Fear of COVID‐19 Scale (FCV‐19S) across countries: Measurement invariance issues

Abstract Aim The threats of novel coronavirus disease 2019 (COVID‐19) have caused fears worldwide. The Fear of COVID‐19 Scale (FCV‐19S) was recently developed to assess the fear of COVID‐19. Although many studies found that the FCV‐19S is psychometrically sound, it is unclear whether the FCV‐19S is invariant across countries. The present study aimed to examine the measurement invariance of the FCV‐19S across eleven countries. Design Cross‐sectional study. Methods Using data collected from prior research on Bangladesh (N = 8,550), United Kingdom (N = 344), Brazil (N = 1,843), Taiwan (N = 539), Italy (N = 249), New Zealand (N = 317), Iran (N = 717), Cuba (N = 772), Pakistan (N = 937), Japan (N = 1,079) and France (N = 316), comprising a total 15,663 participants, the present study used the multigroup confirmatory factor analysis (CFA) and Rasch differential item functioning (DIF) to examine the measurement invariance of the FCV‐19S across country, gender and age (children aged below 18 years, young to middle‐aged adults aged between 18 and 60 years, and older people aged above 60 years). Results The unidimensional structure of the FCV‐19S was confirmed. Multigroup CFA showed that FCV‐19S was partially invariant across country and fully invariant across gender and age. DIF findings were consistent with the findings from multigroup CFA. Many DIF items were displayed for country, few DIF items were displayed for age, and no DIF items were displayed for gender. Conclusion Based on the results of the present study, the FCV‐19S is a good psychometric instrument to assess fear of COVID‐19 during the pandemic period. Moreover, the use of FCV‐19S is supported in at least ten countries with satisfactory psychometric properties.


| INTRODUC TI ON
The threats and consequences of the novel coronavirus disease 2019 (COVID-19) to individual's health and related aspects have been investigated in many different ways, including their psychological health and behaviours from individual and government perspectives (Lin & Cheng, 2020;Rieger, 2020;Shrivastava & Shrivastava, 2020). In addition to the risks of death and serious consequences due to COVID-19 infection, scholars and healthcare professionals have identified the need to assist different populations in tackling mental health difficulties (Holmes et al., 2020;Islam et al., 2020;Usman et al., 2020). More specifically, individuals may have elevated psychological distress and perform inappropriate life-threatening behaviours induced by the elevated distress due to the COVID-19 pandemic (Dsouza et al., 2020;Lin, 2020;. In order to respond to the need of assessing mental health issues, several research teams have developed different instruments to understand the psychological response to COVID-19 (Ahorsu, Lin, Imani, et al., 2020;Lee, 2020aLee, , 2020bTaylor et al., 2020). These instruments include the: (a) five-item Coronavirus Anxiety Scale (CAS) (Lee, 2020a), (b) four-item Obsession with COVID-19 Scale (OCS) (Lee, 2020a), (c) 36-item COVID Stress Scale (CSS) (Taylor et al., 2020) and (d) seven-item Fear of   (Ahorsu, Lin, Imani, et al., 2020;. Moreover, Ransing et al. (2020) conducted a rapid review to summarize the features of these four instruments. Ransing et al. (2020) indicated that one of the most important issues for these instruments was the need to translate, culturally adapt, assess and validate the existing instruments to achieve the maximum utility. Pakpour, Griffiths, Chang, et al. (2020) responded to Ransing et al. (2020) and demonstrated that the FCV-19S had strong features in its brevity with satisfactory psychometric properties shown in different language versions . Indeed, in 3 months of the original Persian FCV-19S being published (Ahorsu, Lin, Imani, et al., 2020;, the scale had been validated in English [in the UK (Harper et al., 2020), in New Zealand (Winter et al., 2020)], Arabic (Alyami et al., 2020), Bangla (Sakib et al., 2020) Italian (Soraci et al., 2020), Hebrew (Bitan et al., 2020), Russian (Reznik et al., 2020), Turkish (Satici et al., 2020), Chinese , Urdu, Malay (Pang et al., 2020), Brazilian Portuguese (Abad et al., 2020), Cuban Spanish (Broche-Pérez et al., 2020) and Greek (Tsipropoulou et al., 2020 for Greek); and construct validity (supported unidimensional or twofactor structure in either confirmatory factor analysis or exploratory factor analysis across all language versions). Although most of the studies conducted to date have reported a unidimensional structure, a couple of studies have reported a two-factor structure Ransing et al., 2020).
Although a two-factor structure has been proposed and tested, the two-factor structure does not have the theoretical background to support it . More specifically, the original FCV-19S (Ahorsu, Lin, Imani, et al., 2020; was developed using Protection Motivation Theory (Rogers, 1975), and was identified as having a single-factor structure using exploratory factor analysis (Ahorsu, Lin, Imani, et al., 2020; with the single-factor structure verified in confirmatory factor analysis (Alyami et al., 2020;Pakpour, Griffiths, Chang, et al., 2020;Sakib et al., 2020;Satici et al., 2020;Soraci et al., 2020;Tsipropoulou et al., 2020). Therefore, the present authors believe that the FCV-19S should have a single-factor structure across different language versions. However, at the time of writing, no studies have examined the measurement invariance of the FCV-19S to verify whether its factor structure is equivalent across different subgroups, including different language versions. Additionally, it is still unclear whether different subgroups (e.g. different ethnic populations, different genders and different age groups) interpret the FCV-19S with similar considerations. Therefore, this is an important missing aspect in the extant literature and the present authors attempted to answer the research question of whether individuals from different countries interpret FCV-19S items similarly.
Consequently, measurement invariance is an important issue for an instrument to satisfy the aforementioned question (i.e. whether different subgroups interpret FCV-19S items similarly). If the psychometric testing on measurement invariance supports the invariance across subgroups, this indicates that individuals in the subgroups interpret the instrument concept and content (e.g. FCV-19S in the present study) in the same way (Limbers et al., 2008;Lin et al., 2013). Moreover, with the use of measurement invariance, the underlying concept can be compared in a more accurate way than using the comparison with observed scores (Vandenberg & Lance, 2000).
For example, some studies have used observed quality of life (QoL) scores (i.e. summing up all the item scores) to compare the quality of life between groups (Bodur & Cingil, 2009;Su et al., 2013Su et al., , 2014. However, this practice of comparisons does not consider any measurement errors or measurement weights. In other words, such comparisons using observed QoL scores are not accurate. In contrast, comparisons using the latent scores with the consideration of measurement invariance, instead of the observed scores, tackle the aforementioned measurement issues . Therefore, testing measurement invariance is important for an instrument to help healthcare providers and researchers meaningfully compare an underlying concept (e.g. fear of COVID-19 in the FCV-19S) between subgroups.
The present study asserts that the FCV-19S should be evaluated for its measurement invariance in three aspects: different ethnicity, gender and age (i.e. children aged below 18 years, young to middleaged adults aged between 18 and 60 years, and older people aged over 60 years). Cultural differences and the different actions and policies made by different governments internationally may make different ethnic populations respond differently to FCV-19S items.
For example, Western people as compared with Eastern people are prone to individualism (Dubois & Beauvois, 2005;Harkness et al., 2000). Therefore, Western people intend to respect personal freedom whenever such freedom does not break the laws. In contrast, Eastern people are more collectivist (King & Bond, 1985;Tsai et al., 2015) and care more about the harmony in the community. Therefore, Eastern people may be more likely to perform behaviours that the society wishes even though such behaviours are not required by the legislation. As a result, Rieger (2020) found that a sample of European students intended not to wear mask if there was no legislation.  (Chen, Jyan, et al., 2020;Lin & Cheng, 2020).
With respect to gender, prior evidence has shown that females (as compared with males) tend to be more sensitive to stress and, therefore, usually have higher levels of fear when encountering various life events (Tolin & Foa, 2006;Vlassoff, 2007). Therefore, it is not known whether males and females interpret items in the older people are more vulnerable to COVID-19 infection and usually have a more serious prognosis than younger cohorts (Dariya & Nagaraju, 2020;Moccia et al., 2020). Therefore, relative to older people, younger people may not be as aware of the seriousness of COVID-19 and have little in the way of a psychological response to COVID-19.
In order to fully understand the interpretation of FCV-19S items among different ethnic populations, different genders and different age groups (children, young to middle-aged adults and older people), the present study used data derived from ten countries to examine the measurement invariance of the FCV-19S. The single-factor structure of the FCV-19S was re-examined in the large sample from a diverse cultural background. More specifically, the present study compared the FCV-19S scores between ten countries comprising Bangladesh, United Kingdom, Brazil, Taiwan, Italy, New Zealand, Iran, Cuba, Pakistan and Japan. Additionally, the FCV-19S scores between genders and age groups were assessed.

| Participants and recruitment procedure
The present study included the datasets from ten countries that have validated the FCV-19S in their respective countries. A short sampling description is given herewith, details can be found in the original papers (Abad et al., 2020;Broche-Pérez et al., 2020;Chang, Hou, et al., 2020;Harper et al., 2020;Mailliez et al., 2021;Masuyama et al., 2020;Pakpour, Griffiths, Chang, et al., 2020;Sakib et al., 2020;Soraci et al., 2020;Winter et al., 2020). More specifically, all the participants used in the present study were recruited through convenience sampling. Some were recruited using online surveys and some were recruited using paper-based (offline) surveys because most of the validations were carried out independently by different research teams and the respective teams had different resources in the different countries. However, there was no serious bias in using the two types of survey data collection and there is prior evidence showing that online and offline surveys are measurement invariant (Martins, 2010). All the study designs were cross-sectional.
Moreover, general populations were the target sample in most of the countries (Table 1). Table 1 also reports the data collection period for each country and a related figure concerning COVID-19 infection at the time of the study.

| Data analysis
The participants' age, gender distribution (male, female, and other),  configural models to examine whether different subgroups shared the same factor loadings. Finally, CFA models with factor loadings and item intercepts constrained equally across subgroups were constructed and compared with the models with factor loadings constrained equally to examine whether different subgroups shared the same item intercepts. ΔCFI > −0.01, ΔRMSEA < 0.01 and ΔSRMR < 0.01 support the full measurement invariance in every two nested models' comparisons (Lin et al., 2019). However, if the full measurement invariance was not achieved, partial invariance was tested using the process of relaxing factor loadings or item intercepts in the constrained models. Moreover, the data relating to "other" gender was not used for DIF or multigroup CFA because there were only 27 participants reporting their gender as other.
Given the huge difference in sample sizes (27 "other" gender, 7,723 male gender, and 8,363 female gender), carrying out invariance testing on such a small sample size would be problematic.
A model with structural equation modelling (SEM) was then constructed to examine the associations between age, gender, and fear of COVID-19. In the SEM model, young to middle-aged adults aged between 18 and 60 years and being male were reference groups.

| Measurement invariance and factor loading findings for the FCV-19S
Differential item functioning contrasts across different countries, gender and age groups are presented in Abbreviations: Infit MnSq, information-weighted fit mean square; Outfit MnSq, outlier-sensitive fit mean square.

TA B L E 3 Item properties of the Fear of COVID-19 Scale (FCV-19S)
the multigroup CFA. Only partial invariance was supported for the FCV-19S across countries (with the relaxed factor loadings of items F2 and F3; relaxed item intercepts of items F1 and F3 to F6). About gender and age groups, full invariance was supported for the FCV-19S. However, the ∆CFI, ∆RMSEA and ∆SRMR were larger in the multigroup CFA across age groups than in the multigroup CFA across gender (Table 6). Table 7 presents the factor loadings and item intercepts of the FCV-19S items across countries, gender and age groups in the con-

| D ISCUSS I ON
In order to respond to the need of assessing mental health difficulties and associated behaviors (e.g. problematic use of the internet,  . In contrast, New Zealand and Taiwan had good strategies to eliminate the has an impact of COVID-19 (Chang, Strong, et al., 2020;Cheng et al., 2020;Winter et al., 2020 Another explanation is cultural differences. Western people embrace individualism (Dubois & Beauvois, 2005;Harkness et al., 2000) and respect their freedom substantially. Therefore, whenever a behaviour is not prohibited by the legislation (e.g. wearing mask), Western people are not likely to violate their will to perform this behaviour. In contrast, Eastern people in collectivism cultures are prone to satisfy the community harmony. Therefore, even a simple behaviour such as wearing mask is not required by the laws, Eastern people are likely to perform this behaviour to align themselves to the society norm (King & Bond, 1985;Tsai et al., 2015).
Apparently, Westerners and Easterners have different attitudes toward hygiene behaviours (e.g. Westerners tend not to wear face masks) (Rieger, 2020)  in a similar way irrespective of their gender and age. Further comparisons on the FCV-19S between gender and age indicate that females had greater fear of COVID-19 than males, and young to middle-aged adults had greater fear than children and older people. The higher fear found among females can be explained by their higher sensitivity toward stress than males (Tolin & Foa, 2006;Vlassoff, 2007).
With the high sensitivity toward stress (e.g. COVID-19 pandemic), females are likely to develop greater fear than males.
Children as compared with young to middle-aged adults showed lower levels of fear toward COVID-19. This may be explained by the different perceptions of COVID-19. As children are not like young to middle-aged adults in encountering difficult challenges resulting from COVID-19 (e.g. financial burden), children may not consider COVID-19 a serious problem and therefore do not have high levels of fear toward COVID-19 (Chen, Jyan, et al., 2020). Surprisingly, older people as compared with young to middle-aged adults also had lower levels of fear toward COVID-19. This finding may be seen as surprising given the fact that older people with COVID-19 have higher mortality rates than young to middle-aged adults (Dariya & Nagaraju, 2020;Moccia et al., 2020). However, it may be that older people may consider that they have little to lose as they have already

TA B L E 5 (Continued)
had relatively long lives. Therefore, they may have made more preparations in relation to their own death than young to middleaged adults and subsequently lowered their fear of COVID-19.
However, additional evidence is needed to examine such speculation. Moreover, although the present findings showed a lower fear of COVID-19 among children and older people, but healthcare providers and policymakers should not ignore the importance of mental well-being in these two populations. More specifically, prior evidence has shown the need to tackle their mental health issues during COVID-19, including increased psychological distress, sleep problems, and problematic use of the internet Chen et al., in press;.
In addition, different factor loadings were observed across countries, genders, and age groups (Table 7). More specifically, most of the factor loadings were strong and consistent (i.e. between 0.5 and 0.9) across the countries for all the seven FCV-19S items, except for Items F1 (loading = 0.471) and F2 (loading = 0.373) in Iran. Therefore, Iranians might be interpreting words such as "afraid" and "uncomfort- observed scores Vandenberg & Lance, 2000). Note: The loadings and intercepts were calculated using constrained models (M5 in Table 5 for country; M8 in Table 5 for gender).
a Relaxed for factor loadings. b Relaxed for item intercepts.
to nursing students at all levels and areas. The present study has the following limitations. First, all the data used in the present study were collected using convenience sampling methods. Therefore, the representativeness of the participants in each ethnic population is low. Also, the characteristics between the ten ethnic populations were not directly comparable and the comparisons of their FCV-19S scores are somewhat biased.
For example, the New Zealanders were aged below 20 years and the Taiwanese were aged above 50 years. With a 30-year of difference in age, the comparison of FCV-19S scores between New Zealanders and Taiwanese is obviously biased by age. Similarly, the Italian sample comprised extremely few males (8.0%) and the FCV-19S score obtained was, therefore, more representative of females.
Second, there were no other psychometric instruments used in the present study. Therefore, the present study could not examine how the FCV-19S associated with other validated instruments and the concurrent validity of the FCV-19S could not be concluded from the present findings. Third, the data collection periods were not comparable across the countries. With different policies and procedures to inhibit the spread of COVID-19, and different numbers of cases and deaths, individuals' fear of COVID-19 may have been different due to these factors. The changes of fear of COVID-19 may thus be a potential confounding factor for the present study.
Finally, given the importance of LGBT+ community, it is important to assess whether the FCV-19S is invariant across "other" gender compared to male and female genders. Unfortunately, the present sample had too few participants of "other" gender to reliably calculate measurement invariance in this group. Future studies are, therefore, needed to recruit a large enough sample size of other gender to provide robust analysis on this important gender variant.

| CON CLUS ION
Based on the results of the present study, the FCV-19S is a good psychometric instrument to assess fear of COVID-19 during the pandemic period. Moreover, the use of FCV-19S is supported in at least ten countries with satisfactory psychometric properties.
However, only partial invariance rather than full measurement invariance of the FCV-19S was supported across the ten ethnic populations. Therefore, future studies on FCV-19S comparisons across different ethnic populations should be cautious with the measurement non-invariance. Such studies may need to consider the use of multigroup CFA rather than simply summing up the FCV-19S scores if they want to make the comparisons. Nevertheless, the full measurement invariance of the FCV-19S was supported across gender and age groups. Therefore, future studies can reliably use the FCV-19S to compare the fear of COVID-19 between gender and age groups.

ACK N OWLED G EM ENTS
The authors would like to thank the participants of the study who generously gave their time and shared their experiences.

CO N FLI C T O F I NTE R E S T
The authors declare that they have no conflict of interest.

AUTH O R CO NTR I B UTI O N S
All authors: Conceptualization, data analysis, final draft collection, data collection, data analysis, data interpretation, final discussion and review of the article.

E TH I C A L A PPROVA L
All procedures performed in this study involving human par-

DATA AVA I L A B I L I T Y S TAT E M E N T
The authors elect not to share data. Research data are not shared for ethical and confidentiality reasons.