Which of the following is considered an acceptable statistical practice?
1) Choosing the model to test based on a model selection criterion
2) Choosing a model to test based on the desired controls for the research question
3) Comparing many models and publishing the one with the most significant results
4) Comparing many models to search for future research questions