How to Interpret R² in Plain English

R² is one of the most frequently reported — and most frequently misunderstood — statistics in student research. Here's what it actually tells you, without the formula derivation.

What R² Actually Measures

R² (R-squared) tells you what proportion of the variation in your dependent variable is explained by your independent variable(s) in a regression model. It's expressed as a value between 0 and 1 (or 0% to 100%).

Example: If a regression of exam scores on study hours produces R² = 0.42, this means 42% of the variation in exam scores across your sample is explained by differences in study hours. The remaining 58% is explained by other factors not included in your model — motivation, prior knowledge, teaching quality, and so on.

Is There a "Good" R² Value?

There's no universal cutoff, and this is where a lot of confusion happens. What counts as a strong R² depends heavily on your field:

An R² of 0.25 in an education study examining one factor's effect on student performance is not a weak result — it simply reflects that student performance has many contributing causes, and no single study can reasonably explain all of them.

What R² Does NOT Tell You

A high R² does not mean your model is correct, that your variables are causally related, or that your predictions will be accurate for new data outside your original sample. R² only describes how well your specific model fits the data you already collected. It's also possible to inflate R² artificially by adding more predictor variables to a model, even ones with no real relationship to the outcome — which is why adjusted R² is often reported alongside it in models with multiple predictors, since it penalizes unnecessary complexity.

A Practical Way to Report It

Rather than just stating "R² = 0.42," it's more useful to state what that means in context: "study hours explained 42% of the variation in exam scores, suggesting other factors — likely including prior academic performance and access to study resources — account for the majority of the remaining variation." This framing is both more accurate and more useful to a reader than the bare statistic alone.

Frequently Asked Questions

Field/context	Typical R² expectations
Physical sciences, controlled experiments	Often 0.90+ expected, since noise is minimal
Social sciences, education research	0.20–0.40 is often considered meaningful, since human behavior has many unmeasured influences
Single-predictor models with complex outcomes	Even 0.10–0.20 can be a genuinely useful finding

What is a good R² value?

There's no universal answer — it depends on your field. In education and social science research, an R² of 0.20-0.40 is often considered meaningful, since human outcomes typically have many unmeasured contributing factors.

Does a high R² mean my independent variable causes the outcome?

No. R² only describes how well your model fits the data you collected; it does not establish causation. A high R² can occur even when the underlying relationship isn't truly causal.

What's the difference between R² and adjusted R²?

R² always increases (or stays the same) as you add more predictors to a model, even irrelevant ones. Adjusted R² corrects for this by penalizing unnecessary model complexity, making it more reliable when comparing models with different numbers of predictors.