Reliability and Validity of Biostatical Analysis used for Pharmacy Research Questionnaire in Saudi Arabia

Phone no: +966504417712 E-mail: yalomi@gmail.com ABSTRACT Objectives: To declare reliability and validity of Biostatical Analysis used for Pharmacy Research Survey in Saudi Arabia. Methods: It is a cross-section survey developed by the authors and the research team. It’s based on the updated literature, national and international accreditation standards organizations. The Internal consistency, reliability through inter-rater reliability, item-item coloration, item-total coloration, Split half reliability (Gutmann’s λ6) and McDonald’s ω, Cronbach alpha. The validity contained of face content validity, construct validity through exploratory factorial analysis and confirmatory factor analysis. All analysis had been done through Statistical Package of Social Sciences (SPSS), Statistical Package of Social Sciences-Analysis of Moment Structures (SPSS-Amos) and Jeffrey’s Amazing Statistics Program (JASP). Results: A total of 209 pharmacists responded. The majority of responders were Saudi 185 (88.52%). The among responders were males [108 (61.77%)] and females [101 (48.33%)]. The three tests had been done with reliability of 31 questions. The completed responders (185) (mean ± SD) was 3.236 ± 0.326 and McDonald’s ω, Cronbach alpha and Gutmann’s λ6 were 0.980, 0.980 and 0.990, respectively with CI 95% (0.975-0.983) and inter-item coloration was 0.607, while the item-total coloration >0.53 and McDonald’s ω, Cronbach alpha and Gutmann’s λ6 value if deleted was >0.97. By using Exploratory Factor Analysis (EFA), the Kaiser-Meyer-Olkin Measure of sampling adequacy was 0.966 and Bartlett’s test of sphericity with approximate chi-square was <0.001. The commonalities extraction for all questions was >0.57, the related components were four of all 31 questions in four components as suggested. They were not confirmed by confirmatory with statistically significant (p<0.001) of the factor model, by factor analysis, by scree plot and pathway analysis and fit not with the original survey changed to 3 factor loading. The confirmatory factor index was (0.761), Tucker-Lewis Index (TLI) was (0.737), Goodness of fit index (GFI) was (0.844) and expected cross validation index (ECVI) (9.029). The collinearity of 23 questions was autocorrelation (2.609e -5) with not statistically significant (p=0.951). The majority of 23 question had Enjuone value close to number 1, while 11 questions had condition index more than 30. All of the 23 questions had The Variance inflation factor (VIF) less than 10 except four questions and had tolerance more than 0.1 except four questions. Conclusion: The reliability and validity related to the corrected survey of biostatistics analysts used in pharmaceutical research in the Kingdom of Saudi Arabia were high. The pharmacy practice was properly used in further future research in Saudi Arabia.


INTRODUCTION
Reliability and validity are a common concept used in the pharmacy research field. The reliability was defined as scale or test is reliable to the extent that repeat measurements made by it under constant conditions will give the same result" 1 that's you need tools to assure to give the same results once repeated the test. The reliability consisted of various types that are including test-retest reliability or, parallel from reliability and inter-rater reliability, or internal consistency reliability. The internal reliability needs various biostatistical analysis to implement that including Cronbach alpha and Machdolan W. The validity was defined as "the degree to which the researcher has measured what he has set out to measure". 1 The validation had several types with common which were content and construct validity. The content validity easy to use, while contract validity needs biostatistical analysis with principal component and factorial analysis. The validity had several advantages that's including unified the goal of the project; it is part of quality management processes, to keep the question of the survey within one field and target. The construct validity needs special education and training to implement through software biostatistical programs. In the past years, the pharmacy strategic plan involved the pharmacy research and validated the role of the pharmacists with a clinical and economic impact in the pharmacy practice. [1][2][3][4][5] Besides, pharmacy indicators of follow-up and a successful pharmacy strategic plan were implemented, 1 including patients and pharmacist satisfaction. As a result, various publications related to the pharmacist with outcome in practice revealed. Also, multiple cross-sections study through a self-survey about patient satisfaction with pharmacy services and pharmacist job satisfaction also revealed. 6 Both surveys extracted from previous literature with a new design appropriate for pharmacy practice in the Kingdom of Saudi Arabia. Even though both studies were very edifying, but reliability and validity are highly suggested for both surveys. Some studies in the biostatistical knowledge of the healthcare professional including pharmacist; they had been done with a very brief discussion about reliability and validity or eve not existed. [7][8][9][10][11] Another study done in pharmacy research knowledge of biostatistics with reliability checked and without detail information about the reliability or validity of the study. 12 The authors were not familiar with an investigation about the reliability or validity of the biostatistics analysis knowledge in Saudi Arabia. The aim of the current study is to declare the reliability and validity of the biostatistics analysis of a pharmacy research survey in the Kingdom of Saudi Arabia.

Survey Development
It is a cross-section survey developed by the authors and hit the research team. It's based on the updated literature and national and international accreditation standards organizations. [13][14][15][16] The survey contained two parts: the first section about demographic data for responders including genders, nationality with dichotomous data and age with ordinal data. The rest of the data uses as ordinal information, including the responder's qualifications, responders background, education, the board of pharmaceutical certificate and the current job and experience. The other section of demographic information was hospital data with ordinal data and included hospital bed capacity based on the ministry of health classification; the university updated hospital accreditation status from national and international accreditation institutions. The second part of the survey about patient satisfaction of pharmacy services. The section divided into several domains and each domain had several questions related to the domain. The answers of the domains were likely with 1 (I do not need this knowledge), 2 (I do not have knowledge), 3 (Weak knowledge), 4 (Incomplete knowledge) and 5 (Complete knowledge). A pilot study was done through the authors and the team distributes electronically or manually to target responders 20-30 as a pilot. Sometimes they interview patients to assure all the questions clear and understood by the responders. All comments brought for discussion. The correction of the survey had been done based on the agreement of most research members. 17 The research team tested the McDonald's ω and Cronbach alpha for internal reliability in the pilot responders by using Statistical Package of Social Sciences (SPSS), Statistical Package of Social Sciences-Analysis of Moment Structures (SPSS-Amos) and Jeffrey's Amazing Statistics Program (JASP). 18

Internal Consistency Reliability Item-Item Correlation
The method used to measure each question to another one, with high coloration results in more than 0.7 that is high internal consistency reliability survey. 18,19

Item-Total Correlation
The method to measure the total questions correlates with each question alone. The high results more than 0.7, the high correlate internal consistency reliability of the survey. 18,19 Split Half Reliability (Gutmann's λ6) The method used through the SPSS and JASP program with the scale option and reliability section. All questions to scale or ordinal data need to be measured. The test used a split-half option. The SPSS or JASP will split the question into two half and measures the coloration of the two groups. The high results more 0.9 about coloration means higher reliability with internal consistency. 17,20,21 McDonald's ω, Cronbach Alpha The research team applied McDonald's ω and Cronbach alpha for internal reliability by using SPSS and JASP. All questions with a scale or ordinal data included in the analysis. The scale more than 0.9; it will be excellent internal consistency, 0.7-0.9 means good reliability, 0.3-0.6 means not acceptable reliability and the score less than 0.3 means weak reliability. 17,22,23

Face Content Validity
The principle authors planned the survey and team research revised independently. Each member revised all survey content questions based on the updated literature and experience. Any violations had been sent to all research team for further discussion and agreements. The survey had been corrected and agreement from the research team. One of the team members transferred all surveys to the Arabic language and double-checked by all team members again for content and accurate translation. 17,19 Construct Validity Exploratory Factorial Analysis The method was used for the construct validity of the survey. The factor used univariate description and Kaiser-Myer-Olin measure of sampling adequacy and Bartlett's test sphericity. The extraction was used principal components analysis, the Eigen values greater than 1 with the maximum iteration of convergence 25 and display through un-rotated faction solution and scree plot. The rotation used Varimax. 15,21

Confirmatory Factor Analysis
The test was done through SPSS-AMOS and JASP software programs with factor variances, R-Sequated, fit measurements, factor loading, without emulation, error calculated with CI 95% and robust method, it was with the auto-estimator and without standardization, it was with pathway analysis. 15,21 Collinearity The test was done through JASP with linear regression for collinearity diagnostics including Eigen value and condition index, the coefficient used with CI 95% tolerance and variance inflation factor, the model fit through ANOVA and auto-correlation with Durbin-Watson. 24

Statically Analysis
Various biostatical analysis was done in the current study like the Mc-Donald's ω, Cronbach alpha and Gutmann's λ6 for calculation reliability. The Kaiser-Meyer-Olkin measure of sampling adequacy and Bartlett's test of Sphericity with approximate chi-square for Exploratory Factor Analysis (EFA) was used during the study. The factor variances, R-Sequated, fit measurements, factor loading, without emulation, error calculated with CI 95% and robust method, it was with the auto-estimator and without standardization, it was with scree plot and pathway analysis. Collinearity had been diagnostician through linear regression, the variance inflation factor was calculated, the model fit through ANOVA and auto-correlation with Durbin-Watson. All biostatistical analysis was done by the Statistical Package of Social Sciences (SPSS), SPSS-AMOS and Jeffrey's Amazing Statistics Program (JASP).

Validity
By using Exploratory Factor Analysis (EFA), the Kaiser-Meyer-Olkin measure of sampling adequacy was (0.966) and Bartlett's test of Sphericity with approximate chi-square was <0.001 (Table 3). The commonalities extraction for all questions was >0.57, the related components were four of all 31 questions in components 4 as suggested as declared with scree plot (Figure 1). They were not established by confirmatory with statistically significant (p<0.001) of the factor model, by factor and pathway analysis and fit not with the original survey and changed to 3 factor loading. (Figure 2

Collinearity
The correlation coefficients of 23 questions was R2 were (0.943) and RMSE was (0.376) with statistically significant (p<0.001), while the autocorrelation was (2.609e -5) with not statically significant (p=0.951). The majority of 23 question had Enjuone value had close to number 1, while 11 questions had condition index more than 30. All 23 of the questions had the Variance inflation factor (VIF) less than 10 except four questions and had tolerance more than 0.1 except for questions (Table 6).

DISCUSSION
Biostatistical analysis is a crucial part of pharmacy research. The biostatistics sciences were an essential part of the study of Pharm D graduate in the Kingdom of Saudi Arabia. All research pharmacists should have a basic knowledge of biostatistics to prevent any mistakes in the results of the research. As a result, the authors and his colleagues did a study to assess the level of knowledge of basic biostatistics among the pharmacists.

Figure1: Exploratory Factor Analysis Scree Plot
Figure2: confirmatory analysis pathway diagram  The study was cross-sectional with a survey that had been distributed to the pharmacist in the Kingdom of Saudi Arabia. The authors did a brief validation of the survey. In the current study, the authors try to evaluate the reliability and validity of the survey. The findings showed a very high reliability of internal consisted of three biostatistical tests either in the pilot responders or completed responders pharmacist and higher than previous medicine residents study. 8

CONCLUSION
The knowledge of biostatistical analysis was used in pharmacy research with the corrected survey had high reliability and validation scale level. Further study with a corrected survey in the future with an adequate sample size is suggested in the Kingdom of Saudi Arabia.

CONFLICT OF INTEREST
None.