COMPARISON OF FCQ RESULTS FOR SUMMER 2006 (ONLINE) VS. SUMMER 2005 (PAPER) Perry Sailor University of Colorado at Boulder Office of Planning, Budget, and Analysis September 2007 Summary Background and procedure. Although online administration of FCQs has been an available option for Boulder instructors since Spring 2003, the vast majority have always chosen to administer FCQs in class on paper forms (with the exception of online courses). However, because of time demands on the FCQ staff caused by the planned introduction of a new FCQ form in Fall 2006, the Office of Planning, Budget, and Analysis (PBA) made online administration mandatory for all FCQs in Summer 2006 for that term only. This provided the opportunity for a natural experiment on the effect of mode of administration (paper vs. online) on ratings and response rates. Instructor ratings, course ratings, and response rates from summer 2005 (paper administration) and summer 2006 (online administration) were compared. We did two sets of comparisons between terms: (1) all sections, and (2) restricted to sections of the same course taught by the same instructor in each term. Key results Response rate: The average response rate was considerably lower when the FCQ was administered online, compared to paper administration. Ratings of course and instructor: The effect of online administration on instructor and course ratings differed by course level. For undergraduate sections, average ratings were slightly lower with online administration, “slightly” meaning sufficient to move a section 2 to 6 percentile points lower in the distribution of section ratings. For graduate sections, the opposite was true, with course and instructor ratings slightly higher when FCQs were administered online than on paper. Variation in ratings: For undergraduate sections, variation within sections in course and instructor ratings were larger when the FCQ was administered online. For graduate courses, variation was higher with online than with paper administration for course ratings, but lower for instructor ratings. As with average ratings, all differences between paper and online variation were small. Section size: The negative effect of online administration ratings was larger in bigger sections; there was little or no effect on instructor or course ratings in sections of less than 20 students, but a bigger one in sections of 40 or more. Instructor type: We compared results for tenured/tenure track instructors (TTT), teaching assistants (TA), and others. Differences among the three groups were small and inconsistent for course and instructor ratings. For response rates, the negative effect of online administration was much less pronounced for TTT instructors than the others, but was still sizable. Discipline College: Unlike other colleges, average instructor ratings and course ratings in Engineering were actually higher with online administration than with paper, and online response rates were almost the same as paper (although it was because paper Paper vs. Online FCQ Administration Page 2 of 10 response rates were low, not because online response rates were high). However, these results were based on only 9 sections. Details There were 677 sections in summer 2006 (online), and 644 sections in summer 2005 (paper)1. Section means, standard deviations, and return rates from the two terms were compared. The unit of analysis was the section – that is, the means and SDs were calculated over 677 (2006) and 641 (2005) section means, not over the 7,404 (2006) and 11,282 (2005) individual forms in the sections. Table 1 shows results for all sections. “Effect size” is the difference between the mean from the online administration and the mean from the paper administration, divided by the standard deviation of the paper administration. Thus, it is the difference between the two modes of administration, expressed in units of standard deviation. Table 1. Summer 2005 (Paper) and 2006 (Online) results for all sections. 2005 2006 (Paper) (Online) Measure Mean Mean Difference Effect Size Instructor Rating Section Mean 3.45 3.36 -.09 -.19 Course Rating Section Mean 3.27 3.24 -.03 -.06 Response Rate 73% 48% - 25% pts. -1.00 Instructor Rating Section SD 0.66 0.73 .06 .18 Course Rating Section SD 0.76 0.81 .05 .17 N sections 641 677 N forms 11,282 7,404 All of the online vs. paper differences were statistically significant, except course rating. However, with the exception of response rate, where there was a very large difference, the effect sizes are small, all less than 0.2 standard deviation. Because statistical significance is largely a function of sample size – with large samples, even tiny differences of no practical significance can be statistically significant -- effect size is a more meaningful measure. (See http://www.colorado.edu/pba/ia/statrules/eff_size.html for more information on effect sizes.) There were 250 cases over the two terms in which the same course was taught by the same instructor, once using paper FCQ administration, once using online. All further analyses in this report include only these 250 cases, involving the same instructors teaching the same courses in both terms. Table 2 shows the overall results from these sections: 1 Note: Some 2005 FCQs were administered online; these were eliminated from analyses. D:\98881671.doc 7/12/2016 5:57:52 AM Paper vs. Online FCQ Administration Page 3 of 10 Table 2. Summer 2005 (Paper) and 2006 (Online) results for sections of the same course taught by the same instructor in both terms. Measure Instructor Rating Section Mean Course Rating Section Mean Response Rate Instructor Rating Section SD Course Rating Section SD 2005 2006 (Paper) (Online) Mean Mean Difference Effect Size 3.49 3.40 -.09 -.20 3.32 3.28 -.04 -.09 72% 48% - 24% pts. -.97 0.65 0.74 .09 .29 0.75 0.82 .06 .21 Again, the effect sizes for course and instructor ratings are small, although slightly larger than in the comparison of all sections reported in Table 1. How meaningful is a difference of .09 points (instructor rating) or .04 points (course rating)? Applying the same mean ratings to the same courses using paper-administered FCQs in fall 2005, the difference between a 3.49 mean instructor rating and a 3.40 would be tantamount to a move from the 56th percentile to the 50th. The difference between a 3.32 course rating and a 3.28 would be a move from the 66th percentile to the 64th. Box plots of course and instructor ratings for 2005 (paper) and 2006 (online) are attached in the appendix (Figure 1), as are box plots of response rates (Figure 2). Differences by section size Table 3 shows course ratings, instructor ratings, and response rates for the two administration types broken down by section size, involving only the subset of 250 sections of the same course taught by the same instructor in both terms. There appears to be a bigger effect on instructor ratings in larger sections; that is, online administration vs. paper makes a bigger difference for instructor ratings as section size gets larger. The same is true for course ratings, where virtually the entire difference between paper and online ratings is for sections of 40 or more students. The effect sizes for response rate are much larger for sections of 10-19, 20-39, and 40 or more students than for very small sections of 1-9 students. Unfortunately, the biggest section in the sample was less than 100 students, so we couldn’t check to see if differences between paper and online ratings continued to grow as section sizes reached the 200, 300, or even 400 or more students that occur in fall and spring terms. Table 3. Ratings for summer 2005 (paper) vs. summer 2006 (online), by class size, for sections of the same course taught by the same instructor in both terms. N in Class 1-9 10-19 20-39 40-97 N 12 83 121 34 All 250 Instructor Rating Paper Online ES 3.68 3.63 -.08 3.53 3.46 -.16 3.52 3.42 -.24 3.26 3.10 -.33 3.49 D:\98881671.doc 7/12/2016 5:57:52 AM 3.40 -.20 Course Rating Paper Online 3.55 3.58 3.36 3.33 3.34 3.30 3.09 2.97 3.32 3.28 ES .05 -.07 -.09 -.28 -.09 Response Rate Paper Online ES 57% 41% -.36 75% 50% -.89 74% 48% -1.24 66% 47% -.90 72% 48% -.97 Paper vs. Online FCQ Administration Page 4 of 10 Instructor Ratings by Section Size Average Section Rating Summer 2005 (paper) vs. Summer 2006 (online) 3.8 3.6 3.4 Paper 3.2 Online 3.0 2.8 1-9 10-19 20-39 40-97 Section Size Course Ratings by Section Size Summer 2005 (paper) vs. Summer 2006 (online) Average Section Rating 3.8 3.6 3.4 Paper 3.2 Online 3.0 2.8 1-9 10-19 20-39 40-97 Section Size Response Rate by Section Size Summer 2005 (paper) vs. Summer 2006 (online) Average Response Rate 80% 70% Paper 60% Online 50% 40% 1-9 10-19 20-39 Section Size Differences by class level D:\98881671.doc 7/12/2016 5:57:52 AM 40-97 Paper vs. Online FCQ Administration Page 5 of 10 As table 4 shows, online instructor and course ratings were actually higher than paper in graduate level classes, the opposite of undergraduate classes. Effect sizes were small. Return rate differences between paper and online administration were similar for graduate and undergraduate classes. Table 4. Ratings for summer 2005 (paper) vs. summer 2006 (online), by class level. Level Grad Undergrad N 17 233 Instructor Rating Paper Online ES 3.67 3.71 .12 3.48 3.38 -.22 All 250 3.49 3.40 -.20 Course Rating Paper Online 3.53 3.56 3.30 3.26 3.32 3.28 ES .07 -.09 Response Rate Paper Online ES 76% 45% -.84 72% 48% -1.00 -.09 72% Instructor Ratings by Course Level Summer 2005 (paper) vs. Summer 2006 (online) Average Section Rating 3.8 3.7 3.6 Paper 3.5 3.4 Online 3.3 3.2 Grad Undergrad Course Level Course Ratings by Course Level Summer 2005 (paper) vs. Summer 2006 (online) Average Section Rating 3.6 3.5 3.4 Paper 3.3 Online 3.2 3.1 Grad Undergrad Course Level D:\98881671.doc 7/12/2016 5:57:52 AM 48% -.97 Paper vs. Online FCQ Administration Page 6 of 10 Response Rate by Course Level Summer 2005 (paper) vs. Summer 2006 (online) Average Response Rate 80% 70% Paper 60% Online 50% 40% Grad Undergrad Course Level Differences by instructor type Table 5 shows ratings and response rates for the two administration types broken down by instructor type (tenured/tenure track (TTT), teaching assistant (TA), and other). Differences in effect sizes among the three groups are small and inconsistent for course and instructor ratings. For return rates, mode of administration made much less difference for TTT instructors than for TAs and others. Recall that the effect size for mode of administration was likewise less for very small sections than for larger ones; so we checked to see if perhaps more TTT instructors taught very small sections. However, this was not the case. Table 5. Ratings for summer 2005 (paper) vs. summer 2006 (online), by instructor category. Category TTT TA Other N 68 32 150 Instructor Rating Paper Online ES 3.59 3.53 -.17 3.27 3.15 -.20 3.50 3.39 -.26 All 250 3.49 3.40 -.20 Course Rating Paper Online ES 3.46 3.44 -.05 2.98 3.04 .12 3.33 3.25 -.18 3.32 3.28 -.09 Response Rate Paper Online ES 68% 52% -.52 68% 39% -1.26 75% 48% -1.23 72% 48% Differences by Discipline College Engineering sections did not show the same pattern as other colleges; average instructor ratings and course ratings were actually higher with online administration than with paper administration. Moreover, unlike every other college, online response rates in engineering were almost the same as paper (albeit because paper response rates were low, not because online response rates were high). However, the reader should note that these results are based on only 9 sections in each condition, and so should be regarded cautiously. D:\98881671.doc 7/12/2016 5:57:52 AM -.97 Paper vs. Online FCQ Administration Page 7 of 10 Online administration was associated with a particularly large negative gap in course and instructor ratings in business sections. We checked to see if engineering sections tended to be particularly small or business sections particularly large, since class size was related to mode-of-administration effect (see Table 4); neither was the case. Table 6. Ratings for summer 2005 (paper) vs. summer 2006 (online), by discipline college. Instructor Rating Paper Online ES 3.48 3.40 -.24 3.50 3.40 -.22 3.36 3.14 -.47 3.64 3.60 -.12 3.49 3.52 .07 * * * * * * 3.58 3.58 .00 College N Arch & Plan 6 Arts & Sci 180 Business 27 Education 10 Engineering 9 Journalism 2 Law 4 Music 12 All 250 3.49 3.40 -.20 Course Rating Paper Online 3.34 3.32 3.31 3.28 3.17 3.00 3.57 3.52 3.28 3.51 * * * * 3.56 3.44 3.32 3.28 ES -.10 -.07 -.33 -.13 .70 * * -.26 -.09 *Too little data to for meaningful comparison; however “All” includes these colleges. Instructor Ratings by College 3.7 3.6 3.5 Paper 3.4 Online 3.3 3.2 Discipline College D:\98881671.doc 7/12/2016 5:57:52 AM us ic M n En gi ne er in g Ed uc at io Bu sin Sc i & Ar ts Pl a & Ar ch es s 3.1 n Average Section Rating Summer 2005 (paper) vs. Summer 2006 (online) Response Rate Paper Online ES 71% 60% -.92 73% 47% -1.04 76% 52% -1.71 92% 45% -5.87 53% 51% -.06 * * * * * * 65% 48% -.65 72% 48% -.97 Paper vs. Online FCQ Administration Page 8 of 10 Course Ratings by College Summer 2005 (paper) vs. Summer 2006 (online) 3.6 3.5 3.4 Paper 3.3 Online 3.2 3.1 3.0 us ic M n En gi ne er in g Ar ts Ed uc at io Bu sin Sc i & Pl a & Ar ch es s 2.9 n Average Section Rating 3.7 Discipline College Response Rates by College 100% 90% 80% 70% 60% 50% 40% Paper Discipline College D:\98881671.doc 7/12/2016 5:57:52 AM us ic M n En gi ne er in g Ed uc at io Bu sin Sc i & Ar ts Pl a & Ar ch es s Online n Average Response Rate Summer 2005 (paper) vs. Summer 2006 (online) Paper vs. Online FCQ Administration APPENDIX Figure 1. Course, Instr rating distributions Summer 05 and Summer 06 | 4 + | | | | | | | | | | | | | | | | | +-----+ +-----+ 3.75 + | | | | | | | | +-----+ | | | | | +-----+ | | | | | | | | | | | *-----* | | 3.5 + | | | | | | *-----* | | | | | | + | | | | *-----* *-----* | | | + | | | | | | | | | | 3.25 + | + | | + | +-----+ | | | | | | | | | | | | | | | | | | | | | | | | +-----+ 3 + +-----+ +-----+ | | | | | | | | | | | | | | | | | 2.75 + | | | | | | | | | | | | | | | | | | | 2.5 + | | | | | | | 0 | | | | 0 | | | | 0 | 2.25 + | | 0 | | | | 0 | | | | 0 | | | | 2 + 0 | 0 | | 0 | 0 0 0 0 | 0 0 0 0 1.75 + 0 0 | 0 0 | 0 0 * 0 | 0 1.5 + 0 0 | 0 0 * | 0 0 | 0 0 0 1.25 + | 0 0 * 0 | 0 | * * 0 1 + * | 0 * 0 | 0 | 0.75 + * | * ------------+-----------+-----------+-----------+----------Mode (Yr) Paper(05) Online (06) Paper (05) Online (06) var Course Course Instr Instr Top edge of box = 75th percentile Bottom edge of box = 25th percentile Middle horizontal line = Median (50th percentile) + = Mean Vertical lines = Extent of data (to 1.5 interquartile ranges) 0 = Data values 1.5 to 3 interquartile ranges * = Data values more extreme than 3 interquartile ranges D:\98881671.doc 7/12/2016 5:57:52 AM Page 9 of 10 Paper vs. Online FCQ Administration Figure 2. Return rate distributions Summer 05 and Summer 06 Variable: retrate (Return rate) Schematic Plots | 1 + | 0 | | | | | | 0 0.9 + +-----+ 0 | | | | | | | | | | | | 0.8 + *-----* | | | | | | | | | | | + | | 0.7 + | | | | +-----+ | | | | | | | 0.6 + | | | | +-----+ | | | | | | | | 0.5 + | | | | | *--+--* | | | | | | | | 0.4 + | | | | | +-----+ | 0 | | 0 | 0.3 + 0 | | | | 0 | | 0 | 0.2 + | | 0 | | 0 | | | 0.1 + 0 | | | | 0 | | 0 + * 0 ------------+-----------+----------Mode (Yr) Paper(05) Online (06) Top edge of box = 75th percentile Bottom edge of box = 25th percentile Middle horizontal line = Median (50th percentile) + = Mean Vertical lines = Extent of data (to 1.5 interquartile ranges) 0 = Data values 1.5 to 3 interquartile ranges * = Data values more extreme than 3 interquartile ranges D:\98881671.doc 7/12/2016 5:57:52 AM Page 10 of 10