faculty. Follow-up univariate F tests revealed 14 items that differed significantly at the 0.05 level using the Scheffé test. For a number of the items, it appears that there is statistical significance and not necessarily meaningful differences. For ex-ample, the item concerning taking a test for another student was rated 3.97 and 3.84 by students and faculty, respectively, yet was different in statistical signifi-cance. Of the items that are significantly different, 3 stand out as being meaning-ful. Items pertaining to studying from old versions of exams and having the instructors manual that contains test items were rated more severe by students (Ms = 2.75 and 3.53, respectively) than faculty (Ms = 1.64 and 2.91, respectively), whereas faculty rated collaborating on work that was supposed to be done individ-ually as more severe (M = 3.05) than did students (M = 2.70). A broad definition of perceived prevalence of cheating was established by ask-ing faculty and students to rate the percentage of students they believed cheat using a Likert scale with 1 being 0% and 10 being 90% to 100%. Faculty perceived that between 0% to 10% of students cheat (M = 1.89), whereas students perceived be-tween 10% to 20% of students cheat (M = 2.32), a difference that is significant, i(271) = 4.27, p<. 001. Table 1 presents the perceived prevalence results using a narrow definition of the term based on the 40 academically dishonest items. As can be seen from Table 1, both faculty and students perceive that the prevalence of cheating is quite low, with most item ratings having means below 2.00, indicating that 1 % to 10% of stu-dents are perceived as engaging in the behaviors. The behaviors perceived to be most prevalent by faculty were students using old tests without permission, whereas students perceived changing words slightly from an original source as the most prevalent. Although the perception of these behaviors is relatively low, there is greater variance (standard deviations typically above 1.0) in the ratings of per-ceived prevalence when compared to the severity ratings. In general, students have greater variance than faculty in their perceived prevalence ratings. A MANO VA was computed to determine whether significant differences ex-isted between student and faculty ratings of perceived prevalence. Results re-vealed a significant MANO VA, Wilks's A = .70, F(40, 247) = 2.69, p < .001. Follow-up univariate F tests revealed 23 items that differed significantly at the .05 level. Students had higher perceived prevalence ratings on 22 of the 23 items, with faculty rating the use of old tests without permission as more prevalent than stu-dents. Similar to the results regarding severity ratings, many of the significant dif-ferences obtained on perceived prevalence ratings, although statistically significant, do not appear to be meaningful differences. For example, although the item concerning using unauthorized materials such as crib notes is statistically sig-nificant with a student mean of 1.55 and a faculty mean of 1.34, this difference is not a meaningful one.

DOI link for faculty. Follow-up univariate F tests revealed 14 items that differed significantly at the 0.05 level using the Scheffé test. For a number of the items, it appears that there is statistical significance and not necessarily meaningful differences. For ex-ample, the item concerning taking a test for another student was rated 3.97 and 3.84 by students and faculty, respectively, yet was different in statistical signifi-cance. Of the items that are significantly different, 3 stand out as being meaning-ful. Items pertaining to studying from old versions of exams and having the instructors manual that contains test items were rated more severe by students (Ms = 2.75 and 3.53, respectively) than faculty (Ms = 1.64 and 2.91, respectively), whereas faculty rated collaborating on work that was supposed to be done individ-ually as more severe (M = 3.05) than did students (M = 2.70). A broad definition of perceived prevalence of cheating was established by ask-ing faculty and students to rate the percentage of students they believed cheat using a Likert scale with 1 being 0% and 10 being 90% to 100%. Faculty perceived that between 0% to 10% of students cheat (M = 1.89), whereas students perceived be-tween 10% to 20% of students cheat (M = 2.32), a difference that is significant, i(271) = 4.27, p<. 001. Table 1 presents the perceived prevalence results using a narrow definition of the term based on the 40 academically dishonest items. As can be seen from Table 1, both faculty and students perceive that the prevalence of cheating is quite low, with most item ratings having means below 2.00, indicating that 1 % to 10% of stu-dents are perceived as engaging in the behaviors. The behaviors perceived to be most prevalent by faculty were students using old tests without permission, whereas students perceived changing words slightly from an original source as the most prevalent. Although the perception of these behaviors is relatively low, there is greater variance (standard deviations typically above 1.0) in the ratings of per-ceived prevalence when compared to the severity ratings. In general, students have greater variance than faculty in their perceived prevalence ratings. A MANO VA was computed to determine whether significant differences ex-isted between student and faculty ratings of perceived prevalence. Results re-vealed a significant MANO VA, Wilks's A = .70, F(40, 247) = 2.69, p < .001. Follow-up univariate F tests revealed 23 items that differed significantly at the .05 level. Students had higher perceived prevalence ratings on 22 of the 23 items, with faculty rating the use of old tests without permission as more prevalent than stu-dents. Similar to the results regarding severity ratings, many of the significant dif-ferences obtained on perceived prevalence ratings, although statistically significant, do not appear to be meaningful differences. For example, although the item concerning using unauthorized materials such as crib notes is statistically sig-nificant with a student mean of 1.55 and a faculty mean of 1.34, this difference is not a meaningful one.

ABSTRACT

Students and faculty were asked to rate the strength of 21 statements as justifications for cheating, with the response set ranging from 1 {not at all a justification) to 4 {a strong justification). Both faculty and students rated all but two statements as not at all or weak justifications, with most item ratings having means below 2.00, and standard deviations typically below 1.0. To determine whether significant differences existed between student and faculty ratings of the statements as justifications, a MANOVA was computed. Results revealed a significant MANOVA, Wilks's A = .83, F(21,286) = 2.70,/? < .001. Follow-up univariate F tests revealed that each of the 21 items differed significantly at the .05 level. Students had higher justification ratings on all of the 21 items. Similar to the results regarding severity ratings, many of the significant differences obtained on the justification ratings, although statistically significant, do not appear to be meaningful differences. For example, although the item concerning having many opportunities to cheat is statistically significant with a student mean of 1.59 and a faculty mean of 1.15, this difference is not a meaningful one. Of the items that are significantly different, two stand out as being meaningful. Items pertaining to needing high grades to get a good job or into further postgraduate programs and cheating due to panic, pressure, or stress were rated as weak to moderately strong justifications (Ms = 2.03 and 2.39, respectively) by the students as opposed to not at all a justification to a weak justification by the faculty (Ms = 1.24 and 1.55, respectively).