Hello, I designed and sent the questionnaire to participants and already have data for a survey about "International Entrepreneurial Culture" with 35 questions in 6 predefined dimensions. I asked opinions of the experts regarding the relationship of IEC and early internationalization in their previous studies. for analyzing the data I have 2 questions: 1- what is the best method according to the nature of my study? I was thinking about CFA. 2- Since I my goal is to finding the most important dimension(s) in early internationalization and I do not have a dependent variable such as (time of internationalization) what is your suggestion in analyzing data?
P.s. I am using STATA as my statistical software
CFA will simply tell you whether your observed data match your initial hypotheses about the specification of measurement model. It sounds like you do, however, have an independent variable, because you predict that there should be differences due to time of internationalization. This would require a comparison between the measurement models for two (or more) subsamples, divided according the time of internationalization.
One issue you do not mention, however, is sample size. A common recommendation is that you have at least 5 observations for every parameter you estimate. This would require a minimum N of 175 for the basic CFA or 250 for a comparison of two separate measurement models.
CFA will simply tell you whether your observed data match your initial hypotheses about the specification of measurement model. It sounds like you do, however, have an independent variable, because you predict that there should be differences due to time of internationalization. This would require a comparison between the measurement models for two (or more) subsamples, divided according the time of internationalization.
One issue you do not mention, however, is sample size. A common recommendation is that you have at least 5 observations for every parameter you estimate. This would require a minimum N of 175 for the basic CFA or 250 for a comparison of two separate measurement models.
Given that you are using Stata, you might find this book helpful:
- https://www.stata.com/bookstore/discovering-structural-equation-modeling-using-stata/
Notice that Chapter 5 is all about group comparisons.
Dear Shayan, in addition to CFA, you could try with clustering analysis (subject to the number of respondents). You could group the observations based on their characteristics and their similarities. There are alternatives, but perhaps the centroid-based method (with the number of clusters predefined by your 6 dimensions) could work. I assume you selected the 6 dimensions based on the literature (theory/practice). Another option is the partitioning-based method, where you could divide data in subsets. Best regards.
Luis Rivera Yes my six dimensions are adapted from literature. Thank you again for your kind answers.
- Badges
- Science topic
- Similar topics
- Statistics
- Survey