Intermediatemethods
Data Interpretation in Biology: Reading Results Honestly
Overview
Collecting data is only half the work; interpreting it honestly is the other half. This lesson covers biological variability, what statistical significance actually means, effect size versus p-values, correlation versus causation, and the habits of skeptical reading — for your own results and others'.
Learning objectives
- Explain sources of biological variability.
- Interpret p-values correctly and state their limits.
- Distinguish statistical significance from biological importance.
- Explain why correlation does not imply causation.
- Evaluate a claim using effect size, replication, and plausibility.
Biology is noisy
Living systems vary — genetically, environmentally, stochastically. Two genetically identical plants grow differently; enzyme assays wobble between replicates. Variability is not error to hide but information to quantify: means summarize, standard deviations and standard errors describe spread, and every conclusion must survive the noise. Honest interpretation starts by respecting variation.
Biological variability reflects real differences, not just mistakes. Quantify it, report it, and never let a bar chart hide it — show the spread.
What p-values actually say
A p-value is the probability of seeing results this extreme if the null hypothesis (no effect) were true — not the probability the hypothesis is true, and not the probability the result is due to chance. p < 0.05 means 'surprising under the null,' nothing more. It doesn't measure effect size, importance, or truth — and p-hacking (trying analyses until one dips below 0.05) manufactures false 'discoveries.'
It quantifies surprise under the null — not the probability your hypothesis is correct, not the size of the effect, and not its biological importance.
Significance versus importance
With enough samples, trivial differences become 'significant' — a 0.1% growth increase at p < 0.001 is still biologically meaningless. Effect size (how big is the difference?) and confidence intervals (how precisely do we know it?) matter more than the p-value's asterisk. Ask always: is the effect large enough to care about, and is it plausible mechanistically?
Correlation, causation, and controls
Correlations suggest; experiments decide. A gene's expression correlating with disease might mean the gene causes it, the disease alters expression, or a third factor drives both. Causal claims need manipulation — knockouts, interventions, randomized trials — plus dose-response, temporal order, and mechanistic plausibility. The Bradford Hill viewpoints (strength, consistency, specificity, temporality, plausibility) remain the classic checklist for weighing causal evidence.
Reading like a skeptic
Interrogate any result: are the controls valid? Is the n biological or technical? Were analyses pre-planned or data-dredged? Do the error bars support the claim? Does the effect replicate? Extraordinary claims need extraordinary evidence — and ordinary claims need ordinary rigor. The goal isn't cynicism but calibrated belief: strong methods earn strong conclusions.
Controls valid → n biological → analysis pre-planned → effect size meaningful → mechanism plausible → replicated. Six checks before belief.
Key points
- Biological variability must be quantified, not hidden.
- p-values measure surprise under the null — not truth or importance.
- Effect size and confidence intervals matter more than asterisks.
- Causation needs manipulation plus plausibility, not just correlation.
- Evaluate claims with a skeptic's checklist before believing them.
Common misconceptions
✓ Truth: It only says the data would be surprising if there were no effect — a statement about the data under the null, not about the hypothesis's truth.
✓ Truth: Large samples make tiny, meaningless differences 'significant.' Importance depends on effect size and biological context.
Practice questions
Q1. A study with n=10,000 finds a 0.05% difference, p < 0.001. The best interpretation is:
- a major discovery
- statistically significant but possibly biologically trivial
- proof of causation
- a flawed study
Huge samples detect tiny differences; significance without meaningful effect size is not importance.
Q2. Which statement about p-values is correct?
- p is the probability the hypothesis is true
- p is the probability of the data (or more extreme) under the null
- p measures effect size
- p < 0.05 guarantees replication
The p-value quantifies how surprising the data would be if nothing were happening — nothing about truth probability or magnitude.
Q3. The strongest evidence that gene X causes a disease phenotype comes from:
- correlation of X expression with disease
- knocking out X and observing the phenotype change, with rescue
- a single patient case
- X's chromosomal location
Manipulation — removing the gene changes the outcome, restoring it reverses the change — is the causal test correlation can't provide.
Click a question to reveal the answer and explanation.
Frequently asked questions
What does a p-value actually tell you?
The probability of observing data this extreme assuming the null hypothesis is true. It does not give the probability that the hypothesis is true, the size of the effect, or its importance.
What is the difference between statistical significance and biological importance?
Statistical significance means the result is unlikely under the null; biological importance means the effect is large enough to matter. Big samples can make trivial effects 'significant.'
Why doesn't correlation imply causation?
A third variable may drive both, or the causation may run backward. Establishing causation needs experimental manipulation, temporal order, dose-response, and mechanistic plausibility.
What is p-hacking?
Trying multiple analyses, subgroups, or endpoints until one yields p < 0.05, then reporting only that — which inflates false positives. Pre-registration of analyses prevents it.