Time attitudes affecting psychological health during COVID-19 pandemic: a two-wave, six-month prospective study in Taiwan

Abstract

The COVID-19 pandemic has induced traumatic and fear responses globally. Time attitudes, which refer to one’s feelings toward the past, present and future, may have certain effects on psychological adaptations during this crisis period. This study employed a person-centered approach and a two-wave prospective design to investigate how people with different time attitude profiles change differently in their PTSD symptoms and COVID-19-related fears from a low-risk stage to the first big COVID-19 outbreak in Taiwan. Participants were 354 adults with a mean age of 27.79 years. The result provided support for the theoretical six-factor structure of the traditional Chinese Adolescent and Adult Time Inventory-Time Attitudes Scale (AATI-TA). Four clusters of time attitude profiles were identified (Positives, Negatives, Past Negatives and Pessimists). At both waves, Positives had lower levels of PTSD severity and COVID-19-related fears than most of the other groups, and the reverse was noted for Negatives. As for time effects, people across all profiles were significantly affected during the outbreak, but Negatives showed a greater increase in PTSD severity than other groups. In conclusion, mental health services should put efforts into early identification of those with highly negative time attitudes and implement interventions that nudge people toward a more balanced or positive attitude in each temporal frame, especially during adversity such as the COVID-19 pandemic.

Introduction

Coronavirus disease 2019 (COVID-19) was declared a global pandemic by the World Health Organization (WHO) on March 11, 2020. While many countries experienced a sudden outbreak with an escalation in COVID-19 cases, Taiwan went through about 250 consecutive days without a locally transmitted case from mid-April to late December 2020. It was not until mid-May 2021 when Taiwan faced its first big COVID-19 outbreak. Thereafter, a nationwide level-three epidemic alert and its restriction measures were implemented. People were advised to stay at home and work from home. Schools and all nonessential businesses were closed. The sudden disruptions to daily life, social isolation, and fear of contagion have made psychological health an issue of public concern. It is thus urgent to investigate the psychological impact of the crisis, as well as to examine the resilient and vulnerable factors that contribute to different psychological outcomes (Holmes et al., 2020).

The COVID-19 crisis was considered a type of mass traumatic event (Griffin, 2020; Horesh & Brown, 2020). When confronted with highly infectious, rapidly transmitted, and deadly diseases, posttraumatic stress reactions were common among the general public (Yuan et al., 2021). Indeed, the prevalence rates of PTSD symptoms during the COVID-19 outbreak were profoundly higher than before, and almost one-quarter to one-third of the populations exhibited some PTSD symptoms due to the pandemic (e.g., Chi et al., 2020; Liu et al., 2020; Megalakaki et al., 2021; Qiu et al., 2021; Shevlin et al., 2020; Shi et al., 2020). Unlike the general view that traumatic stress responses were due to past events, evidence revealed that people without direct exposure to the COVID-19 virus still exhibited PTSD symptoms as a result of the present ongoing stress and negative anticipations about the future (Dixon et al., 2022). Present stressful events, such as getting laid off, experiencing lockdown, and having trouble buying supplies, were all predictive of PTSD symptoms during this period (Bridgland et al., 2021). In addition, negative anticipations about contagion and other future negative events, such as fearing that loved ones or oneself would be infected or losing jobs, were also top factors leading to PTSD symptoms (Bridgland et al., 2021; Di Crosta et al., 2020; Messman et al., 2022). Therefore, one’s appraisals of the current situation and anticipations about the future may be particularly important factors that affect psychological reactions during the pandemic.

Another syndrome specifically describes the COVID-19-related anxiety and stress reactions known as the COVID Stress Syndrome (CSS). CSS, proposed by Taylor et al. (2020a), consists of five dimensions, that is, COVID-19 danger and contamination fears, socio-economic concerns, xenophobia, compulsive checking and reassurance seeking, and traumatic stress symptoms about COVID-19. The syndrome was prevalent during the COVID-19 outbreak; for example, it yielded 38% of moderate level and 16% of severe level of CSS in the general population in Canada and the U.S. (Taylor et al., 2020b). People with a higher level of CSS also exhibited greater depression and anxiety symptom severity, and more maladaptive behaviors, such as panic buying, excessive avoidance, and substance abuse (Taylor et al., 2020c, 2021; Taylor, 2021). The dimension of COVID-19 danger and contamination fears, hereinafter referred to as “COVID-19-related fears”, was found to be the core symptom of CSS through network analyses (Taylor et al., 2020b). This implies that, during the COVID-19 crisis, the central concern of the general population that led to tremendous stress and anxiety reactions was worries about being infected. It is speculated that such a worry was related to one’s appraisals of the current situation and anticipations about the future.

If the traumatic stress reactions and COVID-19-related fears during the pandemic were mainly attributed to present and future threats, the way people appraise the current situation and anticipate the future may be particularly important for psychological adaptations during this period. Additionally, because future thinking is closely related to reconstructions of past experiences (Szpunar, 2010), the way people reflect on the past may also have a role. Time perspective (TP; Zimbardo & Boyd, 2015) and time attitudes (Mello & Worrell, 2015) are two well-known temporal psychological constructs that assess individual differences in perceptions and evaluations toward the past, present, and future. Both TP and time attitudes have been proved to be associated with multiple psychological adaptations (Baird et al., 2021; Cole et al., 2017; McKay et al., 2017).

Concerning TP and post-trauma adaptations, Zimbardo and Boyd’s TP model has been the most frequently used in studying the relationship between TP and PTSD. According to their model, TP refers to one’s habitual orientation to a particular time frame (past, present, and future), combined with a specific attitude (positive or negative). Five distinct but correlated dimensions are accordingly composed, including Past-Negative, Past-Positive, Present-Hedonistic, Present-Fatalistic, and Future (Zimbardo & Boyd, 2015; Stolarski et al., 2018). Evidence has revealed that in the aftermath of the September 11 terrorist attacks, people who felt negative about the past, or had a higher level of Past-Negative, were more psychologically distressed and had lower levels of positive affect and life satisfaction than others over the three years following the attacks. On the contrary, people who were future-oriented and goal-directed, or had a higher level of Future TP, revealed the opposite and more favorable outcomes (Holman & Silver, 2005; Holman et al., 2016). Moreover, a more balanced TP profile (i.e., low on Past-Negative and Present-Fatalistic, high on Past-Positive, and moderately high on Future and Present-Hedonistic) was found to be associated with a lower level of PTSD severity after motor vehicle accidents (Stolarski & Cyniak-Cieciura, 2016). Accordingly, when confronting adversity, people who are stuck in the negative past or feel hopeless about the present and future may be more psychologically vulnerable, while those who are present-oriented, goal-directed, and positive about the future be more resilient.

Although Zimbardo and Boyd’s TP model is the most-used temporal psychological framework in studying post-trauma adaptations, there are some limitations. First, the TP model comprises motivational, cognitive, and affective components (Zimbardo & Boyd, 2015), and is composed of a range of level-down constructs, such as time attitudes (Mello & Worrell, 2015) and time orientation (Shipp et al., 2009). Therefore, it is difficult to differentiate the effects of attitudes from those of orientation. Moreover, the emotions that arise when one thinks about the past, present, or future are often determined by the emotional valence of the thought (Smallwood & Andrews-Hanna, 2013), rather than the temporal orientation itself. Therefore, the attitude aspect of TP may be more relevant to psychological adaptations.

Second, positive and negative valence among some temporal frames are missing in this model (Stolarski et al., 2018). Specifically, Present-Hedonistic represents a present-pleasure-seeking attitude, but not positive feelings toward the present. Additionally, Future TP represents a general orientation to the future with goal-directed and planful thinking. However, one can be future-oriented but with a negative view, for instance, worrying or expecting negative outcomes. Indeed, the Future with a negative view, which is lacking in the TP model, may be particularly predictive of distress and anxiety during the pandemic crisis.

Time attitudes, proposed by Mello and Worrell (2015), were built on the work of Zimbardo and Boyd’s TP model, but focus exclusively on the affective aspect of TP. In this construct, positive and negative valence toward the three temporal frames are fully covered. Although there is mounting evidence that time attitudes have strong associations with well-being and psychological symptomology, its effect on psychological adaptations has not been fully studied in the context of adversity or trauma.

Time attitudes are defined as the positive and negative feelings toward the past, present, and future (Mello & Worrell, 2015). The Adolescent and Adult Time Inventory-Time Attitudes Scale (AATI-TA; Mello et al., 2016) is the most popular and psychometrically robust measure of time attitudes to date (McKay et al., 2020), consisting of six subscales, including Past Positive, Past Negative, Present Positive, Present Negative, Future Positive, and Future Negative. Studies using the AATI-TA have revealed that a positive attitude toward each temporal period was associated with favorable psychological outcomes, such as higher levels of social and emotional self-efficacy, well-being, optimism, hope and perceived life chances, and lower levels of perceived stress, anxiety and depression symptoms, while the associations were reversed for negative time attitudes (Cole et al., 2017; Chishima et al., 2019; Donati et al., 2019; Konowalczyk, McKay et al., 2018; McKay et al., 2015; Şahin-Baltacı et al., 2017).

Because an individual holds attitudes toward the past, present, and future simultaneously (Mello & Worrell, 2015), a person-centered approach, rather than a variable-centered approach, is a more reasonable and comprehensive way to study time attitudes and has thus been increasingly adapted by the researchers in this field (e.g., Andretta et al., 2014; Cole et al., 2017; McKay et al., 2017; Worrell, Andretta et al., 2021; Worrell, Mello et al., 2021). To date, adult research applying this approach focused mainly on studying anxiety and depression symptoms. For example, through cluster analyses, Cole et al. (2017) found five clusters of time attitude profiles and named them as Positives, Negatives, Optimists, Pessimists, and Ambivalents. Positives, who had high positive and low negative attitudes toward all temporal periods, reported significantly lower levels of anxiety and depression than most of the other groups, while Negatives, who had high negative and low positive attitudes toward all temporal periods, reported the opposite results. Another research found four clusters of profiles (i.e., Past Negatives, Negatives, Positives, and Pessimists) also revealed that Positives were substantially less depressed than Negatives (McKay et al., 2017). Generally, people with a positive view toward their past, present, and future seem to be the most resilient against the development of psychological symptoms, while people with a whole negative view are the most vulnerable.

In sum, it is speculated that time attitudes can be indicators of psychological adaptations not only in a normal context, but also during adversity, such as the ongoing COVID-19 pandemic. Since the traumatic stress responses during the pandemic were mainly attributed to negative present experiences and negative future anticipations (Bridgland et al., 2021), and also based on the evidence of TP and post-trauma adaptations (Holman et al., 2016; Stolarski & Cyniak-Cieciura, 2016), it is expected that people with a more positive time attitude profile would reveal fewer PTSD symptoms during the pandemic, while those with a more negative profile reveal more PTSD symptoms. Regarding COVID-19-related fears, because the fears were about the present threats and imagined future risks (Taylor et al., 2020a), and based on the evidence that people with a more positive profile were less anxious (Cole et al., 2017), it is speculated that people with a more positive profile would experience a lower level of fears, while those with a more negative profile experience a higher level of fears.

The present study, by employing a person-centered approach and a two-wave longitudinal design, sought to investigate how people with different time attitude profiles varied in PTSD symptoms and COVID-19-related fears under various impacts of the crisis. The baseline data were collected at a mild epidemic phase, and the follow-up data were collected six months later during the first big outbreak in Taiwan. We first developed the traditional Chinese AATI-TA and examined the internal consistency and structural validity of its scores. We hypothesized that the six-factor structure would be supported, and all the subscale scores would be internally consistent. Based on data collected at time one (T1), we expected to discover at least two types of time attitude profiles, namely, the Positive (i.e., high positive and low negative attitudes toward the three temporal periods) and Negative (i.e., high negative and low positive attitudes toward the three temporal periods) profiles. We expected to discover interaction effects between time attitude profiles and measurement time on PTSD symptoms and COVID-19-related fears. Specifically, people with the Negative profile would experience the greatest increase in PTSD severity and COVID-19-related fears from T1 to time two (T2) and vice versa for those with the Positive profile. In addition, regarding the between-group differences at T1 and T2, we hypothesized that people with the Positive profile would have the lowest level of PTSD and COVID-19-related fears at both time points, with the reverse being true for those with the Negative profile.

Method

Participants and procedures

This study was part of a larger prospective project. In this study, data were collected in two waves; T1 was from September 30 to November 24, 2020, and T2 was from June 2 to June 20, 2021. Figure 1 shows the numbers of cases, total deaths, and major events along the timeline. The study used convenience and snowball sampling methods through posts on social media and an online survey. The participants were Taiwanese citizens and at least 20 years old. Confirmed COVID-19 cases, frontline workers, and those who had experienced quarantine were excluded. A total of 576 participants completed the T1 survey and those who agreed to be followed up completed the T2 survey. Of the 354 individuals in the final sample, as shown in Table 1, their mean age was 27.79 years (SD = 6.92; range = 20–62), 65.0% were female, 54.5% were employed, and 92.1% had at least a college degree. Regarding the longitudinal two-wave data, the attrition analyses revealed no difference between the attrition and nonattrition groups in all the demographic variables and scale scores. All procedures were approved by the Institutional Review Board at National Taiwan University (202005HS001).

Fig. 1.

Timeline of COVID-19 Daily Confirmed New Cases, Total Deaths and Major Events in Taiwan

Note. The vertical axis represents the number of cases, and the dotted-line squares represent the period of data collections.

Table 1.

Demographic Characteristics of the Sample (n = 354)

	n/M	%/SD
Gender
Female	230	65.00
Male	124	35.00
Age	27.79	6.92
Highest education level
High school	28	7.90
College	206	58.20
Postgraduate	120	33.90
Employment
Student	125	35.30
Employed	193	54.50
Unemployed/Retired	36	10.20

Measures

Adolescent and adult time inventory-time attitudes scale (AATI-TA)

The AATI-TA (Mello et al., 2016; Worrell et al., 2013) comprises six five-item subscales assessing past positive, past negative, present positive, present negative, future positive, and future negative attitudes. Items are rated on a 5-point Likert scale from 1 (totally disagree) to 5 (totally agree). AATI-TA scores have proven to be internally consistent and structurally valid in many languages (Chishima et al., 2019; Donati et al., 2019; Konowalczyk, Mello et al., 2018; Şahin-Baltacı et al., 2017; Worrell et al., 2013). In the present study, the traditional Chinese version was used and generated through a translation and back-translation process. The original developers (Frank C. Worrell and Zena R. Mello) had confirmed its equivalence with the original version. The Cronbach’s α of the subscale scores ranged from 0.77 to 0.93 in this study.

Posttraumatic diagnostic scale for DSM-5 (PDS-5)

PDS-5 (Foa et al., 2016) was used to assess PTSD symptoms in the past month according to the impact of the COVID-19 crisis. It contains 20 items split into four subscales assessing the symptom clusters of intrusion, avoidance, changes in mood and cognition, and arousal and hyperreactivity. Items are rated on a 5-point Likert scale from 0 (not at all) to 4 (six or more times a week/severe). The traditional Chinese PDS-5 total scores yielded excellent internal consistency among Taiwanese adult samples (αs = 0.94–0.95) (Su et al., 2020). In this study, the Cronbach’s α was 0.95 to 0.96.

COVID stress scales-danger and contamination fears subscale (CSS-DAN)

The COVID Stress Scales (Taylor et al., 2020a) comprises 36 items split into five subscales assessing the COVID Stress Syndrome in the past week. The 12-item danger and contamination fears subscale (CSS-DAN) was adopted to assess COVID-19-related fears in the present study, which assesses worries about the dangerousness of COVID-19 and fears of contact with fomites potentially contaminated with the coronavirus. Items are rated on a 5-point Likert scale from 0 (not at all) to 4 (extremely). The internal consistency was excellent for CSS-DAN scores among Canadian and American general adult samples (αs = 0.94–0.95; Taylor et al., 2020a). The scale had been translated into traditional Chinese in this study, and the Cronbach’s α was 0.94 to 0.95 for CSS-DAN.

Statistical analyses

To evaluate the structural validity of the traditional Chinese AATI-TA, we used confirmatory factor analyses to examine the model fit for the hypothesized six-factor model in comparison with a two-factor valence model and a three-factor temporal model. All analyses were executed using the maximum likelihood estimation method based on raw scores using Amos version 24. Multiple indicators were used to evaluate the models: the Tucker–Lewis index, the comparative fit index, the standardized root mean square residual (SRMR), the root mean square error of approximation (RMSEA) and its 90% confidence interval. Values of both indices indicate an excellent fit at or above 0.95 and an acceptable fit between 0.90 and 0.95 (Byrne, 2008; Hu & Bentler, 1999). Values of SRMR and RMSEA indicate an excellent fit at or below 0.05 and an acceptable fit between 0.05 and 0.08 (Marsh et al., 2004).

We used cluster analyses to generate time attitude profiles based on AATI-TA scores collected at T1. The scores were standardized as T scores (M = 50, SD = 10) before clustering. SAS Enterprise Guide (v8.3) software was used to conduct Ward’s hierarchical clustering and evaluate a set of possible solutions using two stopping rules: pseudo F and pseudo T-squared statistics. K-means iterative partitioning was then applied to validate Ward’s solutions and generate cluster assignments through SPSS (v25) software. Finally, T scores were plotted to examine the distinctions between and across potential profiles.

To investigate the interaction effects between time attitude profiles and measurement time on psychological outcomes, two-way repeated measures analyses of variance (ANOVAs) were performed. The between-subject factors were time attitude profiles, and the within-subject factors were the two time points. Post hoc multiple comparisons were conducted with Bonferroni corrections. For all the analyses, the p values were two-sided, and the statistical significance was set at p < .05. Effect sizes were measured by partial eta squared in ANOVAs. SPSS (v25) software was used to conduct most of the analyses, except for the software specifically mentioned above.

Results

Examination of factor models of the traditional chinese AATI-TA

Results from the confirmatory factor analyses indicated that, as shown in Table 2, most of the fit indices for the two-factor and three-factor models were not satisfactory. The hypothesized six-factor model (model 4) had all the indices within their acceptable range. However, one Future Negative item, item 25, yielded a low standardized coefficient (i.e., 0.22). After eliminating this item, the model fit improved slightly (model 5), but still did not achieve an excellent fit. The modification indices indicated that four pairs of errors were correlating, with each pair belonging to the same temporal dimension, including two Past Positive items (item 3 “I have very happy memories of my childhood” and item 9 “I have good memories about growing up”), two Present Positive items (item 11 “I am pleased with the present” and item 14 “I am content with the present”), two Future Positive items (item 19 “I am excited about my future” and item 28 “Thinking about my future excites me”) and two Future Negative items (item 4 “I doubt I will make something of myself” and item 10 “I don’t think I’ll amount to much when I grow up”). Upon allowing these errors to correlate, the fit indices showed remarkable improvements (model 6), in which all the indices were acceptable or reached an excellent fit. Thus, this modified 29-item, six-factor model, with four pairs of error covariances, was accepted.

Table 2.

Fit Indices of Confirmatory Factor Analyses for Traditional Chinese AATI-TA (Maximum Likelihood Estimation Method, n = 354)

Model		df	TLI	CFI	SRMR	RMSEA	90% CI
1. Null	8988.78***	435
2. 2-Factor (Valence)	3580.38***	404	0.600	0.629	0.130	0.149	0.145, 0.154
3. 3-Factor (Temporal)	1671.70***	402	0.839	0.852	0.063	0.095	0.090, 0.099
4. 6-Factor (Theorized)	1052.73***	390	0.914	0.923	0.049	0.069	0.064, 0.074
5. 6-Factor (Modified-1)a	981.567***	362	0.918	0.927	0.047	0.070	0.064, 0.075
6. 6-Factor (Modified-2)b	766.231***	358	0.946	0.952	0.045	0.057	0.051, 0.062

Note. AATI-TA = Adolescent and Adult Time Inventory-Time Attitudes Scale; TLI = Tucker–Lewis index; CFI = comparative fit index; SRMR = standardized root mean square residual; RMSEA = root mean square error of approximation; CI = confidence interval

^a With 29 items (one Future Negative item, item25, removed)

^b With 29 items (one Future Negative item, item25, removed) and four pairs of correlated errors

*** p < .001

Figure 2 illustrates the standardized path coefficients, single residual variances, and error correlations for the modified six-factor model (model 6). The standardized coefficients ranged from 0.52 to 0.91, indicating very good to excellent. Table 3 shows the descriptive statistics and Cronbach’s alpha values for time attitudes at T1 and outcome variables for both waves. Internal consistencies were all good to excellent. Altogether, the results warranted this study using the six-factor indices to establish time attitude profiles. Accordingly, the final modified 29-item, six-factor model was employed for the following analyses.

Fig. 2.

Modified 29-Item Six-Factor Model for Traditional Chinese AATI-TA Scores

Note. n = 354. All coefficients are standardized maximum likelihood parameter estimates, and significant at the p < .001 level. AATI-TA = Adolescent and Adult Time Inventory-Time Attitudes Scale.

Table 3.

Descriptive Statistics and Internal Consistency Estimates for all Variables (n = 354)

Variable/Assessed time point	Potential range	M	SD	α	Skew	Kurtosis
AATI-TA/ T1
PSP	1–5	3.24	0.78	0.89	-0.38	0.12
PSN	1–5	2.67	0.95	0.92	0.25	-0.56
PRP	1–5	3.24	0.83	0.93	-0.38	-0.12
PRN	1–5	2.94	0.93	0.91	0.06	-0.62
FTP	1–5	3.18	0.85	0.93	-0.48	0.28
FTNa	1–5	2.81	0.89	0.81	0.09	-0.24
PDS-5/ T1	0–80	6.45	10.40	0.96	2.72	8.46
PDS-5/ T2	0–80	14.76	13.42	0.95	1.37	1.99
CSS-DAN/ T1	0–48	11.75	8.76	0.94	0.90	0.22
CSS-DAN/ T2	0–48	21.81	10.65	0.95	0.28	-0.80

Note. Mean scores for the AATI-TA subscales were derived from the item means for the subscales. Mean scores for the outcome variables were based on the scale total scores. AATI-TA = Adolescent and Adult Time Inventory-Time Attitudes Scale; PSP = Past Positive; PSN = Past Negative; PRP = Present Positive; PRN = Present Negative; FTP = Future Positive; FTN = Future Negative; PDS-5 = Posttraumatic Diagnostic Scale for DSM-5; CSS-DAN = danger and contamination fears subscale in the COVID Stress Scales; T1 = time one; T2 = time two.

^a Descriptive statistics and Cronbach’s α for FTN (Future Negative subscale) are based on four items (without item25).

*** p < .001.

Cluster analyses of time attitude profiles

A four-cluster solution emerged as the best among all possible solutions. The profiles were plotted in T scores for each dimension and are presented in Fig. 3. The labeling of the profiles was guided by the criteria of ± 0.5 SDs around the sample mean for each dimension. Profile 1 was labeled as Positives (n = 133, 37.6%) and characterized by substantially above-average positive attitudes and substantially below-average negative attitudes toward all three temporal periods. Profile 2 was labeled as Negatives (n = 47, 13.3%), which was the inverse of Positives. Profile 3 was labeled as Past Negatives (n = 127, 35.9%) and characterized by an above-average past negative attitude and a slightly below-average past positive attitude (≈ 0.47 SDs), with all the other scores close to the mean. Profile 4 was labeled as Pessimists (n = 47, 13.3%) and characterized by above-average past positive and below-average past negative attitudes, with reversed attitudes toward the present and future.

Fig. 3.

Time Attitude Profiles Based on AATI-TA Scores Collected at T1

Note. Profiles were plotted using T scores (M = 50, SD = 10). A score of 0 in the midpoint indicates the average score relative to the sample, which is 50 in T score. AATI-TA = Adolescent and Adult Time Inventory-Time Attitudes Scale; PSP = Past Positive; PSN = Past Negative; PRP = Present Positive; PRN = Present Negative; FTP = Future Positive; FTN = Future Negative.

Changes of PTSD symptoms and COVID-19-related fears from T1 to T2 in individuals with various time attitude profiles

The assumption of sphericity was met in all two-way repeated ANOVAs based on Mauchly’s sphericity test. Table 4 presents the results of ANOVAs. A significant interaction effect was found between time attitude profiles and measurement time on PTSD symptoms. As shown in Fig. 4, while people in all groups of profiles experienced significant increases in PTSD symptoms from T1 to T2 (all significant at p < .001), this time effect was particularly profound on Negatives, suggesting a greater increase in PTSD severity than in all the other groups. Regarding the between-group effects at each time point, Positives had lower PTSD severity than Negatives (p < .001) and Past Negatives (p = .001) at T1, while Negatives revealed greater PTSD severity than all the other groups (Positives, p < .001; Past Negatives, p = .001; Pessimists, p = .007) at T2. Regarding COVID-19-related fears assessed with CSS-DAN, no significant interaction effect was found, but the main effects of time and group were all significant. As for the time effect, people in all groups of profiles experienced significant increases in the severity of COVID-19-related fears from T1 to T2. As for the group effect, Positives showed less fear than Negatives (p = .007) and Past Negatives (p = .001) across time.

Table 4.

Results of Two-Way Repeated Measures ANOVAs for Psychological Outcomes by Time Attitude Profile Across Time

Variable	T1			T2			Effect
	M	SD		M	SD		Time	Profiles	Time × Profiles
PDS-5
Positives	3.32	5.01		11.06	10.69		F(1, 350) = 163.49	F(3, 350) = 12.82	F(3, 350) = 3.13
Negatives	10.74	13.12		23.79	17.92		p < .001	p < .001	p = .026
Past Negatives	8.19	12.43		15.19	13.01		= 0.318	= 0.099	= 0.026
Pessimists	6.34	10.38		15.06	12.10
CSS-DAN
Positives	9.53	7.63		18.92	10.10		F(1, 350) = 412.44	F(3, 350) = 6.58	F(3, 350) = 2.32
Negatives	13.68	9.19		24.28	11.81		p < .001	p < .001	p = .075
Past Negatives	13.58	9.38		23.05	10.71		= 0.541	= 0.053	= 0.020
Pessimists	11.17	8.31		24.17	9.12

Note. PDS-5 = Posttraumatic Diagnostic Scale for DSM-5; CSS-DAN = danger and contamination fears subscale in the COVID Stress Scales; T1 = time one; T2 = time two

Fig. 4.

Interaction Effects for PTSD by Time Attitude Profile Across Time

Note. PDS-5 = Posttraumatic Diagnostic Scale for DSM-5.

Discussion

The present study is the first to examine the role of time attitude profiles in psychological responses under the context of adversity, specifically, the COVID-19 pandemic. We applied a person-centered approach and a two-wave prospective design to investigate the differences in PTSD symptoms and COVID-19-related fears among various time attitude profiles along different pandemic stages in Taiwan. The results provided support for the six-factor structure of the traditional Chinese AATI-TA. Based on the six-factor indices, four groups of interpretable time attitude profiles emerged (i.e., Positives, Negatives, Past Negatives, and Pessimists). As the severity of the pandemic increased from T1 to T2, Negatives experienced a greater increase in PTSD severity than all the other groups. Additionally, at both T1 and T2, Positives revealed lower levels of PTSD severity and COVID-19-related fears than most of the other groups, and the reverse was noted for Negatives.

Concerning the structural validity of the traditional Chinese AATI-TA, the six-factor model provided the best fit to the data, which is consistent with previous findings (Alansari et al., 2013; Donati et al., 2019; McKay et al., 2015; Mello et al., 2016; Worrell et al., 2013). However, the final six-factor model excluded one Future Negative item (i.e., item 25 “Thinking ahead is pointless”) because of its low factor loading. Four pairs of error covariances were also added in the final model. Similar modifications have been made in some studies. For instance, item 25 showed weak factor loadings in Turkish and Italian versions (Şahin-Baltacı et al., 2017; Worrell, Mello et al., 2021). This item may not be a good indicator of future negative attitude in some languages or cultures. Concerning error covariances, some studies also found that the six-factor model fitted the data better after adding error covariances (Chishima et al., 2019; Donati et al., 2019; Mello et al., 2019). In this study, each pair of the added correlated error terms had similar wordings and were within the same factor, which may be the causes of the covariances. Future research with different samples is needed to evaluate whether these issues are generalizable or only occur in specific samples, languages, or cultures.

Four clusters of time attitude profiles were found in this study. Of all, Positives and Negatives, two of the most often identified clusters in previous studies, emerged as expected. The proportions of both clusters (37.6% for Positives and 13.3% for Negatives) are also similar to previous findings (Andretta et al., 2014; Alansari et al., 2013; McKay et al., 2017; Worrell, Andretta et al., 2021 ). It seems that people are more likely to have a whole positive view toward their past, present, and future than a whole negative view. The other two clusters in this sample, Past Negatives and Pessimists, had also been identified in other studies (Cole et al., 2017; McKay et al., 2017; Worrell, Andretta et al., 2021). Past Negatives had negative past attitudes, but balanced present and future attitudes. These individuals had likely experienced some adverse events in the past, but somehow their present and future attitudes were not negatively skewed by those experiences. Somewhat opposite to Past Negatives, Pessimists were positive about their past but felt negative about the present and future. Among all the profiles, Positives were the only group holding strong positive attitudes to the three temporal frames, whereas Negatives were the only group holding a strong negative view to all temporal frames.

For the outcomes among various profiles, an interaction effect was found between time attitude profiles and measurement time on PTSD symptoms. Although people in all groups of profiles experienced a significant increase in PTSD symptoms from T1 to T2, Negatives showed an even greater increase. These findings support our hypotheses. As the risk of the pandemic increased, those who were highly negative about the past, present, and future were the most traumatized. Past Negatives, who were only negative about their past, and Pessimists, who were pessimistic about their present and future, were all less affected than Negatives. One possible reason is that their negative attitudes were not as strong as those of Negatives. Another reason may be that both Past Negatives and Pessimists had at least a balanced or positive time attitude to a temporal frame that buffered the traumatic impact of the crisis. This buffering effect may come from the balanced present and future attitudes for Past Negatives and the positive past attitude for Pessimists. Thus, attitude to the past, present, and future may all have a role in PTSD symptomology when confronting a crisis, in which a more balanced or positive time attitude is associated with fewer traumatic reactions.

The hypotheses regarding the traumatic responses for Positives are partially supported. Although Positives had stronger positive attitudes about the past, present, and future than all the other groups, they still unexpectedly experienced a significant increase in PTSD symptoms right after the outbreak. This implies that even someone with highly positive time attitudes, can still be affected right after an outbreak of a deadly disease and become more distressful and anxious than they used to be. However, in comparison with other groups at T1 and T2, Positives were less traumatized than most of the other groups at both time points as expected. On the contrary, Negatives were more traumatized than most of the other groups. Thus, though people with a Positive time attitude profile may exhibit traumatic responses in a crisis, they still tend to be more psychologically resilient than those with other profiles.

The above findings extend prior knowledge in several ways. Previous evidence has revealed that time perspective (TP) could influence post-trauma adaptations, in which positive and future-oriented thinking or a more balanced TP profile was associated with better outcomes (Holman & Silver, 2005; Holman et al., 2016; Stolarski & Cyniak-Cieciura, 2016). By separating the attitude aspect of TP from other aspects, such as time orientation, this study provides novel evidence that time attitudes alone can influence PTSD symptomology. Specifically, without considering one’s habitual time orientation, people with a more positive or balanced time attitude profile reveal fewer traumatic stress reactions during a crisis. Furthermore, for the study field of time attitudes, because most of the previous work was done in a normal context or only included depression and anxiety symptoms as outcomes (Cole et al., 2017; McKay et al., 2017), this study further contributes to this field through providing the first evidence that time attitudes are associated with PTSD symptomology during adversity.

The hypotheses regarding COVID-19-related fears are partially supported. No interaction effect was found. People in all groups of profiles experienced a significant increase in COVID-19-related fears from T1 to T2. Interestingly and unexpectedly, Positives and Negatives experienced a similar amount of increase. Based on the evidence that a Positive profile was associated with a lower level of anxiety symptoms (Cole et al., 2017), Positives were expected to experience a lower amount of increase in fears. However, according to our result, the protective effect of a Positive profile for anxiety and fear responses may not be as profound as previously thought, especially in a context of a pandemic outbreak. Indeed, COVID Stress Syndrome, whose core symptoms were COVID-19-related fears, was suggested to be conceptualized as a pandemic-related adjustment reaction (Taylor, 2021). Hence, COVID-19-related fears may be common psychological first responses for most people, regardless of their positive attitudes.

The findings of between-group differences in COVID-19-related fears are more in line with our expectations. In comparison with other groups, Positives had a lower level of fears across time, while Negatives had a higher level of fears. Therefore, although people with highly positive time attitudes may not be immune to contamination fears when facing a pandemic outbreak, their level of fears tend to be lower compared to people with other profiles.

Overall, our results suggest that people who are highly negative about their past, present, and future are the most vulnerable when confronting a crisis. A more positive time attitude profile is more likely to buffer the impact of a crisis. However, during a serious and ongoing crisis, such as a pandemic outbreak, traumatic and fearful responses seem to be common for most people. Even those who are usually positive and optimistic can be affected to a certain degree.

Strengths and implications

There are several strengths and implications of this study. First, because people hold all six dimensions of time attitudes simultaneously, unlike most of the research that only examined the independent relationship between each dimension of time attitudes and psychological outcomes (Chishima et al., 2019; Donati et al., 2019; Konowalczyk, McKay et al., 2018), this study, by adopting a person-centered approach, considers all the dimensions together. The results extend the current understanding of how all six dimensions of time attitudes interact to affect one’s psychological adaptations during a crisis. Specifically, it reveals that people who are highly negative toward all temporal frames are the most vulnerable during a crisis, and those who are highly positive toward all temporal frames are the most resilient. In addition, having at least a balanced or positive attitude toward a temporal frame may lower the risk of developing psychological symptoms.

The second merit of this study is that it applies a two-wave prospective design. Most of the research used a cross-sectional design and only confirmed the concurrent effects of time attitudes on psychological health (Andretta et al., 2014; Cole et al., 2017; McKay et al., 2017; Worrell, Andretta et al., 2021). This study provides further evidence that a more positive time attitude profile is related to better future psychological health, at least for a half year. The reverse result is also true for a more negative profile.

Lastly, this study is one of a few that has investigated the effects of time attitudes on psychological outcomes in the context of adversity, specifically, during a pandemic outbreak. According to the results, time attitudes are not only influential in psychological health in a normal context, but also traumatic stress and fearful reactions during a crisis.

There are also a number of practical implications. With the results suggesting that individuals with a more negative time attitude profile tend to be more traumatized and fearful during a crisis, the AATI-TA may be a useful screener for early identification. Intervention programs that help to change time attitudes and develop more adaptive views about the past, present, and future could then be implemented. Even a change in attitude toward a single temporal frame may be beneficial to some extent. In this study, people with all clusters of profiles were significantly affected by the pandemic. Advocacy on the importance of public mental health and programs for public education and services should be in order, especially while the whole world is still at the peri-pandemic stage.

Limitations and future directions

Some limitations of this study should be considered. First, the sample consisted of more female, young adults, and highly educated participants, while those who had experienced quarantine or been confirmed to have COVID-19 were excluded. Thus, the sample was not nationally representative. The generalizability of the results should be taken with caution. Future research can consider using randomized sampling methods to verify our findings. It would also be informative to investigate whether time attitude profiles have a role in adaptations among those exposed more in a crisis, such as confirmed cases or frontline workers during a pandemic. Second, this study did not include any time orientation measure. Future research may assess both time attitudes and time orientation to examine if these two central components of TP have different effects on psychological adaptations in a crisis. Third, concerning time attitude profiles, person-centered analyses are indeed data-driven. As a result, whether the profiles identified in this sample can be replicable for other samples in the same population is yet to be examined. More research is needed to replicate our findings. Additionally, some clusters identified in previous studies did not emerge in our sample, such as Ambivalents (i.e., people with balanced attitudes to all temporal frames; Cole et al., 2017) and Optimists (people who are highly negative to the past, but highly positive about the future; Alansari et al., 2013). Research with larger and more diverse samples could provide the opportunity to compare outcomes among more profiles. Moreover, it may be promising to investigate some preceding factors, such as personality, social, or cultural factors, that contribute to the evolvement of different time attitude profiles. Fourth, the follow-up data in this study were collected right after the outbreak, and the psychological reactions may thus be acute. How people with different time attitude profiles adjust in the long run remains unknown. Future work may track a longer period with multiple assessment time points to evaluate the long-term associations between time attitudes and psychological adaptations during a crisis. Lastly, this study reveals the relationships between time attitudes and psychological adaptations during a pandemic outbreak. Whether time attitudes could also have a role in psychological health under the impacts of other crises, such as earthquakes or traumatic interpersonal events, is a worthwhile topic of future investigation.

Conclusion

The present study first verified the traditional Chinese version of AATI-TA, and via a two-wave prospective design, examined PTSD symptomatology and COVID-19-related fears, as well as symptom changes in individuals with various time attitude profiles before and during the first big COVID-19 outbreak in Taiwan. Using a person-centered approach, four profiles were identified. In comparison with the first assessment before the outbreak, people in all groups of profiles were significantly affected during the outbreak, and especially those with the Negative time attitude profile experienced a greater increase in PTSD symptoms over time. Altogether, this study and its findings extend the current understanding regarding the role of time attitude profiles on psychological symptoms during adversity, especially an unparalleled trauma-like event.

Author Contribution

Wei-Chun Chiang: Conceptualization, Methodology, Formal analysis, Investigation, Writing - Original Draft, Writing - Review & Editing, Project administration, Funding acquisition. Sue-Huei Chen: Conceptualization, Writing - Review & Editing, Supervision, Project administration, Funding acquisition.

Funding

This work was supported by joint grants from the Ministry of Science and Technology of Taiwan (109-2420-H-002-007, 110-2420-H-002-005, and 108-2926-I-002-002-MY4) and Ministry of Education of Taiwan (110L9A00703).

Data Availability

The datasets generated during and/or analyzed during the current study are available in the Open Science Framework repository, https://doi.org/10.17605/OSF.IO/H86GY.

Declarations

Competing interests

The authors have no competing interests to declare.

Ethics approval

The questionnaire and methodology for this study was approved by the Institutional Review Board at National Taiwan University (Ethics approval number: 202005HS001).

Informed consent

Informed consent was obtained from all participants included in the study.