The Pearson correlation coefficient (r) is one of the most fundamental tools in quantitative research. First derived by Karl Pearson in 1895, it provides a single standardised number — ranging from −1 to +1 — that concisely captures both the strength and direction of a linear relationship between two continuous variables. Its mathematical elegance lies in being scale-invariant and dimensionless, allowing researchers to compare associations across variables measured in entirely different units.

When applying Pearson correlation, four conditions should be confirmed: both variables should be continuous (interval or ratio scale), the relationship should appear linear on a scatter plot, both variables should be approximately normally distributed (especially important for small samples), and no major outliers should be present. Violations of these assumptions — particularly influential outliers and non-linearity — can render the Pearson r seriously misleading. In such cases, Spearman rank correlation or Kendall's tau are more appropriate and robust alternatives that should be computed and reported alongside Pearson r for transparency.

A distinction every researcher must internalise is the difference between statistical significance and practical significance. With large samples (n ≥ 200), even very weak correlations (r = 0.10) may reach p < 0.05 — but such a correlation explains only 1% of variance, which is scientifically trivial in most contexts. Conversely, with very small samples (n = 8), a genuinely moderate correlation (r = 0.55) may not reach significance simply due to insufficient statistical power. This is why researchers must always report the confidence interval for r alongside the p-value: the CI communicates both the precision of the estimate and the plausible range of true population values.

The coefficient of determination (r²) is often the most practically meaningful output. If r = 0.70, then r² = 0.49, telling us that approximately 49% of the variance in Y is explained by its linear relationship with X. This is far more informative than the p-value alone, because it directly answers the question "how much does knowing X help us predict Y?" Always report r² (or the equivalent effect size) in research papers, dissertations, and technical reports — it is required by APA 7th edition guidelines and increasingly expected by peer reviewers.

In ecological and wildlife research, Pearson correlation is routinely applied to examine relationships between environmental gradients and biological responses — for example, how annual rainfall amount predicts plant biomass, how prey density influences predator activity patterns, or how GPS-derived movement parameters correlate with habitat quality metrics. Field biologists working with camera-trap data, GPS telemetry, or morphometric datasets will frequently report correlation coefficients as a foundational step before applying regression modelling or multivariate ordination methods such as PCA or NMDS.

Finally, and most critically: correlation does not imply causation. No matter how large or statistically significant a Pearson r value is, it cannot, by itself, establish a causal mechanism. Unmeasured confounding variables, reverse causation, or coincidental co-variation may entirely explain an observed association. Proving causation requires controlled experimental designs with random assignment, or advanced causal inference frameworks (instrumental variables, regression discontinuity, difference-in-differences). Use Pearson correlation as a powerful tool for describing and quantifying bivariate associations — but always report it completely (r, df, p, CI, r²), interpret it cautiously, and embed it within the broader context of your study design, theoretical framework, and research question.