Step 1: Identify the nature of the data first: proportions are categorical, so the right test must operate on frequency counts rather than means.
Step 2: The chi-square statistic, $\chi^2 = \sum \frac{(O-E)^2}{E}$, contrasts observed counts $O$ with expected counts $E$, making it the standard tool to test whether two or more proportions differ beyond chance.
Step 3: Mean-based tests ('t' test, ANOVA) and relationship tools (correlation, regression) are designed for quantitative data, so they are unsuitable here. The proportion comparison belongs to chi-square. \[\boxed{\text{Chi square test}}\]