Study design and setting
A cross-sectional survey was conducted in Selangor, Malaysia. The state has the surface area of 8,104 km2, an equivalent of 3129 sq. mi. In 2017, Selangor has a population of 6.39 million and Malaysia has an approximate population of 31.19 million which makes Selangor approximately 20.1% of the population of Malaysia. The rationale behind the cross-sectional intercept survey was to achieve a true response and hard-to-reach people living in urban areas. Studies had shown that the intercept survey method is a means of assessing the hard-to-reach province of urban areas and it also has a higher response rate compared to other methods of survey (13). Construct validity and reliability test of the questionnaire was conducted separately through a pilot study among the target population prior to the final data collection. Data were collected from May 2018 to October 2018 at train stations and shopping malls at different periods of time.
Sampling and sampling size
In order to obtain a sample size with a robust statistical power and incorporation of diseases prevalence for sensitivity and specificity, sample size calculation from Buderer, 1996 (14) was adopted. The prevalence of dengue fever in Selangor in the year 2017 was calculated to be 0.007, thus, the required minimum sample size is 550 participants. Incomplete questionnaires were excluded during data entry and cleaning. A total of 560 individuals provided fully completed questionnaires.
The questionnaire consists of three main sections: the first part of the questionnaire which represents the hypothetical construct of knowledge about dengue has 30 items, the second part which represents the hypothetical construct of attitude towards dengue has 13 items and the third part presents hypothetical construct of preventive practices against dengue with 14 items. The Cronbach’s alphas of each construct of the questionnaire were 0.72, 0.70 and 0.61 respectively which was a satisfactory scale on internal consistency (15). The validated pre-tested questionnaire was administered to the respondents who volunteered to participate in the survey.
Exploratory factor analysis (EFA) was used to examine the dimensionality and complex interdependence among each construct of the KAP questionnaire. Convergence validity was further used to affirm the degree of relationship between the items and accordingly, sub-constructs were designed. The construct of knowledge on DF and vectors were factored into six sub-constructs, the construct of attitude towards DF and vectors was factored into four sub-constructs and the construct of preventive practices against DF was also factored into four sub-constructs. Partial least square path analysis (PLS-path analysis) was used to explore the data. The statistical analyses were performed with two software namely IBM SPSS Statistics software Version 23.0 (SPSS; Armonk, NY, IBM Corp., 2015) and R Programming Language (/R/win-library/3.5).
Let X1, X2, …, Xn be the manifest variables or items which are used to indirectly measure the approximate representation of the sub-constructs such that each response to an item xi is 1 or 0 (1 for the correct answer and 0 for the wrong answer) and n = number of questions. Each hypothetical construct (KAP) was considered to be caused by the formation of the sub-constructs (cause-effect relationship). Table 1 illustrates the definition of sub-constructs.
Based on the definition of preventive practices against dengue infection and vectors, which are the ways some target population demonstrates attained knowledge and preconceived ideas through actions, the KAP on dengue model was constructed on two simple theories: 1) Better knowledge influence better attitude and 2) Better knowledge and attitude influence better preventive practices.
= f Knowledge…………………………………….. (2.4.1)
= f Knowledge, Attitude………………………….. (2.4.2)
The ellipses represent the inner models showing the relationship between the main hypothetical constructs. The rectangles are the outer model showing the relationship between each construct and its indicative sub-constructs which form blocks for each of the hypothetical constructs (KAP). The black arrows labeled 1, 2, and 3 represents the relationship between the structural model (inner model) and the unlabeled arrows represent the relationship between the indicative sub-construct and the main construct which is referred to as the measurement model (Figure 1).