In the traditional Popperian framework of the scientific method, the order of operations is sacred: first formulate a falsifiable hypothesis, design a controlled experiment, collect empirical data, and finally determine whether the evidence refutes the hypothesis. But in actual academic practice, this sequence is routinely inverted through harking scientific bias (Hypothesizing After the Results are Known): presenting an unexpected post hoc finding as if it had been predicted a priori from the very beginning.
The Epistemic Fraud of Just-So Storytelling
Coined by social psychologist Norbert Kerr in 1998, harking scientific bias distorts the literature by converting exploratory data mining into pseudo-confirmatory science. In an exploratory phase, discovering an unexpected correlation is legitimate and valuable, it generates new hypotheses for future testing. But when an investigator writes up the paper and presents the discovered correlation as the original hypothesis, the nominal significance tests become mathematically fraudulent.
HARKing is the scientific embodiment of the Texas Sharpshooter: an amateur marksman fires 50 bullets at a blank barn wall, walks over with a can of red spray paint, draws a bullseye around the tightest cluster of bullet holes, and claims to be an elite sniper. When you hypothesize after observing the data, you are painting the target around random clusters of observational noise.
The Structural Damage Inflicted by HARKing
The ubiquity of harking scientific bias contaminates entire disciplines through three reinforcing mechanisms:
| Pathological Mechanism | Scientific Distortion | Long-Term Consequence |
|---|---|---|
| Concealing Failed Hypotheses | The original theoretical model that failed is completely omitted from the published manuscript. | Other researchers waste decades attempting to re-test the failed theory because its failure was suppressed. |
| Inflating Theoretical Plausibility | Authors construct elaborate, highly convoluted post hoc theoretical explanations for random statistical flukes. | The discipline accumulates ‘epicycles’ of increasingly bizarre theories that describe zero real phenomena. |
| Unfalsifiable Literature | Because hypotheses are tailored to fit observed noise, every published paper appears to confirm its thesis. | Subsequent independent replication attempts fail, producing the modern replication crisis. |
The Antidote
The only architectural defense against harking scientific bias is cryptographic temporal separation: freezing the study design, sampling plan, and primary hypothesis in an immutable public registry before a single byte of data is gathered. When the protocol is locked in advance, hypothesizing after the results are known becomes structurally impossible.
Continue Your Exploration
This series translates the core principles of the replication crisis, statistical power, causal DAGs, Bayesian false positive risks, and adversarial collaboration into actionable mental models. For formal mathematical proofs, empirical literature reviews, and institutional protocols, consult our foundational treatise: Replication Crisis: 5 Empirical Rules of Epistemic Independence.
Empirical Precedent
In 1998, social psychologist Norbert Kerr formally defined HARKing (Hypothesizing After the Results Are Known) as an epistemic pathology that converts exploratory data dredging into counterfeit confirmatory science. Kerr demonstrated that when unexpected patterns in sample data are retrofitted with post-hoc theoretical justifications, the researcher creates an illusion of predictive power while concealing the true probability of Type I error.
The destructive potential of HARKing was highlighted in 2011 when Daryl Bem published nine experiments in the Journal of Personality and Social Psychology claiming empirical evidence for precognition (extrasensory perception). Bem presented statistically significant p-values for participants allegedly predicting future random computer outputs. Critical post-publication audits revealed that experimental endpoints, outlier rejection criteria, and trial blocks had been adjusted post-hoc to isolate rare statistical fluctuations. When independent researchers attempted preregistered direct replications with fixed protocols, the effect vanished entirely.
The Replication Crisis & Epistemic Independence: 5 Empirical Rules
An empirical exploration through the foundational dynamics, systemic trade-offs, and practical architecture of the replication crisis & epistemic independence: 5 empirical rules.