📸 Point-biserial correlation in SPSS — calculated as Pearson r
What Is Point-Biserial Correlation?
Point-biserial correlation (rpb) measures the strength of association between one continuous variable and one genuinely dichotomous binary variable (e.g., pass/fail, male/female, treatment/control). It is mathematically identical to Pearson r when the binary variable is coded as 0 and 1, so no special formula or SPSS menu is needed.
Running in SPSS
Step 1: Ensure the binary variable is coded 0/1.
Step 2: Analyze → Correlate → Bivariate. Add both the continuous and binary variables. Method: Pearson → OK. The resulting r IS the rpb.
📸 rpb output — Pearson r with binary variable = point-biserial r
Converting rpb to Cohen's d
To compare with other effect size metrics: d = 2r / √(1-r²). For rpb=.312: d = 2(.312)/√(1-.097) = 0.654 — a medium effect.
APA Reporting
Point-biserial correlation indicated a significant association between gender and exam score, rpb(118)=.312, p=.003, 95% CI [.136, .470], with female students scoring significantly higher (M=74.2) than male students (M=68.4).