Hello,
I have little doubts about the statistical model that I am using to analyze my data. I have two groups of residue studies data Group 1 n=7 and the other group 2 n=47. they are independent and the studies are expensive and rare so I couldn't increase the sample size by any means.
I have tested the normality of both groups using the SPSS and found to be not normally distributed then I transformed all the data to fit the normal distribution using the square-root calculation - they fit the normal distribution p= 0.2 (more than 0.05).
which test should I use especially that I am using SPSS package
thanks
It has been a few months without an answer probably because you haven't said what your research question(s) is(are). Further, accepting the null hypothesis on the basis of 7 and 8 cases is of course not warranted. I assume you have gone on with your analyses, but if not you might want to include more information in your question.
Thank you for your reply. Actually I am testing two kinds of pesticides residue (Dietary and non dietary) and would like to test if there is a significant difference between them or not are the difference in the disspaition of residue on leaves (non dietary) is the same or faster compared to those on crops ? This is the research question. and due to the high costs of the studies they are very limited and confidential and used for the pesticide registration purposes. These actually the studies available at the moment to compare and I cant really increase the sample size by anyway.
I used a normal T test with considering difference in the sample variances (Welch’s test) what do you think?
If you are interest in whether a difference in means as large as observed is reasonable if there is no differences whatsoever from their populations, then Welch's test would often be used. Bootstrapping is tricky with small samples, and ranked based tests aren't testing means. But, you said you took the square root of the data, so also this is also not about a test of means. Is the test you want about population means, or something else? If by "if there is a significant difference between them", are you meaning a difference of a specific size? I am concerned here that you are using some phrases that get used in some statistical jargon.
ONe possibility is that if you are just interested in whether the values of one of these populations tend to be higher, there are rank based procedures. The Wilcoxon rank sum test is popular, but many people argue against it (e.g., https://www.researchgate.net/post/How_can_I_compare_the_medians_of_groups_with_different_sample_sizes ).
Hello Mohamed Badawy. Take a look at this BMJ Statistics Note by Bland & Altman (2009):
https://www.bmj.com/content/338/bmj.a3166.long
Notice especially the final paragraph. I suspect you are not the first person to measure pesticide residues. What is known about their (population) distribution from other studies? Are means and SDs sensible and defensible for description? If yes, then a t-test will likely be fine.
HTH.
I think you should use the Mann Whitney U test. Because your groups are not suitable for test normality. So you should not use t-test in that analysis.
Good luck with your study.
Haider Raid Talib thanks
I cant use it as I need to compare the mean of the two groups rather than ranking them.i had to transform the data using log or square root to achieve the normality before comparing them using Welch test
Bruce Weaver Thanks this is really helpful .yes as you said most of studies are not normally distributed And all of them back transforming the data to achieves the normality and then they apply Welch test or T-test. Thanks again
By the way, I believe it is "silly or pointless" to test for normality as a precursor to a t-test (or some other parametric test). I explained why in this conference presentation.
Presentation Silly or Pointless Things People Do When Analyzing Data: 1. ...
Apologies to readers who are not old enough to recognize cartoon character Quick Draw McGraw and Baba Looey (which I misspelled as Baba-Louie in the presentation). :-(
- Badges
- Science topic