I don't know how to adjust bonferroni correction to my data.
Firstly, I tested the levels of several types of biomarker candidates in normal and cancer serum using ELISA. I found three types of potential biomarkers (A, B, C) for discriminating cervical cancer from normal cytology.
Secondly, I combined three types of biomarkers using logistic regression model.
Then, I wanna using bonferroni correction to reduce type I error but I have no idea how to adjust the bonferroni correction to my data.
Is the bonferroni corrected p value < 0.016? (
Yes. Each of your tests is significant if and only if p < 0,016. That is, instead of reaching p < 0,05 each test has to reach 0,016 to be called significant.
I didn't fully understand what exactly you were testing. You have to adjust for every statistical test you perform on the same data. If you performed 3 statistical tests you have to adjust like you did: 0,05/3
It's not clear, but from the description, it sounds like you don't need a p-value adjustment for the logistic regression model.
I'll also note that bonferroni is a very conservative method. There are other methods that are usually more desirable. These may control the familywise error rate (FWER) or the false discovery rate (FDR). I would look into these. The more conservative the p-value adjustment method, the more likely you are to miss significant effects or hypotheses ("type-ii error").
Yes, for Bonferroni correction you did correct. Basically, here are 2 ways of doing it and both lead to the same result: one way, as you did: when deviding threshold level (0.05) by number of tests. And the second way: to multiply obtained p-values for each test by the number of tests and if the adjusted p-value is still below 0.05, consider as significant. This is about Bonferroni correction.
But now, before doing the Bonferroni correction, clarify for yourself: did you do multiple logistic regression of univariate? If you did multiple logistic regression, you do not need any further correction. As I understood, you combined them, meaning that you did multivariate logistic regressoin analysis. So, just use the p-value you obtained. You could apply the Bonferroni correction to the initial analysis, when all biomarker were analysed between normal and cancer serum.
- Badges
- Science method
- Similar topics
- Biological Science
- General Biochemistry
- Biochemical Techniques
- Cytology