Hi, I collected my data using questionnaires from 3 different firms in different industries, and when I used one-way ANOVA, the results showed that the differences in 3 groups are big. How could I control it?
SET A NEW HYPHOTHESIS that the effect I want to investigate is different in 3 firms.
OR
EMERGE GROUPS WITH SIMILAR RESULTS?
Thanks a lot if anyone could give me some advice.
TYPE 1 ERROR: It appears you are engaging in Type 1 error by trying to fix the data to fit your needs. in research, we take the data as they are, not trying to fix the data or the result. ANOVA is a calculation with specific formula: F = MSR / MSE. Whatever the result of the calculation, we accept it "as is." To insist otherwise or try to fix it so that it becomes what we want is Type 1 error at the least, at the worst, it is unethical.
FAILED VARIABLES must be removed, as irrelevant or non-significant. No effort should be spent trying to "fix." report you findings, not report the "fix" or "doctored findings."
DIFFERENT FIRMS THAT ARE DIFFERENT.... if that is the findings, report that finding. If you have a hypothesis that does not agree with the data, it just means that you hypothesis is wrong. Don't try to "fix" any thing so that the result will come out as the way you want it.
Paul Louangrath Thank you for your reply. I would consider it seriously. I haven't considered about the differences issue before I collected data. And after collecting data from 3 firms which come from different industries, I just used the whole data to do the data analysis and the result is okay. But I started to notice that I need to control the differences since there are 3 different types of companies.
Therefore, what I mean fix it was is there any possible methods to emerge the 2 groups into one or should I add a new hypothesis as you mentioned to compare the 3 firms. And thank you a lot for your opinion for I was just started to do research and I was lack of many research knowledge and experience but I'm trying to learn. Thanks.
If you design is proper no need to fix it. Analysis also you cannot fix it your design you have to correct it if needed and if every is done without any error then you have to accept the results as it appear you cannot change that since that are the true result of yours research.I hope so
- Badges
- Science topic
- Similar topics
- Econ
- Manufacturing
- Industry