Study design/setting
This study was a descriptive cross-sectional community-based study conducted among Ajman University (AU) students and staff to develop and validate a novel self-reporting scale in the English language to measure the identification rate of falsified hand sanitizer among the public in the UAE. Respondents were sent a survey link by email, and data collection occurred between March 3, 2020 and March 25, 2020.
Study participants (inclusion/exclusion criteria)
The target population for this research was any resident of the UAE, national or non-national, aged 18 or over, who was willing to participate. Participants aged less than 18 years and those who did not want to participate were excluded.
Pilot testing
The pilot study began at Ajman University on October 20, 2019. As of November 1, 2019, 350 respondents had undertaken satisfactory completion of the questionnaire with no apparent difficulty. The outcomes of the pilot study were employed to calculate the sample size needed for the main research and to check the reliability of the test.
Sample size/sampling technique
To calculate a sample size for this survey, a pilot study was used. The questionnaire was sent to 400 students and staff at Ajman University, from which 350 respondents were achieved, yielding a response rate of 87.5%. The sample size calculation was based on the question, ‘Do you know how to identify falsified hand sanitizer?’ According to the pilot study, the proportion of people who answered yes to this question was approximately 55%. The alpha level was set at 5%, giving a 95% confidence interval. Precision (D) for the 95% confidence interval was fixed at 5% so that the 95% CI would have a maximum width of 10%. On the basis of these assumptions, a sample size n of 1270 was required, assuming that nonresponse rates would be approximately 70%. Ajman University's Admissions and Registration Department supplied us with an Excel spreadsheet containing the names of students and staff and their colleges, study year, and email addresses. Basic random-sample selection was used to choose the sample, with ID numbers employed for random selection, stratified by department and college. A total of 1280 participants were chosen for the final sample.
Administration of questionnaires
The survey instrument was designed as a self-administered survey to be used by preselected respondents chosen at random from the spreadsheet provided by the Admissions and Registration Department. Participants were sent a web-based electronic link to the survey via email. The first page of the survey comprised an explanation of the type of and reason for the study. When participants moved to the next page, this was registered as their consent to participate. Reminder emails were sent two times in a month from the inception of the survey. On the completion of survey the respondents were acknowledged with a ‘thank you’ message and were not given any incentives.
Research instrument conceptualization/development
Face/content validity
The questionnaire's first draft was subjected to face and content validity testing. An expert panel was convened comprising an industrial pharmacist, an academic, two regulatory pharmacists, and two community pharmacists. The panel assessed the content validity of the scale. Additionally, public consultations formed part of the process. The content validity ratio and content validity index (CVI/CVR) were calculated by requesting that the experts classify every item in the instrument as essential or nonessential. Good content validity is demonstrated when the CVR reaches 0.78 or above. Any individual item that does not meet this level would generally be removed from the final draft. The CVI was then calculated by taking the mean of all CVR values for all items that were kept in the final draft, i.e., those with a CVR above 0.78.16, 17
Construct validity
Exploratory factor analysis (EFA) was used to test construct validity. Factor analysis was undertaken using principal component analysis (PCA); then, varimax rotation with Kaiser-Mayer-Olkin (KMO) and Bartlett's sphericity test were conducted to assess factor numbers. The construct validity criteria were satisfied when there were eigenvalues of 1 and item loadings of 0.40 or above with no cross-loading.18 Confirmation of the model was then performed using partial confirmatory factor analysis (PCFA) employing maximum likelihood analysis using oblimin rotation. Calculations were undertaken of the incremental fit indices of the Tucker-Lewis index (TLI), normed fit index (NFI), and comparative fit index (CFI). In addition, we reported the absolute fit index using root mean square error of approximation (RMSEA).19, 20
Internal consistency/reliability analysis
Intraclass correlation coefficients (ICCs) and Cronbach alpha was computed to ascertain test-retest reliability and internal consistency. A value of 0.7 or above is acceptable for Cronbach alpha.19 For ICCs which is a ratio also, is considered highly reliable when the value is near to 1. In the current research Rosner’s criteria is employed to interpret ICCs, which reflects ICC < 0.40 corresponds to poor agreement, 0.40 ≤ ICC < 0.75 corresponds to fair/good agreement and ICC ≥ 0.75 relates to excellent agreement.21 For the estimation of item internal consistency (IIC) Pearson correlation was employed. IIC involves the relationship every item has with its hypothesized domain or factor. The IIC demands that the adjusted scale score should correlate with the item r ≥0.4.22 Test-retest reliability across two points in time was assessed following a six-week gap using Pearson's correlation coefficient (ρ). A (ρ) value above 0.75 and a p-value <0.05 are regarded as a correlation with strong significance.19, 20
Known group validation
We hypothesized that respondents with higher levels of education would have a better ability to identify falsified hand sanitizer than those with lower levels of education. A one-way ANOVA test was used.
Statistical analysis
SPSS version 24 was used to conduct data analysis. Frequencies and percentages were employed to summarize the demographic/baseline characteristics from the study sample. One-way ANOVA was used to calculate the correlation between demographics and the ability to identify falsified hand sanitizer. A p-value below 0.05 was regarded as statistically significant.