According to the classification outlined above, 123 students in the sample were categorized as
chemical engineering students (29.3% of which were female) and 250 students were categorized
as chemistry students (49.8% of which were female). The chemical engineering students were
comprised of 72% freshman, 21% sophomores, and 7% upperclassman. Similarly, the chemistry
students were comprised of 73% freshman, 16% sophomores, and 11% upperclassman. This
distribution is not surprising given the fact that they survey was administered in a required
introductory English course.
For the questions with linear responses (SAT scores), a Welch’s t-test was used to compare the
mean responses of chemical engineering with chemistry students.14 A chi-square test was
utilized for dichotomous variables to assess whether there is a statistically significant difference
in the responses of the two groups.15 A Wilcoxon rank-sum test was used for all Likert-type
questions.16 For all tests performed in this analysis, the maximum probability of Type-I error (e.g.
a false positive result) that was permitted was 5%. Note that only survey items pertaining to
student preparation, background, and attitudes was analyzed in this paper. All analyses were
conducted using the statistical software system R.17
According to the classification outlined above, 123 students in the sample were categorized aschemical engineering students (29.3% of which were female) and 250 students were categorizedas chemistry students (49.8% of which were female). The chemical engineering students werecomprised of 72% freshman, 21% sophomores, and 7% upperclassman. Similarly, the chemistrystudents were comprised of 73% freshman, 16% sophomores, and 11% upperclassman. Thisdistribution is not surprising given the fact that they survey was administered in a requiredintroductory English course.For the questions with linear responses (SAT scores), a Welch’s t-test was used to compare themean responses of chemical engineering with chemistry students.14 A chi-square test wasutilized for dichotomous variables to assess whether there is a statistically significant differencein the responses of the two groups.15 A Wilcoxon rank-sum test was used for all Likert-typequestions.16 For all tests performed in this analysis, the maximum probability of Type-I error (e.g.a false positive result) that was permitted was 5%. Note that only survey items pertaining tostudent preparation, background, and attitudes was analyzed in this paper. All analyses wereconducted using the statistical software system R.17
การแปล กรุณารอสักครู่..