Outliers Case Study
After deleting all univariate outliers and multivariate outliers, reliability of counter-productivity and personality were α=.59, α=.76 respectively. According to Spiliotopoulou, 2009, a larger sample size might increase alpha value, this could offer a possible explanation for why reliability of counter-productivity decreased when outliers were deleted. Therefore, in order to avoid the scale being underestimated, we decided to keep the outliers for any further statistical …show more content…
This positive correlation suggested that the higher the score in customer service orientation questionnaire (i.e. which means you are less agreeable, less conscientious and less emotion stable), the more likely that you are a counter-productive employee. Result suggested that there is indeed a relation between psychometric test and job performance.
Results from Tett, Jackson and Rothstein’s (1991) meta-analysis supported the relation between job performance and psychometric tests (r= .24), and confirmed the validity of using personality scale for employee selection. Moreover, Ashton (1998) found that overall workplace misbehavior scale was strongly related to agreeableness (r= -.25) and conscientiousness (r= -.30) scales in the big five personality scales, however this correlation was not obtained in emotion stability scale.
In the current research, the correlation is stronger than both abovementioned studies, therefore, this psychometric test would be useful in selecting for sales …show more content…
(our data was combined into one index (i.e.customer service orientation))
Combining Likert scales into indexes adds values and variability to the data. If the assumptions of normality are met, analysis with parametric procedure can be followed. (Christopher and Elaine, 2007)
Creswell(2008) further proposed that in order for Likert data to be regarded as interval data there is need to develop multiple categories within a scale (in our questionnaire, it was measuring counter-productivity, aggreeableness, conscientiousness, and emotion stability), establish equality of variance between each value on the scale and normality of the data.
In Murray’s (2013) experiement, she used scale data from a study with 111 participants which measured the 5 variables (mathematics self-efficacy, academic self-regulation, availability of academic resources and learning styles). The study asked participants to choose the best fit answer from 3 different Likert 5 point scales. After completing data analyses, she concluded that both parametric and non-parametric analyses (e.g. Pearson and Spearman rho) did not change the conclusions drawn from the results of Likert scale data. Carifio and Perla (2008) and Pell (2005) both stated that it is no problem/suitable to use combined likert scales to perform parametric