Undergraduate quantum mechanics (QM) uses a variety of notations, each with their own advantages and constraints, for representing quantum states and carrying out individual calculations. An example of this can be seen when calculating expectation values, which can be solved using several different methods. Analysis of written exam data given at three universities (teaching spins-first QM) showed students were more likely to use matrix or integral calculation in situations where it is much simpler to use a summation method. By adapting the structural features framework by Gire and Price, we are able to answer questions regarding how and why students use different methods. We analyze student responses to expectation value problems administered in interviews conducted at two universities in the middle and end of the semester to highlight specific areas of difficulty and features of the methods which led to students' choices of one method over another.