I have the data of maize drought trials. There are some traits for which data were collected based on scores. For example, 1=severer rolling, 5=no rolling. So, do I need to transform these kind of data? If I do need then what kind of transformation? Do I also need to standardize the data? What will be the consequences if I do not do any of these?
More details about your experimental design and your plans for analysis are needed before anyone can help you in your specific case.
Here are some general rules though.
Transforming data is often used when the data does not meet the assumptions of an analysis in order to get it to meet the assumptions. For example, ANOVAs assume normality and homogeneity of variance. If you data is non-normal or heteroscedastic, then you could transform it to try to get it to meet the assumptions. So the answer in your case is that it depends on your data and your analysis.
Standardization is often used when you using multivariate statistics and you have variables with vastly different scales. For example, you might measure leaf thickness and leave lengths. For most plants the measurements of thickness will be tiny and lengths huge (relatively). If you use non-standardized data the results will be driven by the length measurement which would completely overwhelm any effect of thickness even if there is a huge effect size in thickness. Standardizing helps with this problem. So again, in your case it depends on the analysis you intend to run.
- Badges
- Science topic
- Similar topics
- Geoscience
- Asia
- India