Describe before you model
Report the median and interquartile range alongside the mean — real data is skewed and averages hide it. Understand the distribution's shape, the sampling process and how missing values arose before fitting anything.
Inference
A confidence interval expresses the uncertainty around an estimate; a p-value is the probability of data at least this extreme if the null hypothesis were true. It is not the probability the hypothesis is false, and statistical significance is not practical importance.
Traps
Multiple comparisons inflate false positives — correct for them. Simpson's paradox can reverse a trend when groups are pooled. Survivorship bias, selection bias and confounding all produce confident, wrong conclusions; correlation only implies causation under a design that supports it.