Graphic Organizers Can Facilitate Selection of Statistical Tests Part 1: Analysis of Group Differences

Abstract

Background and Purpose. The literature on teaching statistics at the university level provides support for the use of tools such as graphic organizers to help students understand the often confusing and sometimes complex process of selecting the appropriate statistical test for a given research question or design. This is the first of 2 reports that present several graphic organizers (also known as decision trees or flow charts) as decision-making guides for students to aid in the selection of appropriate statistical tests representative of those commonly used in physical therapy. Method/Model Description and Evaluation. To use the graphic organizers, the student learns 4 fundamental statistical concepts and then follows a decision-making process that leads to an appropriate statistical analysis. A prospective 2-group, reverse-design study on students who were familiar with the statistical textbook but not with the graphical organizer showed significant differences between the graphical organizer and textbook in both the time taken to select an appropriate statistical test and student preference. The graphic organizer led to correct statistical selection 3 times more often than the textbook. Outcomes. For 4 years, we have used the graphic organizer in the curriculum for students enrolled in a Master of Physical Therapy (MPT) program, receiving excellent student feedback. Discussion and Conclusion. The graphic organizers presented in this publication can be used to simplify the decision-making process for professional (entry-level) or postprofessional physical therapist students in choosing a statistical analysis for research questions that investigate group differences. The graphic organizers are convenient to use and can provide a visual organization of the decision-making process while highlighting the concepts discussed in a traditional statistics course. Key Words: Graphic organizer, Statistics, Research design, Group differences. BACKGROUND AND PURPOSE This is the first of 2 reports that presents a decision-making guide for the appropriate matching of statistical tests to research designs. The article presents several graphic organizers (also known as decision trees or flow charts) that focus on research designs and statistical approaches representative of those commonly used in physical therapy. We based the content of the graphic organizers on a compilation of statistical analyses reported in Physical Therapy and the Journal of Orthopedic and Sports Physical Therapy from January 2001 through June 2002. This analysis showed that approximately 65% of articles examined group differences, while approximately 35% were correlation/ regression designs (Table 1). This paper will focus on statistics for the former. The literature on teaching statistics at the university level provides support for the use of tools such as graphic organizers to help students understand the often confusing and sometimes complex process of selecting the appropriate statistical test for a given research question or design. As suggested by Pan and Tang,1 the use of a statistics cheat sheet can decrease graduate students' statistics anxiety without damaging the learning process, since the students must know the concepts in order to use the information. Additionally, an illustration that is clear, self-contained, easily grasped, and provokes reflection may enhance the Aha! experience of the statistics student.2 Several authors3-6 have noted that even though students can learn how to use statistical analysis, they often do not know how to apply the analysis to research scenarios. Statistical tests are commonly taught 1 test at a time without selection skills;7 the integration of selection skills could enhance the practical application of that knowledge.3-6,8 Manual computational procedures may not be as necessary to learn with the availability of computer programs, but instructors still need to help students develop skills necessary to select the appropriate statistical test. …