Download presentation
Presentation is loading. Please wait.
1
Effects of Targeted Troubleshooting Activities on
Student Confidence In a Statistics Computer Lab Meredith A. Henry THE CHALLENGE CHANGES IN AVERAGE STUDENT CONFIDENCE ASSESSING CONFIDENCE *** Successful psychology students are competent conducting and reporting statistical analyses. But, students often struggle with/hold negative attitudes towards statistics and research design. (Mills, 2004) One reason may be lack of confidence/intimidation when confronted with analytical packages such as SAS (Statistical Analysis System). Previous experience teaching a graduate statistics lab also suggests: A) Students struggle with troubleshooting SAS code/analyses B) Students express preference for hands-on class activities The current study tested the effectiveness of a series of exercises, designed to give students experience troubleshooting the SAS program, in increasing student confidence across different domains of statistical skills. *** Pre-course survey: During the first class of the semester, students self-reported Overall confidence using the SAS program Confidence for skills related to conducting analyses (i.e., “Rate your confidence in using SAS to conduct dependent samples t-test.”) Confidence for skills related to troubleshooting analyses (i.e., “Rate your confidence troubleshooting issues related to SAS code for t-tests.”) Post-course survey: During the last course of the semester students Self-reported on all items again Rated 7 components of the course—Including “Make it Work” (MIW) exercises—from “most effective” to “least effective” Provided qualitative feedback on the MIW exercises Students became significantly more confident over the course of the semester. Repeated measures t-tests revealed that students in both years reported more confidence post-course than pre-course in all three domains: Overall confidence: t(12) = 7.90, p < .001; t(11) = 13.00, p <.001 Analysis: t(11) = 9.54, p < .001; t(11) = 11.46, p < .001 Troubleshooting: t(12) = 8.70, p < .001; t(11) = 7.93, p <.001 GROUP A vs. GROUP B COMPARISONS *** Topics covered in the course were divided into 2 groups: Group A topics had a MIW exercise assigned in both 2014 and 2015. Group B topics only had a MIW exercise assigned in 2015. In Fall 2014, students reported significantly more confidence gain for Group A topics than for Group B topics for both analytic [t(11) = 4.12, p < .005] and troubleshooting [t(12) = 2.50, p < .05] confidence. This suggested that the MIW exercises caused greater confidence gains. *** THE STUDENTS WHAT DID THE STUDENTS THINK? PY 716-L is the lab section of the first of a three-course graduate statistics sequence for students enrolled in UAB’s three psychology graduate programs. Student characteristics: Traditional course structure: Lecture gives students appropriate code to run a variety of analyses in SAS 9.3. Assignments ask students to modify code to successfully run, interpret, and report analyses A group project involves conceiving a study design, creating data, choosing analyses, running analyses, interpreting results, and reporting conclusions. Students had a positive opinion of the MIW exercises Across both semesters, 60% of students ranked MIW as one of the top three most useful aspects of the course. Qualitative Feedback: “I found them helpful because they featured common mistakes.” “Helpful—hands on experience with the code is always good.” “They were helpful; a few more would have been great.” [2014, emphasis added] “I think the more practice you get at picking up on those small errors can make a big difference.” “I like them. I thought they were very helpful in understanding the syntax via troubleshooting. I just think doing more of them would be better. Maybe start the class with that?” [2014, emphasis added] I liked the opportunity to work independently; I was able to process and retain the info better this way compared to during lecture. If so, we would expect no significant difference between Group A and Group B in Fall 2015. However, we saw the same pattern of results. Students still gained more confidence in both domains for Group A vs. Group B topics [t(11) = 4.19, p < .005; t(11) = 2.73, p <.05] In an attempt to explain this surprising finding, we looked back at the properties of the MIW exercises themselves. Fall 2014 Fall 2015 Total students 13 14 Male 5 2 Beh’l-Neuroscience 3 4 Developmental Medical-Clinical 6 7 PRACTICED vs. UNPRACTICED SKILLS *** Statistics is a discipline that often “builds” upon itself Some SAS skills (e.g., entering data, running tests of normality, etc.), while their own topic of instruction, are also necessary parts of more advanced analyses. As such, some skills will appear in multiple lectures, assignments, and MIW exercises. We proposed that students would gain more confidence for these “practiced” skills vs. “unpracticed” skills. 2015 students did report significantly more confidence for “practiced “skills, t(11) = 3.68, p < .005. However, 2014 students reported no difference in confidence based on amount of “practice”, t(11) = 2.109, p = n.s. CONCLUSIONS By chance, many Group A topics were also “practiced” skills. Fall 2014 students were exposed to “practiced” skills multiple times in lectures, but didn’t have as many MIW opportunities for hands-on engagement. Fall 2015 students had MIW activities for all topics, but still ended up “practicing” some of them more. It is possible that the consistently higher gains in confidence for Group A skills, and the failure of Fall 2014 students to benefit from “practice” both reflect the same mechanism. We propose that neither the MIW exercises nor repeated exposure to SAS skills alone is sufficient to optimize gains in student confidence. Rather, targeted, hands-on activities like the MIW exercises, administered multiple times will lead to the greatest levels of student confidence in statistical skill. Briefly, 1) The MIW exercises are effective in increasing student confidence, 2) Greater confidence was associated with higher grades (r = 0.82, p = .001), and 3) student enjoy the MIW exercises. (see feedback above right) Thus, there is sufficient evidence supporting continued use & refinement of MIW exercises to improve the statistics lab. MAKE IT WORK EXERCISES LIMITATIONS & FUTURE DIRECTIONS Both sample sizes were small (N=13; N=14). This reduced power and made a control group unrealistic Graduate students may have different attitudes towards statistics, which may make them more receptive to MIW Future studies should attempt to look at larger and more diverse samples There should also be an attempt to more objectively study the possible MIW*practice interaction Selections of “bad code” given to students Designed to include “common” SAS errors Students’ task: 1) Use knowledge from lectures and error messages in the SAS log window to fix coding errors; 2) Interpret results of analyses once run; 3) Report results in appropriate written form Exercises assigned at the beginning of the class. Students have 10 minutes to complete them, then the whole class reviews and discusses. Contact Meredith A. Henry Phone: (205)
Similar presentations
© 2024 SlidePlayer.com. Inc.
All rights reserved.