If the reviewer want linear regression and I have one dependent non parametric parameter (10 sample size) and ten parametric parameters (each parameter 5 sample size).
Can I perform Pearson correlation and linear regression?
Yes, You can still perform Pearson correlation and linear regression in your scenario, but there are some important considerations to keep in mind.
Pearson correlation assesses the linear relationship between two continuous variables. It presumes that the variables are normally distributed and have a linear relationship. While Pearson correlation can be calculated between a parametric and a non-parametric variable, it is important to note that the interpretation may not be entirely meaningful, particularly if the non-parametric variable does not meet the normality and linearity assumptions. Furthermore, with a sample size of ten, the power of the correlation test may be low.
Linear regression models the relationship between a dependent variable and one or more independent variables. Assumptions such as linearity, normality, error independence, and homoscedasticity etc are critical linear regression. If the nonparametric dependent variable violates these assumptions, the linear regression results may be unreliable.
Due to small sample sizes and potential assumptions violations, it's advisable to use non-parametric regression techniques or bootstrapping methods to handle smaller sample sizes and relax traditional parametric assumptions. Additionally, explore resampling techniques like permutation tests to assess variable relationships without relying heavily on assumptions.
Thanks a lot for your help. If normalization is done for the non parametric dependent parameter but two parametric parameter (out of 10) are not homogenous; can multiple linear regression, result may be reliable?