We examined the influences of pre-solving pause time, algebraic knowledge, mathematics self-efficacy, and mathematics anxiety on middle-schoolers' strategy efficiency in an algebra learning game. We measured strategy efficiency using (a) the number of steps taken to complete a problem; (b) the proportion of problems completed on the initial attempt; and (c) the number of resets prior to completing the problems. Using the log data from the game, we found that longer pre-solving pause time was associated with more efficient strategies, as indicated by fewer solution steps, higher initial completion rate, and fewer resets. Higher algebraic knowledge was associated with higher initial completion rate and fewer resets. Mathematics self-efficacy and mathematics anxiety was not associated with any measures of strategy efficiency. The results suggest that pause time may be an indicator of student thinking before problem-solving, and provide insights into using data from online learning platforms to examine students' problem-solving processes. [This paper was published in "Learning and Individual Differences."]