Good measures mean reliable and valid measures One thus needs to assess

Good measures mean reliable and valid measures one

This preview shows page 90 - 92 out of 194 pages.

Good measures mean reliable and valid measures. One thus needs to assess carefully the measures used in terms of their reliability and validity. In this final section of the chapter, issues related to the reliability and validity of this study are brought up and efforts made to avoid possible bias and increase the credibility of the findings are discussed. 4.7.1 Reliability The reliability of a measure mainly concerns two aspects, the repeatability (how consistent are the results when data is collected in the same way at another point in time) and the internal consistency (how stable is the measurement across its items) (Zikmund 1994). In order to test for the repeatability of a measurement, additional data would have to be collected; however, doing so is not feasible in the scope of this study due to constrains in time and money. Reliability will thus be assessed in terms of the measurements internal consistency. One of the most well established measures for internal consistency is the Cronbach alpha ( ơ ), which is also used in this study to assess reliability. Cronbach’s alpha is calculated based on the correlations of the individual scale items with each other and is the most commonly used indicator of scale reliability. This value should ideally be above .7 (Hair et al. , 2007). Furthermore, the average variance extracted (AVE) is calculated, which represents the average amount of variance explained among the indicators of the latent construct (Hair et al ., 1998). A common threshold for the AVE is .5. The results are reported in the following analysis chapter. 4.7.2 Validity Particularly in social science research where we need to measure rather abstract concepts with relatively concrete measures, it can be difficult to establish whether the instruments measure what they are supposed to measure. This is referred to as the issue of validity. There are several ways to test for the validity of an instrument, but there is no real way to prove validity - it can only be argued (de Vaus, 2002).
Image of page 90
_______________________________________Methodology __ 73 There are three primary types of validity: (1) content validity, (2) construct validity, and (3) criterion validity (Hair et al. , 2007). Content (or face) validity refers to the extent to which a measure taps the different aspects/dimensions of a construct (de Vaus, 2002). Content validity is ensured by having experts go through and assess the items to ensure that each construct is well reflected. In terms of construct validity, it can be differentiated between convergent and discriminant validity. Convergent validity exists when the items of a measure are highly correlated. Discriminant validity addresses the notion whether two different constructs in the model really are distinct from one another (de Vaus, 2002). Finally, criterion validity can be assessed by comparing the outcome of a measure with the outcome of a well established other measure of that construct and determining whether they are correlated (de Vaus, 2002). A problem with this, however, is that it is difficult to know whether the established measure can be considered valid (de Vaus, 2002). Yet, if the construct in combination with other
Image of page 91
Image of page 92

  • Left Quote Icon

    Student Picture

  • Left Quote Icon

    Student Picture

  • Left Quote Icon

    Student Picture