It is important to be specific when writing a scientific manuscript.Are you presenting a descriptive study, or do you wish to test a hypothesis. If you have a study hypothesis, what is it? How has it been operationalised into statistically testable null hypotheses? Is the study design cross-sectional or longitudinal? In the latter case, is it a cohort study or a case-control study? What associations are you studying? Cause - effect relations are crucial for the analysis. If they are unclear, you have to work with assumptions. Can your estimates be interpreted clinically? This is not always the case with odds ratios or hazard ratios. If the estimates cannot be interpreted, they are not useful. Change the statistical method!