With the criterion, eigenvalue greater than one I have run factor analysis in SPSS for six items. KMO and Bartlett's tests are significant. First component is 2.988 with 49.795 % variance. Second component is 1.002 with 16.702 % variance only. But in rotated components matrix component 1 factor loadings are -.033, .650, .420, .842, .829, .773 and component 2 factor loadings are .884, .474, .582, .132, .024, .177, respectively. I am doubtful whether eigenvalue of 1.002 is significant. Can we ignore it? Because, in other instances of using same data but with different sample size, I have obtained only one extracted component solution therefore I have claimed single dimension i.e. construct validity. In this context also I want to establish uni-dimensionality of the items. How should I proceed? What influences factor solution? Can any one help me. Thank you.
Dear all
I have found answer. I have identified outliers through box plots in SPSS and I have deleted more biased response which consequently reduced my sample from 931 to 930. Later on I have again conducted tests of reliability and validity in SPSS. Now, my instrument proved reliable and valid for all dimensions. Thank you.
- Badges
- Science method