A psychologist is interested in whether changing the temperature in a classroom will increase student grades on a math test (measured on a 1-100 scale). At the school, there are three rooms the students are randomly assigned to take their tests in. Each room is set at one of the following temperatures: 65 degrees, 75 degrees, or 85 degrees. What method of describing results would make the most sense for this study (comparing group percentages, correlating individual scores, or comparing group means) and why? What could be done to the study design to decrease the margin of error?