3.6.2. SampleAccording to Sekaran (2009), a sample is a part of the population. Frompopulation the researcher get 100 respondents to analyze. The sampling frame is alisting of the members of the target population that can be used to create and/ordraw the sample (E.Stevens, 2006). In this research the author will use non-probability sampling. The data collection was done by using the formula because of thepopulation is unknown :n = Za2E2n = 1.960.202= 96.0425
Whereby :N = number of sample Za2= standard value which the level of confidence (a) 95%E = the level of provision that is used to express the maximum error is 20% According to the calculation below, can see the minimum of example in thisresearch were 96 respondents. So the researcher will take 100 respondents to dothe study. The crtitria are :-Female or male.-Muslim.-Age 15 – 50 years old.-Aware about halal issues.-Know about Wardah Cosmetic3.7. Data analysis methodThe data that has been collected in this research will be analyzed inStructural Equation Modeling (SEM) and Path Analysis. According to Ghazali(2008) explained that Structural Equation Modeling (SEM) is a model ofstructural equation which is a simultaneous equation that focuses on thepredictions which is capable of describing latent variables (indirect measurable)and indirectly measured based on indicators (manifest variables). SEM for socialscience researchers provides the ability to provide path analysis with latentvariables. SEM analysis has a higher flexibility for researchers to link existingtheory. Analyzed in SEM is the relationship between the latent variable(unobserved variable) and not between the manifest variable or between theindicator variables (observed variable). Hox and Bechger (2002) stated that SEMis a combination of corfirmatory factor analysis, regression analysis and pathanalysis. SEM is suitable for the present study because SEM can directly measurethe relationships between latent and observed variables (Hair et al., 2010). WhilePartial least squares is a method for construct preductive models when the factors26
are many and highly collinear. Partial least squares (PLS) regression is atechnique that reduces the predictors to a smaller set of uncorrelated componentsand performs least squares regression on these components, instead of on theoriginal data. The rationale of the use of SEMPLS in this research is similar to thedefinition by Ghozali (2006), which is PLS is a soft modeling analysis methodbecause it is not assume data that must be measured on a certain scale, whichmeans the sample size can be small.3.7.1. Evaluation of measurement model (outer model)184.108.40.206 Validity TestThe research will be valid if able to be measured what should be measure asthe researcher want. Several types of validity test are used to test the goodness ofmeasures. (Sekaran, 2009).
You've reached the end of your free preview.
Want to read all 36 pages?
- Fall '19