I would like to ask a question on CrossValidated about statistics reporting, rather than statistical analysis itself. The question is - Is it necessary to report on the step-by-step results from a model-building process, or is it ok to just describe the methods behind the best fit model selection process, and then report the actual results solely on the best fit models themselves? In this case, I am building mixed effects growth curve models. I am fairly confident that there may be some element of opinion for an answer to this sort of question, and there is certainly likely to be variation across field. I am within the field of health / psychology / medicine / social science, but I would welcome insight from any field because I have been researching this topic and have found it very challenging to hone in on an answer. Is this a sort of question that is appropriate for me to ask here? Do you have any feedback about what would make it most appropriate?
↧