I’m doing my master degree on role and impact of use of instant messengers on project team performance. My questionnaire consists of several questions for each construct, both independable and dependable. A 7-point rating scale is used to measure responses.
Everyone says it is very simple to analyze data via SPSS. The process is straightforward indeed when you have, for example, one dependable variable measured by one single question. I looked through all methods, and I believe that statistical techniques to explore relationships among variables, such as correlation and regression, are more suitable to me. However, I’m confused about how I am supposed to proceed with my case.
As far as I know, I have two options: create a composite construct, like use of IM and team performance out of my questions using Factor Analysis or go for SPSS AMOS, to which I have access too, but with which I am not familiar.
I have around 170 responses by now, so my case should be suitable for Factor Analysis. For example, I wanna create composite variable for use of Instant Messengers (IM). I have 10 variables (IM1-IM10) that measure that construct. When I run Dimension Reduction processes, I got two factors with eigenvalue over 1 that explain most of the variance. By the way, I assume that questions fall into two categories, where one explains use of IM within a project team, and the second group explains use of IM outside the team. Then I have those tables, Component Matrix and Patter Matrix. The pattern matrix shows the factor loadings of each of the variables. Component Matrix sometimes has values in both columns of factors. So which values I need to use in order to calculate composite score?
I'm quite new in statistics field. My assumption for composite score is something like that IM= (IM1*x+IM2*x+…+IM10*x) / (IM1+IM2+..+IM10) . I don't think that I can simply summarize all the values from IM1 to IM10, or can I?
i think this link may help you https://www.researchgate.net/post/How_do_I_combine_8_different_items_into_one_variable_so_that_we_will_have_6_variables_using_SPSS
if you want to test the dependence relationship structural equation modelling(SEM) is most appropriate one.
if you want to test the correlationtional relationship, confirmatory factor analysis (CFA) is most appropriate one.
you can perform these two tests in AMOS.
if you are not familiar with SEM, the alternative approach is to create the factor score or summated scale from the results of the factor analysis and use them as the variables in regression analysis.
further discussion in this regard is most welcome
Dear Ruslan,
You can either carry out a scale reliability analysis on each of your groups of questions or an exploratory factor analysis on the whole questionnaire. I would not recommend using SEM at this stage. Both techniques allow you to reduce the dimensions in your data in a systematic way, turning your variables into scale data.
- Badges
- Science topic
- Similar topics
- Mathematics
- Statistics
- Statistical Techniques