = 0.04, β = −.07, p = .260). This was also true for information seeking (perceived knowledge: b = −0.00, SE = 0.08, β = −.00, p = .986, preventive behavior: b = −0.02, SE = 0.03, β = −.03, p = .540). Therefore, we found support for H3 cross-sectionally but not longitudinally.

Answering RQ1, the SEM showed that distraction was not related to perceived knowledge cross-sectionally (b = 0.01, SE = 0.06, β = .02, p = .853) or longitudinally (b = −0.06, SE = 0.09, β = −.05, p = .475). Distraction was positively associated with preventive behaviors cross-sectionally (b = 0.08, SE = 0.03, β = .23, p = .022, BH: n.s.) but not longitudinally (b = 0.02, SE = 0.04, β = .03, p = .657).

In addition to the effects of coping on the outcome variables, we also included the reversed paths in the model. None of the tested outcomes predicted coping behaviors over time (see Figure 1).

Moderation Effects

To test the assumption that using media with certain characteristics would increase or decrease the associations with coping effectiveness and the coping strategies, we ran regression models and compared models with the interaction effect to the models without the interaction effects (see Tables 2–4). Apart from one exception, entering the interactions between coping strategies and media characteristics did not yield a significant increase in explained variance.

Table 2. Predictors of the Outcome Variables Including Media Characteristics for Distraction.

	Wave 1 Cross-sectional				Longitudinal
	b	SE (b)	β	p	b	SE (b)	β	p
DV: Stress		n = 227				n = 215
Stress at W1	-	-	-	-	0.58	0.06	.59	<.001
Distraction (D)	2.49	1.57	.10	.114	-3.14	1.39	-.13	.025
Hedonic Valence (H)	-2.59	1.62	-.11	.111	0.91	1.44	.04	.529
Semantic Affinity (S)	-1.70	1.56	-.07	.279	1.63	1.36	.07	.230
D*H	4.30	1.63	.18	.009	0.17	1.44	.01	.907
D*S	0.22	1.50	.01	.884	-0.96	1.30	-.04	.464
Δ R^2 to control model		.03		.029		.00		.732
R^2		.20		<.001		.38		<.001
DV: Perceived coping efficacy		n = 227				n = 215
PCE at W1	-	-	-	-	0.60	0.05	.61	<.001
Distraction (D)	-0.00	.05	-.00	.954	0.05	0.04	.07	.230
Hedonic Valence (H)	0.23	0.06	.29	<.001	0.10	0.05	.13	.028
Semantic Affinity (S)	0.10	0.05	.13	.058	-0.00	0.04	-.01	.908
D*H	-0.10	0.06	-.12	.083	0.07	0.04	.09	.117
D*S	-0.02	0.05	-.02	.711	0.07	0.04	.08	.101
Δ R^2 to control model		.01		.220		.01		.129
R^2		.15		.001		.46		<.001
DV: Perceived knowledge (PK)		n = 227				n = 215
PK W1	-	-	-	-	0.64	0.05	.68	<.001
Distraction (D)	-0.05	0.06	-.06	.408	-0.12	0.05	-.13	.014
Hedonic Valence (H)	-0.00	0.07	-.00	.968	0.07	0.05	.07	.190
Semantic Affinity (S)	-0.00	0.06	-.00	.986	0.03	0.05	.02	.630
D*H	-0.01	0.07	-.01	.893	-0.02	0.05	-.02	.641
D*S	-0.08	0.06	-.08	.195	0.5	0.05	.06	.252
Δ R^2 to control model		.01		.422		.00		.390
R^2		.18		<.001		.53		<.001
DV: Preventive behavior (PB)		n = 227				n = 215
PB W1	-	-	-	-	0.64	0.05	.68	<.001
Distraction (D)	0.02	0.03	.03	.693	-0.02	0.02	-.04	.391
Hedonic Valence (H)	0.08	0.03	.18	.011	0.04	0.02	.09	.096
Semantic Affinity (S)	-0.01	0.03	-.02	.805	0.01	0.02	.02	.629
D*H	0.03	0.03	.06	.825	0.00	0.02	.01	.860
D*S	-0.01	0.03	-.03	.686	0.01	0.02	.03	.560
Δ R^2 to control model		.00		.581		.00		.843
R^2		.18		<.001		.57		<.001

Note. DV = dependent variable, Variables are centered. Cross-sectional models use W1, longitudinal models use W2 DVs. Not all estimates shown, control model includes age, gender, risk group membership, being quarantined, household size, number of children, information seeking, distraction, active coping/planning, media characteristics; For longitudinal model also W1 outcome variables. Coefficients of final models including interactions shown.

Table 3. Predictors of the Outcome Variables Including Media Characteristics for Active Coping/Planning.

	Wave 1 Cross-sectional				Longitudinal
	b	SE (b)	β	p	b	SE (b)	β	p
DV: Stress		n =270				n =251
Stress W1	-	-	-	-	0.60	0.05	.60	<.001
Active Coping (AC)	4.64	1.55	.18	.003	1.88	1.37	.08	.169
Hedonic Valence (H)	-4.30	1.43	-.18	.003	-0.99	1.26	-.04	.432
Semantic Affinity (S)	-1.30	1.47	-.05	.377	-0.37	1.26	-.01	.770
AC*H	0.55	1.34	.02	.683	0.06	1.21	.00	.959
AC*S	0.80	1.15	.03	.485	-1.05	1.01	-.04	.300
Δ R^2 to control model		.00		.728		.00		.570
R^2		.20		<.001		.46		<.001
DV: Perceived coping efficacy		n = 270				n = 251
PCE at W1	-	-	-	-	0.63	0.05	.68	<001
Active Coping (AC)	0.04	0.05	.04	.496	-0.02	0.04	-.03	.573
Hedonic Valence (H)	0.120	0.05	.24	<.001	0.01	0.04	.01	.869
Semantic Affinity (S)	0.11	0.05	.13	.033	0.02	0.04	.03	.581
AC*H	-0.02	0.05	-.03	.642	0.01	0.04	.01	.786
AC*S	0.00	0.04	.00	.947	0.04	0.03	.06	.155
Δ R^2 to control model		.00		.894		.00		.363
R^2		.15		<.001		.48		<.001
DV: Perceived knowledge (PK)		n = 270				n = 251
PK W1	-	-	-	-	0.65	0.05	.68	<.001
Active Coping (AC)	0.12	0.06	.12	.055	0.04	0.05	.04	.430
Hedonic Valence (H)	0.01	0.06	.01	.845	0.02	0.04	.02	.656
Semantic Affinity (S)	0.01	0.06	.01	.836	-0.00	0.04	-.00	.984
AC*H	0.08	0.05	.08	.137	-0.03	0.04	-.03	.472
AC*S	-0.06	0.04	-.07	.159	0.03	0.04	.03	.393
Δ R^2 to control model		.01		.114		.00		.483
R^2		.20		<.001		.52		<.001
Preventive behavior (PB)		n = 270				n = 251
PB W1^3	-	-	-	-	0.61	0.05	.64	<.001
Active Coping (AC)	0.08	0.03	.17	.001	-0.00	0.02	-.01	.926
Hedonic Valence (H)	0.05	0.03	.12	.054	0.05	0.02	.12	.013
Semantic Affinity (S)	0.02	0.03	.05	.472	0.00	0.02	.01	.822
AC*H	0.02	0.03	.05	.378	-0.04	0.02	-.09	.064
AC*S	0.00	0.02	.00	.938	0.02	0.02	.04	.379
Δ R^2 to control model		.00		.677		.01		.091
R^2		.15		<.001		.49		<.001

Note. DV = dependent variable. Variables are centered. Cross-sectional models use W1, longitudinal models use W2 DVs. Not all estimates shown, control model includes age, gender, risk group membership, being quarantined, household size, number of children, information seeking, distraction, active coping/planning, media characteristics; For longitudinal model also W1 outcome variables. Coefficients of final models including interactions shown.

Table 4. Predictors of the Outcome Variables Including Media Characteristics for Information Seeking.

	Wave 1 Cross-sectional				Longitudinal
	b	SE (b)	β	p	b	SE (b)	β	p
DV: Stress		n = 299				n = 280
Stress W1	-	-	-	-	0.61	0.05	.62	<.001
Information Seeking (IS)	4.52	1.39	.19	.001	1.61	1.21	.07	.186
Hedonic Valence (H)	−5.20	1.33	−.21	<.001	−0.01	1.18	−.00	.993
Semantic Affinity (S)	−2.14	1.35	−.09	.114	−0.14	1.13	−.01	.904
IS*H	0.88	1.25	.04	.480	0.49	1.08	.02	.648
IS*S	0.32	1.27	.01	.803	−1.18	1.09	−.05	.182
Δ R^2 to control model		.00		.768		.00		.485
R^2		.20		<.001		.43		<.001
DV: Perceived coping efficacy		n = 299				n = 280
PCE at W1	-	-	-	-	0.64	0.05	.67	<.001
Information Seeking (IS)	−0.07	0.05	−.09	.126	−0.01	0.05	−.01	.764
Hedonic Valence (H)	0.19	0.05	.23	<.001	0.04	0.04	.05	.263
Semantic Affinity (S)	0.11	0.05	.14	.014	0.02	0.04	.03	.509
IS*H	−0.06	0.04	−.07	.188	0.02	0.03	.03	.466
IS*S	−0.08	0.04	−.09	.073	0.03	0.03	.04	.411
Δ R^2 to control model		.01		.107		.00		.570
R^2		.18		<.001		.47		<.001
DV: Perceived knowledge (PK)		n = 299				n = 280
PK W1	-	-	-	-	0.67	0.05	.68	<.001
Information Seeking (IS)	0.24	0.05	.26	<.001	−0.04	0.04	−.05	.333
Hedonic Valence (H)	0.04	0.05	.04	.549	−0.01	0.04	−.01	.839
Semantic Affinity (S)	−0.10	0.05	−.11	.065	0.01	0.04	.01	.778
IS*H	0.04	0.05	.04	.447	−0.06	0.04	−.07	.119
IS*S	−0.05	0.05	−.05	.348	0.04	0.04	.04	.360
Δ R^2 to control model		.00		.441		.01		.176
R^2		.16		<.001		.50		<.001
DV: Preventive behavior (PB)		n = 299				n = 280
PB W1^3	-	-	-	-	0.68	0.04	.71	<.001
Information Seeking (IS)	0.10	0.03	.20	<.001	−0.01	0.02	−.01	.796
Hedonic Valence (H)	0.05	0.03	.11	.052	0.00	0.02	.00	.945
Semantic Affinity (S)	0.09	0.03	.18	.002	−0.01	0.02	−.02	.576
IS*H	−0.03	0.03	−.05	.334	0.01	0.02	.01	.788
IS*S	−0.03	0.03	−.05	.331	0.02	0.02	.04	.330
Δ R^2 to control model		.00		.429		.00		.609
R^2		.18		<.001		.53		<.001

Note. DV = dependent variable, Variables are centered. Cross-sectional models use W1, longitudinal models use W2 DVs. Not all estimates shown, control model includes age, gender, risk group membership, being quarantined, household size, number of children, information seeking, distraction, active coping/planning, media characteristics; For longitudinal model also W1 outcome variables. Coefficients of final models including interactions shown.

We found no support for H4, which predicted an interaction between the media characteristics and distraction in predicting PCE and stress such that using more positive content and content which did not cover the COVID-19 pandemic to distract oneself would lead to less stress and more PCE. For stress, including the interaction effect showed a small increase in explained variance and a significant interaction effect between distraction and positive valence cross-sectionally contrary to our expectation (see Table 2). For individuals who used more positive content to distract themselves, this approach was associated with more stress than for individuals who used more negative content to distract themselves. All other interaction effects were not significant.

Unexpectedly, we found that using positive compared to negative media content for distraction significantly positively predicted PCE at W1 (β = .29, p < .001) and longitudinally (β = .13, p = .028, BH: n.s.; see Table 2).

For RQ2, we explored how hedonic valence and semantic affinity moderated the relationship between problem-focused coping strategies and stress and PCE. We found no significant interaction effects (see Tables 3 and 4). For both problem-focused strategies, using positive media content was related to lower stress (active coping/planning: β = −.18, p = .003, BH: n.s.; information seeking: β = −.21, p < .001) and a higher PCE cross-sectionally (active coping/planning: β = .24, p < .001; information seeking: β = .23, p < .001; see Tables 3 and 4).

Lastly, we examined if the media characteristics moderated the relationship between problem-focused or emotion-focused coping strategies and perceived knowledge and preventive behaviors (RQ3). None of the interactions were significant (see Tables 2–4).

We again found significant direct effects of hedonic valence for some of the analyses. Participants who reported using more positive content for active coping/planning also reported more preventive behaviors longitudinally (β = .12, p = .013, BH: n.s.; see Table 3). Similarly, for emotion-focused coping (i.e., distraction), using positive media content was a significant positive predictor of preventive behaviors cross-sectionally (β = .18, p = .011, BH: n.s.; see Table 2). No significant effects on perceived knowledge emerged.

Discussion

This paper investigated how coping strategies and using media for coping related to mental health and situation-addressing coping effectiveness during a situation of acute crisis. We tested a combination of the goodness-of-fit hypothesis and MMT using a short-term longitudinal study about coping using media during the first COVID-related lockdown in Germany. While our predictions received limited support, the short-term longitudinal design revealed interesting results. Our findings imply a double-edged role of problem-focused coping strategies showing relationships with increased stress but also with increased knowledge and more adherence to preventive behaviors. Emotion-focused coping did not relate to more effective coping. In addition, positive media content may be particularly useful for coping with crises. These findings can provide valuable insights into how individuals can effectively cope with crises.

Consistent with previous research (e.g., Finkelstein-Fox & Park, 2019), the goodness-of-fit hypothesis received partial support. Our research suggests that information seeking as a coping strategy could be beneficial as it is related to increased perceived knowledge and preventive behaviors cross-sectionally. Active coping/planning is similarly related to increased preventive behaviors cross-sectionally. However, information seeking and active coping/planning are also related to more stress cross-sectionally. Other studies conducted during the pandemic yielded similar findings, for example, increased worry due to information seeking (e.g., Liu, 2020).

Our findings thus suggest a two-sided role of problem-focused coping in general and information seeking in particular during circumstances with high degrees of uncertainty: Seeking information is an important prerequisite for taking preventive steps but also an ineffective strategy to reduce stress. Overall, this supports suggestions by stress management specialists to seek information but only during limited, predefined periods of the day (e.g., American Psychological Association (APA), 2020). However, as we did not find longitudinal effects, the direction of the effect remains unclear and might be only very short-term.

Distraction as an emotion-focused strategy was found to be ineffective, contradicting the goodness-of-fit hypothesis. Thus, resembling results of prior research (e.g., Conway & Terry, 1992), predictions about the effectiveness of emotion-focused coping were not supported. Moreover, longitudinally, none of the coping strategies positively predicted any of the coping effectiveness outcomes. Given this mixed support, future research should address boundary conditions for the goodness-of-fit hypothesis, particularly the timeframe in which it applies. As longitudinal associations for a 1-week time frame were not found, even shorter timeframes may be more adequate (Wolfers et al., 2020). Because effective coping is often assumed to translate to longer-term outcomes, future research should analyze the way this transition from short-term to long-term effects occurs.

We found no support for our predictions concerning media characteristics. Semantic affinity and hedonic valence of media content did not strengthen the relationships predicted by the goodness-of-fit hypothesis providing no support for the proposed combination of the goodness-of-fit hypothesis and MMT. The only significant interaction supported an opposite direction, suggesting that distraction is related to more stress when using more positive content.

However, using positive content for coping across several coping strategies is directly related to less stress and more PCE and might have promoted stricter adherence to preventive behavior. Along with similar findings from other studies (see Prestin & Nabi, 2020), this implies the potential of positive media content to directly improve coping outcomes. Positive media content could elicit positive emotions, decreasing stress while simultaneously increasing adaptive behaviors (e.g., Tugade & Fredrickson, 2007). Positive emotions might provide a link between different types of coping outcomes, as they may address affective processes and create capacities to engage in more effortful situation-addressing behaviors simultaneously. This would attribute a direct or mediating but not – as implied here – a moderating role to positive emotions. Consuming positive media content could be an easy-to-implement coping approach deserving of more attention in the field of coping with crises. Recommendations for adjusting to stress during the COVID-19 pandemic did not mention media or recommended not using them (e.g., APA, 2020; CDC (Centers for Disease Control and Prevention), 2023). As relationships were mostly found cross-sectionally, exploratory, and not consistently across all outcomes and strategies, future research is needed. If our results hold in future studies, experts could advise consuming positive media content in crisis situations.

Our findings should be interpreted considering our study’s limitations. First, to test the goodness-of-fit hypothesis, low controllability was assumed but not measured. Thus, some individuals might have felt able to alter their situation. Moreover, while controllability is usually measured as a global measure concerning a situation, controllability could also be measured concerning aspects of a situation. Effective coping could then be considered a way of coping in which the controllable aspects of a stressful transaction are identified and approached. This more nuanced view on controllability could be an interesting direction for research on the goodness-of-fit hypothesis (Person & Frazier, 2024).

Second, our small sample resulted in low power to detect small effects. As we tested a relatively high number of effects, some might be chance findings and did not hold after a correction. Replication is necessary to draw more robust conclusions. Third, some of our measures were not based on previously validated scales. While this was due to the ad-hoc nature of this research, further validating our scales or replication with validated scales is necessary. Finally, for the analyses of the interaction terms, we could not model the measurement error as done in the analyses with the larger sample. While we consider this as the more robust procedure due to our smaller sample, it should be noted as a limitation.

Conclusion and Implications

Utilizing a short-term longitudinal design during the first stage of the COVID-19 pandemic in Germany, this study assessed mental health and situation-addressing coping effectiveness in conjunction and investigated the use of media for coping in a crisis situation. Overall, no support was found for combining the goodness-of-fit hypothesis and MMT. Our results implied a double-edged role of problem-focused coping during crises as it is related to more preventive behavior and more knowledge but also to increased stress and less PCE. Additionally, findings show that positive media content is related to reduced stress, increased PCE, and better adherence to behavioral guidelines. Our results thereby provide evidence for potential benefits of using positive media for mood management beyond mood enhancement. As predictions were mostly found cross-sectionally, the direction and duration of these effects remain unclear.

Future research should therefore focus more on the temporal dynamics of using positive media content for coping during crises. Intervention studies (e.g., Prestin & Nabi, 2020) that prescribe positive media content over a longer time frame as well as studies combining surveys with the assessment of tracked media content, for example, on social media (see, e.g., van Driel et al., 2022) could be particularly valuable to determine under what circumstances positive media content promotes successful coping with crisis.

As to practical implications, our findings support guidance given during the COVID-19 pandemic that in times of crisis, seeking information during limited, predefined periods is an adaptive strategy (e.g., APA, 2020). Moreover, our findings suggest that using positive media content could be an adaptive coping technique. Due to its easy accessibility, increasing exposure to positive media content may be a particularly interesting target for interventions promoting successful coping with crises.

Disclosure Statement

No potential conflict of interest was reported by the author(s).

Additional information

Notes on contributors

Lara N. Wolfers

Lara N. Wolfers is an assistant professor of Youth and Media Entertainment at the Amsterdam School of Communication Research (ASCoR) at the University of Amsterdam. In her research she focuses on how media are used for coping with stress and how digital media are used in the family.

Luca Marie Lüpken

Luca M. Lüpken is a PhD student in Comparative Psychology at Heinrich Heine University Düsseldorf. The focus of her research is on the effects of stress and stress neuromodulators on decision-making in the social and economic domain.

Michelle Schimmel

Michelle Schimmel is a Research Master's candidate in Communication Science at the University of Amsterdam. She is interested in the antecedents and consequences of affective polarisation, how different information environments shape citizens' political beliefs, and computational methods.

Sonja Utz

Sonja Utz is a full professor for communication via social media at University of Tübingen and the head of the Everyday Media Lab at Leibniz-Institut für Wissensmedien in Tübingen. Her research focuses on the effects of social media use, especially in knowledge related contexts, human-machine communication, and algorithm acceptance.

Robin L. Nabi

Robin L. Nabi is Professor of Communication and Digital Media at the University of California, Santa Barbara. Her research focuses on how media can be used to enhance coping and well-being.

Franziska Gaiser

Franziska Gaiser is a media research specialist at the Bavarian Regulatory Authority for New Media in Munich and a PhD student at the Leibniz-Institut für Wissensmedien in Tübingen. Her main research focus is on media usage and human-machine communication, particularly in relation to smart speakers.

Notes

1. Hypotheses were preregistered (see https://aspredicted.org/blind.php?x=vy9gk8). Note that we deviated considerably from the preregistration (see OSF).