I am trying to perform CFA uisng AMOS SPSS for a dataset with 34,000 observations. All the features (dataset columns) in the dataset have non normal distribution. The hypothesis that I am trying prove has 3 factors and 16 variables. Because of the non normal distribution, I intend to use bootstrapping method. I have following questions.
1. How to choose the bootstrapping techniques suitable for the mentioned dataset?
2. How to choose the values for bootstrap sample size, PC confidence level and BC confidence level?
3. Which parameters (e.g. Biased - corrected percentile method -> Regression weights -> P values) should be considered to test the hypothesis using CFA?
Hi! Need to check the t value mainly and also p value. depending on the hypothesis interpretations need to be made considering the 1 tailed or 2 tailed test as well. Thanks.
please read this article
https://support.minitab.com/en-us/minitab-express/1/help-and-how-to/basic-statistics/inference/how-to/resampling/bootstrapping-for-1-sample-mean/interpret-the-results/key-results/
Nirupama Rajapaksha, I could be wrong (and I'm happy for others to indicate I am if so), but I wonder whether you are worrying about bootstrapping when there is really no need for it given that you have 34,000 observations - if, by observations, you mean data records, each of which has 16 variables that you want to use in confirmatory factor analysis (CFA).
If I were you, I'd simply go ahead and conduct a CFA. And, if that doesn't work, I'd then use EFA to try to find out why.
Nirupama Rajapaksha,
See our article. It uses Bootstrapping during CFA.
Article Developing and Validating a New Multi-Dimensional Scale for ...
Best,
Ali
- Badges
- Science topic
- Similar topics
- Philosophy
- Moral Philosophy
- Free Will