Happy New Year everyone!
I am currently using JMP for a Cell Culture experiment and working with variables such as Density, Volume, and Speed. I have entered the values on the JMP software as continuous (all of them), but would like to know which test is most suitable to understand my data. Which statistical test can I use to best understand my data?
Thank you!
What do you want to achieve with the data? What are your objectives? etc. All these will determine the answers to your questions.
Hi, If all your variables (dependent and independent) are continues you should use Regression, but if your response variable is continues and independent variables are discrete you should use ANOVA.
I don't see how this is related to JMP. If you want to know how a particzuar analysis is to be performed in JMP, then please let us know what the concrete analysis is and where your particular problem is.
If you just don't konw how to analyse such data, then the standard answer is likely: a (non-)linear model expressing density as a function of volume and speed. Your data set is very small (sparse), so you will either need a solid and reasonable set of assumptions (mainly about the functional relationships) or you simply assume linear relationships, what may limit the quantitative conclusions and the possibility for extrapolations. The simplest model is a multiple linear regression model includeing the interaction of volume and speed. But this this may not be a very sensible model; as nothing much is known and told about the biological (theoretical) background, there can not be any more helpful advice given.
Additional to the previous answers: you mentioned "cause and effect", this cannot be done with statistics only. You need to have additional assumptions, i.e. a temporal relationship of the variables (strictly speaking, you cannot show causality in a cross sectional study) and the exclusion of other possible causes (which may be handled to some degree by the sampling process or controlling techniques). The statistics may only contribute to find the necessary covariance between the variables, and as Jochen pointed out, this may not be linear in nature. Therefore, a null result may not be interpreted as support for no causal relationship.
I am always very skeptical when I read questions such as Which statistical test can I use to best understand my data, because I do not think that test-driven statistics is useful to most people, because results from tests are generally interpreted in a (Neymanian) accept/reject fashion, i.e. P-values are
- Badges
- Science topic