Translation of the Fear of COVID-19 Scale into French-Canadian and English-Canadian and Validation in the Nursing Staff of Quebec

Gélinas, C., Maheu, C., Lavoie-Tremblay, M., Richard-Lalonde, M., Gallani, M. C., Gosselin, É., Hébert, M., Tchouaket Nguemeleu, E. & Côté, J. (2021). Translation of the Fear of COVID-19 Scale into French-Canadian and English-Canadian and Validation in the Nursing Staff of Quebec. Science of Nursing and Health Practices / Science infirmière et pratiques en santé, 4(1), 1–23. https://doi.org/10.7202/1077985ar Résumé de l'article


INTRODUCTION
As of April 27, 2021, the province of Quebec registered more than 345 000 people infected with, and over 10 000 deaths from, the coronavirus disease , being one of the most affected provinces in Canada (Gouvernement du Québec, 2021; Public Health Agency of Canada, 2021a). During the first wave of the pandemic in Quebec in the spring of 2020, 3890 out of 5629 deaths (69.1%) were from older adults of residential and long-term care centres (Protecteur du citoyen, 2020). The province of Quebec went through a second wave of the pandemic in the fall of 2020 and the winter of 2021, and a third wave has started at the end of March 2021 (Gouvernement du Québec).
Nursing staff, such as nurses and licensed practical nurses (LPNs), are at the frontline of the effort to prevent and treat the spread of the SARS-CoV-2, having to work over twice the required overtime compared to the previous year (Carrière et al., 2020). Of the confirmed positive COVID-19 cases reported by the Public Health Agency of Canada for Quebec as of March 2021, 15% were healthcare workers, with 3 reported deaths (Public Health Agency of Canada, 2021b).
Fear is a core emotional response to the perceived threat of COVID-19 during the current pandemic (Van Bavel et al., 2020). Fear of COVID-19 was associated with symptoms of stress in frontline nurses (Labrague et al., 2020) and healthcare workers (Khanal et al., 2020). The Fear of COVID-19 Scale (FCV-19S) was developed to measure the level of fear that people might experience toward COVID-19 (Ahorsu et al., 2020). The original version is in Iranian, but it was also made available in English by the authors. So far, the FCV-19S has been translated, adapted and validated in over 35 countries and in at least 20 languages, sometimes by more than one research group for the same language (see Table 1 and Table  2 in Appendix A). Studies have reported good internal consistency, with Cronbach alphas ranging from 0.68 (Reizer et al., 2020) to 0.93 (Bharatharaj et al., 2020;Chang et al., 2020). FCV-19S mean scores varied across populations, from 13 for the general population in Norway (Iversen et al., 2021) to 27 in a general population in Iran (Ahorsu et al.). Moderate fear levels (mean=19.92, SD=6.15) were reported in 261 frontline nurses from the Philippines (Labrague et al., 2020). Similar fear levels were also found in other studies with healthcare workers, including nurses from Mexico (Garcia-Reyna et al., 2020) and Turkey (Saracoglu et al., 2020).
According to nurses' work characteristics, the study by Labrague and colleagues (2020) reported higher fear scores in part-time nurses compared to those working full-time, and in nurses who did not attend a COVID-19 training. In the same study, a negative association was found between fear of COVID-19 and job satisfaction, and a positive association was found with turnover intention (Labrague et al.).
At the time we proceeded with the translation of the scale (May-June, 2020), no other French version was available. In July 2020, a French version from France was made available in a preprint manuscript (Mailliez et al., 2020). French-Canadian is noticeably different from French spoken in France in addition to the cultural differences. Therefore, validating a French-Canadian version was warranted in light of previous French-translated tools from French-Canadian to French-France (Lonjon et al., 2014).

PURPOSE AND OBJECTIVES
This study aimed to translate the FCV-19S into French-Canadian (French-CA) and English-Canadian (English-CA), and to validate both versions in the nursing staff population from the province of Quebec, Canada. Specific objectives were to evaluate the:

DESIGN AND SAMPLE
This methodological study was part of a larger cross-sectional survey. An anonymous online survey was sent to nursing staff in the province of Quebec. Members of the Quebec Order of Nurses (Ordre des infirmières et infirmiers du Québec -OIIQ) and the Quebec Order of Licensed Practical Nurses (Ordre des infirmières et infirmiers auxiliaires du Québec -OIIAQ) were eligible if they had consented to be contacted for research projects. As established practices in these professional orders, an approximate proportion of 25-30% of these consenting nursing members were randomly selected through their specific order to not overburden members with multiple project invitations within the same period. Out of 28 500 nurses and 24 000 licensed practical nurses (LPN), the invitation to complete the survey was sent to an estimated total of 15 000 members across the province, and we expected a minimum participation rate of 10%.

PROCEDURES
Ethics approval was obtained from the Medical/Biomedical Research Ethics Committee of the Centre intégré universitaire de santé et services sociaux (CIUSSS) du Centre-Ouest-Montréal to conduct this study . OIIQ and OIIAQ members received an email inviting them to complete an anonymous survey online via Qualtrics (2020), a secure web-based electronic data capture system. A link was provided and led them to an introduction letter and consent. If they agreed to participate, they were granted access to the survey. The survey was available from July 22 to November 16, 2020, and could only be completed once per participant. Two reminders at a minimum of two-week interval were sent.
Considering that nurses and LPNs are French or English-Canadian speakers, participants could choose their preferred language (French-or English-Canadian) to complete the survey.

INSTRUMENTS
For discriminative and convergent validation of the French-CA version of the FCV-19S, selected variables were used including the psychological distress K6 scale, the Patient Health Questionnaire on depression symptoms, work satisfaction and turnover intention as well as socio-demographic and work-related characteristics during the pandemic.
The Fear of COVID-19 Scale. The FCV-19S was developed using several steps, including an extensive literature review on general fear scales and item pooling, content evaluation with experts from various disciplines (i.e., health education, psychology, medicine, nursing, and sociology), cognitive interviews and pilot testing with 46 individuals from the general population (Ahorsu et al., 2020). The FCV-19S includes 7 items, each rated on a Likert scale from 1 (strongly disagree) to 5 (strongly agree), for a total score from 7 to 35. Higher scores represent greater fear of COVID-19 (Ahorsu et al.). The original Iranian version of FCV-19S was found to include one main factor with a Cronbach alpha of 0.82. After obtaining permission from the primary author of the scale (Ahorsu), we translated the English version of the FCV-19S into French-CA and English-CA using a double forwardbackward method inspired by Sousa and Rojjanasrirat (2011). Two bilingual French-Canadian speakers (one with experience in health care terminology and one with colloquial phrases) independently translated the English version into French-CA. Any discrepancies were resolved via discussion as part of a committee including the bilingual French-Canadian speakers and two research team members with expertise in tool translation and development to produce a preliminary French-CA version of the FCV-19S. For most items, the exact same wording in French-Canadian was used by both individuals who did the translation. Minor edits were made to some words (e.g., "réseaux sociaux" to "médias sociaux", "mal à l'aise" to "inconfortable"). The translation of "losing my life" (item 4) led to further discussion as it was translated as "perdre ma vie" and "mourir" by each individual. We made the decision to keep "perdre ma vie" as it was felt that the meaning was related to broader aspects of life. This preliminary French-CA version then underwent blind backward translation into English-Canadian by two other bilingual speakers (one familiar with health care terminology, and one with the use of colloquial phrases) blind to the scale. The preliminary French-CA and the two back-translated English-CA versions were compared to the English version of the scale (Ahorsu et al.) and discussed within a larger committee involving all four individuals who translated the scale and three research team members with expertise in tool translation and development. Most items were back-translated with the exact same wording as the English version except for "scared" (item 1 -afraid) and "moist" and "sweaty" (item 3 -my hands become clammy). The committee agreed that the meaning was preserved as terms were synonyms. Again, "perdre ma vie" was back-translated into "losing my life" and "dying." In order to be consistent with our decision and the English version (Ahorsu et al.), it was decided to keep "losing my life." To better fit with the English-Canadian language, minor edits were made to the English version of the scale (item 1: "most afraid" was changed for "very afraid", item 5: "I become nervous or anxious" was moved at the beginning of the sentence). Finally, "coronavirus" was changed for "COVID-19" in both the French-CA and English-CA versions. Both versions of the FCV-19S can be found in Table 4.

Stress-Related
Measures. The K6 psychological distress questionnaire includes 6 items rated on a 5-level descriptive scale (0= none of the time to 4= all of the time) (Kessler et al., 2002). Areas Under the Curve (AUC) >0.80 for the K6 were obtained with DSM-IV, or World Health Organization structured interviews as reference criteria in the general population (Kessler et al., 2002(Kessler et al., , 2003(Kessler et al., , 2010. A cut-off score of 13 or higher has been established to identify individuals with nonspecific serious psychological distress (Kessler et al., 2003). In this study sample, internal consistency was supported with Cronbach alphas of 0.87 and 0.88 for the French and English versions of K6.

5
The 9-item Patient Health Questionnaire (PHQ-9: Kroenke & Spitzer, 2002) is rated on a 4level descriptive scale (0=Not at all to 3=Nearly everyday) with a total score ranging from 0 to 27. The PHQ-9 is a depression screening tool with scores from 0 to 4 indicating no to minimal symptoms, 5 to 9 mild symptoms, 10 to 14 moderate depression symptoms, and scores ³15 severe symptoms. In this study sample, internal consistency was supported with Cronbach alphas of 0.89 and 0.88 for the French and English versions of PHQ-9. Both the K6 and the PHQ-9 are recommended measures by the Institut national de santé publique du Québec (Canuel et al., 2019).
Work Satisfaction and Turnover Intention.
Work satisfaction was measured using a selfreported single-item (I am satisfied with my work) on a 7-level descriptive scale (do not agree at all to very strongly agree). This single-item measure of work satisfaction has shown to have high correlations with other work satisfaction scales (Wanous et al., 1997). Using the same response scale, turnover intention was assessed with 2 items: professional turnover intention and organisational turnover intention (O'Driscoll & Beehr, 1994). In this study sample, acceptable Cronbach alphas of 0.79 and 0.74 for turnover intention were obtained for the French and English versions.

Socio-Demographic and
Work-Related Characteristics. Socio-demographic included selfidentified gender, age and generation. Workrelated characteristics included professional role (nurses or LPN), years of experience (current setting and profession), healthcare setting (acute care, long-term care, and other), perception of preparedness to offer safe care during the COVID-19 pandemic (very well prepared to very poorly prepared), provided care to COVID-19 patients or not, and had provided care to COVID-19 patients who died.

DATA ANALYSIS
Descriptive statistics were obtained for all variables. According to the level of measurement of each variable, frequencies (%), means and standard deviations (SD) or medians and interquartile ranges (IQR) were reported. The FCV-19S items of the French-CA and English-CA versions appeared normally distributed with SDs < means, and kurtosis and skewness indices < ±2 (Kim, 2013). Cronbach alphas were calculated for internal consistency of the French-CA and English-CA versions of the FCV-19S. Cronbach alpha values >0.70 are considered acceptable (Streiner et al., 2015). Exploratory Factor Analysis (EFA) was performed for both versions of the scale to examine its internal structure as a strategy of construct validity. We expected to find one dominant factor with a high eigenvalue (>1) (Tabachnick & Fidell, 2019) and explaining more than 50% of the COVID-19 fear variance. Confirmatory Factor Analysis (CFA) was also performed using IBM SPSS Amos, Version 26.0 (Arbuckle, 2014) to confirm each version's unidimensional structure, and model fit indices were reported. We expected a ratio of c 2 to df <3, comparative fit index (CFI) >0.95, and root mean square error of approximation (RMSEA) <0.06 for good model fit (Schreiber et al., 2006). As a common rule, 10 to 20 participants per item with a minimum sample size of 100 to 200 is required to run factor analysis for scales with well-determined factors (Tabachnick & Fidell). Therefore, our samples of 1517 and 191 nursing staff participants for each version were appropriate to run our factor analyses. We had no missing data for the FCV-19S.
Further analyses included t-tests and one-way analyses of variance (ANOVA) using relevant sociodemographic and work characteristics. Power analysis was done using G*Power 3.1 (Faul et al., 2007). Considering ratios of 0.2 and 0.3 as compared groups were not of equal size, a sample size > 1000 participants was required to perform ttests (a=.01; 80% power; effect size=0.25). A minimum of 275 participants was also required to perform one-way ANOVAs (a=.01; 80% power; effect size=0.25) with up to 5 compared groups. Therefore, other validation objectives were achieved using the French-CA version of the scale. More specifically, t-tests or ANOVAs were performed to examine discriminative validation between groups according to socio-demographic (gender, generation) and work (professional role, healthcare setting, job status, preparedness and exposure to COVID-19) characteristics. Post-hoc tests (Fisher LSD or Tukey HSD) were done to allow for subgroup comparisons. A Pearson or Spearman correlation was calculated between the FCV-19S, psychological distress, depression symptoms, work satisfaction and turnover intention. Subgroup analyses (t-tests, one-way ANOVA) were performed with cut-off scores of psychological distress and depression symptoms. The alpha was set at .01 with Bonferroni correction for multiple tests. We used IBM SPSS Statistics for Windows, Version 24.0. (IBM Corp., 2016). We had less than 3% missing data in independent variables used for discriminative and convergent validation objectives which were not replaced in data analysis.

SAMPLE DESCRIPTION
A total of 1708 nursing staff respondents, for an estimated participation rate of 11%, completed the FCV-19S i.e., 1517 and 191 completed the French-CA and the English-CA versions, respectively. As expected for nursing staff, the majority were women. Participants were in the middle ages, and all administrative areas of the province of Quebec were represented with a proportion of 60% from Montréal, the south shore (Montérégie and Estrie) and the North Shore (Laval and Laurentides). Participants were experienced nursing staff working in acute care, long-term care and other settings (e.g., community care, mental health, public health). Almost half of the participants reported that they provided care to COVID-19 patients, and among them 60% had provided care to COVID-19 patients who died. All sample characteristics are described in Table 3. The level of fear of COVID-19 for the total sample (n=1708) was moderate (mean=17.03, SD=6.18).

OF COVID-19 SCALE: FRENCH-CA AND ENGLISH-CA VERSIONS
Internal consistency was supported with Cronbach alphas of 0.90 and 0.88 for the French-CA and English-CA versions of the scale (Table 4). In the French-CA version, moderate inter-item correlations ranged from 0.47 to 0.74 (p<.001). In the English-CA version, moderate inter-item correlations were also found and ranged from 0.39 to 0.64 (p<.001). The Kaiser measure of sampling adequacy was > .60 and the Bartlett's test of sphericity was significant supporting EFA assumptions. Multicollinearity was absent as all inter-item correlations were lower than .85. EFA revealed one large factor with a high eigenvalue (>4), explaining 63.5% and 59.4% of the fear of COVID-19 variance for the French-CA and English-CA versions. Factor loadings were all above 0.65 in both versions of the scale ( Table 4). CFA of both versions supported the scale's unidimensionality when taking into account covariances between items' error terms. The French-CA one-factor scale showed good model fit with a ratio of c 2 to df=2.67 with CFI=1.00 and RMSEA=0.03. Also, the English-CA one-factor scale showed a good model fit with a ratio c 2 to df=1.38, CFI=0.99, and RMSEA=0.05.

DISCRIMINATIVE VALIDATION OF THE FRENCH-CANADIAN VERSION OF THE FEAR OF COVID-19 SCALE
Fear of COVID-19 mean scores differed according to various socio-demographic and work characteristics (Table 5). Fear levels tended to be higher in women than men. No correlation was found with age. However, we found differences between generations. Levels of fear were higher in generation X (40-56 years old) compared to generation Y (20-39 years old) (Fisher LSD, p=.002) and Baby Boomers (57-74 years old) (Fisher LSD, p=.004).
Many differences were associated with work characteristics. Fear levels were higher in LPNs compared to nurses, and in long-term care facilities compared to acute care and other settings (Fisher LSD, p<.001). A tendency towards significance was identified for job status (F test, p>.01) with higher fear in part-time nursing staff. Higher fear levels were found in nursing staff who reported being poorly prepared to offer safe care during the COVID-19 pandemic. Most group pairs showed different fear levels (Tuckey HSD, p<.01) except between groups who reported somewhat poorly and very poorly prepared (p=.105). No statistically significant differences in fear levels were found between nursing staff who provided care to COVID-19 patients and those who did not. However, higher fear levels were found in those caring for COVID-19 patients who died.

CONVERGENT VALIDATION OF THE FRENCH-CANADIAN VERSION OF THE FEAR OF COVID-19 SCALE
A moderate level of psychological distress was found in this sample with a mean of 8.17 (SD=5.05), and it was positively correlated with fear of p<.001). Nursing staff who reached the cut-off score ³13 had a greater level of fear compared to those who did not (Table 5). Mild depression symptoms were found in this sample with a PHQ-9 mean of 6.44 (SD=5.61) which were positively correlated with fear of COVID-19 (r= 0.31, p<.001). Significant differences in fear scores were found between the four categories of depression symptoms (Table 5). Post-hoc tests demonstrated significantly higher fear scores in most group pairs (Tukey HSD, p<.001) except between nurses with mild and moderate symptoms (p=.116) and between nurses with moderate and severe symptoms (p=.090). Nursing staff with no or minimal depression symptoms had lower fear scores compared to all other groups (p p<.001).

DISCUSSION
We translated the Fear of COVID-19 scale into French-Canadian and English-Canadian using a rigorous translation approach. At the time we did the translation, only the English version provided by Ahorsu and colleagues (2020) was available. Soon after, on June 25 th , the English-US version (Perz et al., 2020) was published, and on July 22 nd , the French-France version was made available (Mailliez et al., 2020). Some differences were noted between our French-CA version and the French version from France. More specifically, the terminology from France included the use of "mourir" in item 4, "réseaux sociaux" in item 5, "mon rythme cardiaque accélère" in item 7 as well as the use of "le" instead of "la" COVID-19 in all items. Our English-CA version was very close to the English-US version. The only difference was related to item 4 in which the term "dying" is included in the English-US version.
A unidimensional scale was found in this nursing staff population of Quebec. Indeed, internal consistency and one main factor of fear behaviors related to COVID-19 were supported for both versions. Findings were consistent with those of previous validation studies of the Iranian version (n=717; Ahorsu et al. 2020), the French-France version (n=316; Mailliez et al., 2020), and the English-US version (n=237; Perz et al., 2020) with Cronbach alpha coefficients >0.80, and unidimensional scale (Lin et al., 2021).
In terms of discriminative validation of the French-CA version of the scale using sociodemographic characteristics, we found a tendency towards higher levels of COVID-19 fear by an average of one point on the FCV-19S in women compared to men. Previous studies from other countries among healthcare workers (n=2930, Mexico;Garcia-Reyna et al., 2020) and students in health disciplines (n=5423, Vietnam;Nguyen et al., 2020;n=370, Israel;Zolotov et al., 2020) such as medicine and nursing reported higher fear levels in women than men. Although age did not influence fear levels in our nursing sample, levels of COVID-19 fear were higher in generation X compared to generation Y and Baby Boomers. Generation X is born from 1964 to 1980 (40 to 56 years old).
Considering that the average maternity age is 30 years old in Quebec, individuals from this generation may have children who live with them (La Presse Canadienne, 2010). Also, they may fulfill caregiver roles for their parents. Kumar et al. (2020) identified infecting family members as the most important factor associated with fear faced by healthcare workers (n=329) in Pakistan during the COVID-19 pandemic. Transmission of COVID-19 to their family was the top source of distress in healthcare workers (n=657 with almost half were nurses) from a large medical centre in New York (Schechter et al., 2020). Therefore, nurses from generation X may have been concerned about getting COVID-19 and infecting their family members.
Regarding discriminative validation related to work characteristics, fear levels were higher in nursing staff who work in long-term care facilities, who provided care to COVID-19 patients who died, and were LPN. High fear level in long-term care facilities is consistent in our provincial context, considering that more than half of COVID-19 deaths were reported in these healthcare settings during the first wave of the pandemic in Quebec (Institut national de santé publique du Québec, 2021). Testimonies of healthcare workers' fear were also stated in the Quebec Ombudsman's Progress Report on the first wave of COVID-19 in long-term care facilities (Protecteur du citoyen, 2020). In alignment with this, higher fear levels were found in nursing staff who provided care to COVID-19 patients who died than those who survived. Higher fear was also found in nursing staff who reported being poorly prepared to offer safe care including appropriate nosocomial infection prevention and control interventions during the COVID-19 pandemic (Observations sur les centres d'hébergement de soins longue durée de Montréal, 2020). Interestingly, in a study by Labrague et al. (2020) with a sample of 261 nurses working in COVID-19 designated hospitals in the Philippines, a lower fear level related to COVID-19 was found in nurses who received a COVID-19 training compared to those who did not. Such a finding supports the importance of providing COVID-19 training to nursing staff. Other studies with healthcare workers have shown greater odds of COVID-19 fear or higher fear levels in nurses compared to other health professionals and support staff (Garcia-Reyna et al., 2020;Khanal et al., 2020). In this methodological study, only nursing staff was included not allowing us to make comparisons with other health disciplines. Moreover, to our knowledge, we are the first research team to include LPN. It must be noted that a higher proportion of LPN (46.4%) worked in long-term care facilities compared to nurses (17.0%) in this study sample. Therefore, higher fear in LPN is coherent in our provincial context, where the highest rate of mortality related to COVID-19 was found in long-term care facilities during the first wave of the pandemic.
According to convergent validation of the French-CA version, positive mild associations (r=0.30-0.38) were found between fear of COVID-19, psychological distress and depression symptoms. We must mention that we were the only research team to use the K6 scale, not allowing us to compare our findings with previous studies. These previous studies used the Hospital Anxiety Depression Scale (HADS) or the Job Stress Scale (JSS). Correlations in the range of 0.40-0.50 with these stress measures were found in nurses (JSS in Labrague et al., 2020) and healthcare workers (HADS in Khanal et al., 2020). Similar correlations were found in the general population for the original Iranian version and the French-France versions of FCV-19S with HADS anxiety and depression scores (Ahorsu et al., 2020;Mailliez et al., 2020). Anxiety was also found to be associated with higher fear scores (English-US version) in a student sample (Perz et al., 2020). The PHQ-9 was used by Saracoglu and colleagues (2020), but they did not report associations between depression and fear of COVID-19. Although our correlations with stress-related measures were lower in comparison to these previous studies, they were positive as expected and almost reach a moderate magnitude (Agoklu, 2018). According to convergent validation with work characteristics, our mild correlations (<0.30) between COVID-19 fear, work satisfaction, and turnover intention are similar to those found by Labrague et al. (2020) who used the same measure for turnover intention and a similar one for work satisfaction. A mild negative correlation was found between COVID-19 fear and work satisfaction, and mild positive correlations were obtained with turnover intention. Turnover intention is an important predictor of nurses' actual turnover behaviour (Hayes et al., 2006).

LIMITATIONS
The participation rate was low, but expected. A potential explanation for this is the timing of the invitation during the summer vacation period. Also, several surveys were launched during the pandemic, and nursing staff was a targeted group in these studies. Nursing staff may also have been solicited to participate in other studies through their healthcare institution. A smaller sample of nursing staff completed the English-CA and due to low power, we could not check for discriminative and convergent validation of the English-CA version of the scale. Our primary goal was to check if the scale structure was consistent and similar for both versions. Our selected stress-related measures differed from those used in other studies, not allowing us to compare our findings with previous studies. As this methodological study was part of a larger cross-sectional survey with broader objectives, we had to select measures with fewer items to minimize the participant's burden.

IMPLICATIONS FOR NURSING
Fear of COVID-19 was high in Quebec nursing staff and was associated with stress-related symptoms. Training and appropriate resources are essential to provide nursing staff and healthcare workers with adequate preparation and protection from COVID-19. Screening for fear of COVID-19 could help identify nursing staff at higher risk of distress. It could guide the development of further interventions to modify or overcome factors that trigger the fear of COVID-19, such as stressors, lack of information, or low perceived control. In addition, the development of support interventions and safety measures at the workplace and at home are important to reassure nursing staff and healthcare workers.

CONCLUSION
The French-CA and English-CA versions of the FCV-19S resulted in a unidimensional scale. Discriminative and convergent analyses of the French-CA version of the scale support its validity in Quebec nursing staff population. As fear of COVID-19 was associated with stress-related symptoms, it should be addressed in training and support programs. The FCV-19S was validated in a large sample of nursing staff, but further validation testing with other multidisciplinary team members is required to support its validity in the healthcare workforce of Quebec.
Authors' contribution: CG, MLT and JC designed the study. CG, CM, MH, and MRL participated in the translation process. CG and MRL collected and analyzed the data. CG organized the article. MRL drafted the introduction and CG drafted the objectives, methods, results, and discussion. All authors (CG, CM, MLT, MRL, MCG, EG, MH, ETN and JC) participated in the interpretation of findings, revised critically and approved the final version of the manuscript. Faul, F., Erdfelder, E., Lang, A.-G., & Buchner, A. (2007