Validated formative assessment tools provide a reliable way to compare student learning across variables such as pedagogy and curricula, or demographics. Such assessments typically employ a closed-response format developed from student responses to open-response questions and interviews with students and experts. The validity and reliability of these assessments is usually evaluated using statistical tools such as classical test theory or item response theory. The suitability of individual questions on an assessment can be examined using either of these methods, but so far little attention has been given to evaluating the suitability of individual response choices available in each question. Here, we use mutual information, a tool rarely used in PER, to quantitatively evaluate the utility of response choices in an assessment. We use the Physics Lab Inventory of Critical thinking (PLIC) as an example to illustrate how assessment developers can use this method for their own assessments. The PLIC was designed to measure three latent constructs and we confirm this structure through a factor analysis. We calculate factor scores that represent performance on each of the proposed constructs and evaluate the suitability of individual response choices in terms of how much information they provide about these factor scores.