Study design and date
This cross-section validation study was conducted during the period from 10th July to 23rd August 2020.
Subjects
The study was carried out on university students in Mansoura University, Egypt, and King Khalid University, Saudi Arabia.
Sample size calculation
There are no absolute rules for the sample size required to validate constructs or study tools due to the various types of tools and the number of their items [19]. Guidelines for the respondent-to-item ratio ranged from 5:1, 10:1, 15:1, or 30:1 [20]. As larger samples are always better than smaller samples, we will use the respondent-to-item ratio of 30:1 for better confirmatory factor analysis [19]. The scale contains 36 items; thus, the final sample size is 1080 students.
Procedures and data collection
Participants were volunteered to enroll using an anonymous online survey, administered by Google Forms to ensure wide reach and easy access. The electronic link was published on social media and student forums; the participants agreed to participate voluntarily without any compensation or incentives. Data were collected anonymously including sociodemographics (e.g., age, sex, nationality, residence, and college) and the Arabic version of COVID-19 stress scale.
The COVID-19 Stress Scales (CSS) prepared by Taylor et al. [18] consists of 36 items, distributed over five dimensions stable factor, corresponding to scales assessing COVID-related stress and anxiety symptoms: (1) danger and contamination fears (12 items), (2) fears about economic consequences (6 items), (3) xenophobia (6 items), (4) compulsive checking and reassurance seeking (6 items), and (5) traumatic stress symptoms about COVID-19 (6 items). Items were rated on a 5-point scale ranging from 0 (not at all) to 4 (extremely). The scales were intentionally designed so they could be readily adapted for future pandemics. The CSS were developed and initially validated in population-representative samples from Canada and the USA. The scale performed well on various indices of reliability and validity. The scales were inter-correlated, providing evidence of a COVID stress syndrome.
Arabic translation of CSS
The English version was translated into the Arabic language by two independent translators, and the two translations were compared to reach an agreed initial version. This initial version was revised from a linguistic, grammar, and other perspective. The Arabic version was back translated into the English language again by another two translators unaware about the original English version, to ensure the accuracy of the translation and the consistency of the synonyms and the translation did not differ between the English and Arabic versions [21].
Content validity
The content validity indices were estimated by 22 experts (11 Egyptian and 11 Saudi) jurors specialized in psychology, psychiatry, and public health. They included 7 professors, 13 associate professors, and 2 lecturers. The Arabic version of the scale was evaluated for clarity, relevance, and translation of the contents. The experts were asked independently to review each item (clarity and relevance) using three-point ordinal scale and translation using correct or wrong. If wrong, they add their suggestions. The content validity index (CVI) was calculated at the item level (I-CVI), expert level (E-CVI), and scale level (S-CVI).
To obtain the CVI at the item level (I-CVI), the number of experts judging the item as relevant or clear (rating 3) were divided by the total number of experts. The CVI per expert (E-CVI) is calculated by dividing number of items scored 3/total number of items. If the CVI is higher than 0.79, the item was appropriate. If it is between 0.70 and 0.79, it needs revision. If it is less than 0.70, it is eliminated. The CVI for the entire scale (S-CVI) was assessed using the S-CVI with the average approach, by summing all I-CVI for relevancy divided by the number of items. The scale as a tool was considered to be valid if S-CVI greater than or equal to 0.90 [22].
I-CVI and S-CVI were calculated using the following formulae [23]:
$$ \mathrm{I}-\mathrm{CVI}=\mathrm{Number}\ \mathrm{of}\ \mathrm{experts}\ \mathrm{rating}\ \mathrm{the}\ \mathrm{item}\ 3/\mathrm{total}\ \mathrm{number}\ \mathrm{of}\ \mathrm{experts} $$
$$ \mathrm{S}-\mathrm{CVI}=\mathrm{Sum}\ \mathrm{of}\ \mathrm{the}\ \mathrm{I}-\mathrm{CVI}\mathrm{s}/\mathrm{total}\ \mathrm{number}\ \mathrm{of}\ \mathrm{items} $$
The CVI for each expert (E-CVI) is number of items scored 3 (relevant)/total number of items.
Reliability was assessed in the forms of internal consistency. To test the reliability of the CSS-Arabic, the final version was completed by1080 students. Internal consistency was examined by Cronbach’s α reliability coefficients. Cronbach’s α value of 0.50–0.70 was acceptable, whereas 0.70 or higher shows good homogeneity among the items [24].
Exploratory factor analysis
Exploratory factor analysis of the 36 items was done using RML rotation using MPlus. Parallel analysis was used to determine the number of factors to retain. The selection of goodness-of-fit indices was based on conventional guidelines [25].
Statistical analysis
Data were analyzed with SPSS version 24 (IBM Corporation, Chicago, IL, USA). Qualitative variables were presented as number and percent, whereas quantitative variables were presented as mean (SD). CVIs were calculated for each item and each expert. Cronbach’s α were calculated to measure the internal consistency between items. P ≤ 0.05 was considered statistically significant. Validity and factor analysis were determined using robust maximum likelihood (RML) rotation using MPlus examining the full scale [26]; the researchers used RML because it is robust to departures from a normality distribution of the data sample.
Internal consistency was assessed by Cronbach alpha coefficients (α), inter-item correlations and corrected item-total correlations. A Cronbach’s α of .70 or higher indicates acceptable reliability [27]. Goodness of fit was assessed according to the following criteria: root mean square error of approximation (RMSEA ≤ .10), comparative fit index (CFI > .90 or more desirably ≥ .95), and standardized root mean square residual (SRMR ≤ .08) [28, 29].