Data dredging (also known as p-hacking, fishing expedition, or post-hoc data mining) refers to the misuse of statistical analysis by performing multiple comparisons or subgroup analyses until a statistically significant result is found — often by chance alone — without a pre-specified hypothesis.
🧠 In scientific critique, data dredging implies: Searching for significance rather than testing a meaningful, pre-defined question.
Multiple subgroup analyses without correction for multiple testing.
Highlighting spurious associations as if they were genuine discoveries.
Retrospective storytelling: interpreting random patterns as clinically or biologically relevant.
🧨 Why it’s problematic: Increases the false positive rate.
Produces results that cannot be replicated.
Undermines the credibility of the research.
Often used to inflate the perceived value of weak or null findings.
📉 Example (in a critical review): “The authors engage in data dredging, slicing the dataset until statistical significance emerges — not to test a hypothesis, but to justify one retroactively.”