Correlation

Correlation analysis quantifies how two variables move together—capturing both the direction (positive or negative) and the strength of a linear relationship. In practical terms, it helps answer business and research questions such as whether social responsibility tracks with university reputation, whether higher prices relate to lower product sales, or whether pay increases correspond to reduced absenteeism. In SPSS, correlation is also used to describe relationships across different measurement levels, with Pearson correlation (R) commonly applied to interval/ratio continuous variables and Spearman correlation used when variables are ordinal.

The correlation coefficient, reported as r (or R), ranges from −1 to +1. A value of +1 indicates a perfect positive relationship: as one variable increases, the other increases exactly. A value of −1 indicates a perfect negative relationship: as one increases, the other decreases exactly. A coefficient of 0 indicates no linear relationship—knowing one variable does not help predict the other. Importantly, the coefficient’s magnitude reflects strength, while the sign reflects direction. However, correlation does not establish cause and effect; it only measures association, so causation requires additional testing beyond correlation.

Interpreting results requires more than computing the coefficient. The transcript emphasizes that statistical significance is determined using the P value. A correlation is treated as significant when the P value falls below common thresholds such as 0.05 (and even more strongly below 0.01). For verbal interpretation, the coefficient is matched to a strength category (e.g., very low for values around 0.1–0.3, very high for values around 0.9–0.99), but the P value is what supports claims that the observed relationship is unlikely to be due to chance.

A worked SPSS example demonstrates the reporting workflow. The dataset includes servant leadership measured through seven items and self-efficacy measured through eight items. Because the analysis needs single variables rather than item sets, the items are combined into latent variable scores by computing the mean for each construct (creating new variables for servant leadership and self-efficacy). Then the analysis proceeds through Analyze → Correlate → Bivariate, selecting Pearson correlation, using a two-tailed test, and flagging significant correlations.

The output yields a Pearson correlation between servant leadership and self-efficacy of r = 0.534, with a P value reported as less than 0.01. The relationship is therefore described as moderate, positive, and statistically significant. For write-up, the transcript recommends reporting the correlation coefficient and P value (rather than adding “insignificant”), and optionally stating that the hypothesis of a significant relationship (H1) is supported—interpreting it as higher servant leadership behavior aligning with higher self-efficacy among followers.

When more than two variables are involved, the approach shifts to a correlation matrix. The example adds a third construct (labeled JS), computes its mean score similarly, and runs another correlation analysis to produce a matrix. Because correlation matrices repeat values symmetrically around the diagonal, formatting guidance focuses on removing redundant rows/columns and presenting a clean table suitable for theses or journal articles. Across both two-variable correlations and multi-variable matrices, the core interpretation logic remains the same: direction and strength come from the coefficient; significance comes from the P value; and causation claims are off-limits.