Grading practices can send a powerful message to students about course expectations. A study by Henderson et al. ("American Journal of Physics" 72:164-169, 2004) in physics education has identified a misalignment between what college instructors say they value and their actual scoring of quantitative student solutions. This work identified three values that guide grading decisions: (1) a desire to see students' reasoning, (2) a readiness to deduct points from solutions with obvious errors and a reluctance to deduct points from solutions that might be correct, and (3) a tendency to assume correct reasoning when solutions are ambiguous. These authors propose that when values are in conflict, the conflict is resolved by placing the burden of proof on either the instructor or the student. Here, we extend the results of the physics study to earth science (n = 7) and chemistry (n = 10) instructors in a think-aloud interview study. Our results suggest that both the previously identified three values and the misalignment between values and grading practices exist among science faculty more generally. Furthermore, we identified a fourth value not previously recognized. Although all of the faculty across both studies stated that they valued seeing student reasoning, the combined effect suggests that only 49% of faculty across the three disciplines graded work in such a way that would actually encourage students to show their reasoning, and 34% of instructors could be viewed as penalizing students for showing their work. This research may contribute toward a better alignment between values and practice in faculty development.