Commonly Used Statistical Methods in Occupational Therapy Research

Method

In a previous study, Roush et al. (2015) identified the top 25 statistical methods used within the physical therapy literature. These methods were used as the basis for this study to examine the occupational therapy literature. Three sources were used to identify high-quality, prolific, and high-impact journals in the field of occupational therapy. In the first source, Rodger et al. (2007) performed a descriptive design study in which 184 published authors were surveyed about their opinions on quality indicators for occupational therapy journals and which journals expressed those factors. In the second source, Potter (2010) used Bradford’s Law of Scattering to map the bibliographic patterns in the occupational therapy literature to identify the most-cited journals. This search revealed nine journals that represented approximately one-third of the 6,804 journal references identified as relevant to occupational therapy. In the final source, Brown et al. (2017) performed a bibliometric analysis of occupational therapy articles and authors published from 1991 to 2014. Among the multiple outcomes they evaluated was the most common journals cited in occupational therapy research.

Using the results from these three articles, we included a list of 12 journals in our analysis (Table 1). A 13th journal, the Open Journal of Occupational Therapy, was included because of wide readership as evidenced by a relatively good h5-index of 17 (Hirsch 2005). Articles were selected for data analysis if they were published from January 2018 to December 2018 and included research reports, systematic reviews, scientific articles, original contributions, clinical investigations, case reports, case studies, and brief reports. Articles were excluded if they were announcements, clinical commentaries, corrections, editorials, letters, opinions, and erratum.

Before the review of the journal articles to categorize the statistics used, an interrater reliability analysis was conducted to determine the level of agreement among the four researchers in identifying statistical methods because only one reviewer analyzed each article similar to Roush et al. (2015). Standards for reviewing the articles were decided on a priori and included examining the entire article as well as determining what type of research methodology was represented (qualitative, quantitative, or mixed) and what statistical tests were used. Investigators recorded the journal name, article name, authors, volume and issue numbers, and inclusive pages.

Any statistical tests found in the article were then documented in an Excel spreadsheet, and results were tallied for the incidence for which the statistical test was cited. Statistics were tallied individually, except for descriptive statistics that included mean, median, mode, range, standard deviation, standard error of the mean, and standard error of measurement (Table 2). Twenty articles were assigned to the researchers (n = 4) to independently record. Fleiss’s κ was used to determine the degree of agreement among raters for each statistical method (Fleiss, 1981). The 1,223 articles were then randomly assigned to the four reviewers for the entire 2018 calendar year.

Results

Fleiss’s κ interrater reliability analysis revealed a level of agreement among the raters that ranged from .77 to 1.0, indicating substantial agreement among reviewers (Fleiss, 1981). The top 25 statistical methods or categories of statistical methods, the number of articles in which they were present, and the percentage in which the representation equates to of all reviewed articles are presented in Table 2. The following top 10 statistical methods or categories of statistical methods were identified: (1) descriptive statistics, (2) t test, (3) confidence interval, (4) χ², (5) effect size, (6) analysis of variance, (7) parametric correlation, (8) Cronbach’s α, (9) post hoc analysis and pairwise comparisons, and (10) thematic analysis and coding.

Methodologies were also identified as quantitative, qualitative, or mixed methods (Table 3). Quantitative approaches were the most common, followed by mixed methods and then qualitative studies.

Discussion

Roush et al. (2015) analyzed the most commonly used statistical methods in physical therapy literature, and we replicated this idea in our study by examining statistical method usage in prominent occupational therapy literature. Because Roush et al. (2015) served as a precedent to this study, we used this literature to guide the inclusion of the top 25 categories of statistical methods that were coded. Roush et al. (2015) reported that their top 25 categories accounted for 97.77% of all the statistical methods used. In our analysis, we found that the top 26 categories accounted for 96.15% of all the statistical methods used, demonstrating remarkable consistency. Additionally, we added new categories of statistical methods that were prominent across occupational therapy literature in our coding that were not reported in Roush et al.’s (2015) study. The aim of this study parallels that of Roush et al.’s (2015) work in that both articles seek to increase knowledge of common statistics within the realms of occupational and physical therapy, respectively, and encourage educators in these fields to plan curriculum in accordance with the determined most common statistical methods.

The results of this study closely resembled those of Roush et al.’s (2015) work, with descriptive statistics being the most commonly used statistical measure, followed by t tests and confidence intervals. In fact, 25 of the top 26 statistical tests reported in our study were represented in Roush et al.’s (2015) article, indicating a large amount of overlap between the physical therapy and occupational therapy literature. The one statistical test commonly reported within the occupational therapy literature that is not found as frequently in the physical therapy literature is the use of thematic analysis and coding. This finding may be because of the larger number of qualitative and mixed-methods studies found within the occupational therapy literature; however, this conclusion is speculative because Roush et al. (2015) did not report on study design and methodology.

In this analysis of statistical methods, we used the studies conducted by Potter (2010), Brown et al. (2017), and Rodger et al. (2007) as a foundation of various journals’ relevance in the realm of occupational therapy. The 13 journals that we included in our study are journals deemed prominent in the field of occupational therapy by previous literature. In addition, we also go a step beyond the work of Potter (2010), Brown et al. (2017), and Rodger et al. (2007) by looking within these most prominent occupational therapy literary journals to determine prevalence among the statistical analyses used in the occupational therapy literature.

The use of evidence-based practice in occupational therapy is a strong pillar of the profession. AOTA (2024) described occupational therapy as a “science-driven, evidence-based profession that enables people of all ages to participate in daily living or live better with injury, illness, or disability” (para. 3). Despite this emphasis on evidence and research in the profession, a gap is present between occupational therapy literature and its application through evidence-based practice because of several barriers. One barrier is that many occupational therapy practitioners express a lack of confidence in their abilities to interpret and synthesize findings in research to adequately apply them to practice (Bennett et al., 2003).

Many occupational therapy practitioners also have noted that the accessibility of quality evidence-based research is often limited and that the presentation of data is not always feasible to decipher (Lyons et al., 2011). According to Brown et al. (2010), the strongest predictor of use of evidence-based practice and positive attitudes toward research implementation is the level of academic degree, with recent graduates more likely to apply evidence-based practice. It is crucial to address these barriers to promote the utilization of evidence-based practice among all occupational therapy practitioners, especially those who are less likely to implement research because of negative attitudes or poor confidence in abilities to analyze data.

In this study, we sought to minimize barriers to the utilization of evidence-based practice by increasing awareness of statistical use in the occupational therapy literature. Because a trend has been found in regard to lowered confidence in analyzing and processing data among some populations of occupational therapy practitioners, we aimed to consolidate data analysis to key statistical methods that should be a focus in occupational therapy curricula and continuing education (Bennett et al., 2003). Exploring prevalent statistical methods in notable occupational therapy literature can aid those performing research in the field of occupational therapy to understand common statistics and their application in research. Moreover, this awareness can influence the curriculum of occupational therapy students regarding the utilization of common statistical methods in research to help improve their learning of evidence-based practice.

Continuing education is a crucial component of the practice of occupational therapy and the profession’s advancement as a whole. For example, occupational therapy practitioners in the United States have continuing education requirements through accumulation of continuing education units to maintain their state licensure and national certifications through the National Board for Certification in Occupational Therapy (AOTA, 2025, p. 5). In addition to informing occupational therapy curriculum, this study provides guidance on the types of statistical methods to include in postgraduate continuing education opportunities. The occupational therapy profession emphasizes continuing education among practitioners of all expertise levels, and the development of continued statistical method learning opportunities for occupational therapy practitioners can help improve their utilization of evidence-based practice. This study demonstrates the prevalence of identified statistical methods in the occupational therapy literature and, thus, informs continuing education instructors on content that supports its consumers. Thus, an incorporation of statistical interpretation instruction into occupational therapy’s continuing education could be instrumental in closing the gap between occupational therapy practice and the utilization of research to inform practice.

Limitations

This study has several limitations. One limitation is that two of the sources used to find the high-impact occupational therapy journals were 8 yr or older at the time of the data collection, suggesting that more high-impact journals might be available at this time (Potter, 2010; Rodger et al., 2007). However, six of the nine high-impact journals listed in Potter’s (2010) study were also listed in Brown et al.’s (2017) study, indicating good consistency in the top-rated occupational therapy journals.

Another limitation of this study is the limited number of journals surveyed. Although the most cited and impactful journals were included, several others were excluded because of feasibility concerns. It was not feasible for us to count the statistics from all the 2018 articles in each of the journals listed.

A third limitation is that we tabulated the most used statistical methods, but we did not discern whether the methods were appropriately used. Although the purpose of this report was to identify the most commonly used statistical measures, researchers in a future study should evaluate whether the most commonly used measures are also the most appropriate.

A final limitation is the inclusion of thematic analysis and coding within the reported top-25 statistical methods. Although not a statistical method, thematic content analysis and coding are heavily documented in the occupational therapy literature as an analysis method of qualitative data. Therefore, in an effort to inform educators of prevalent statistical methods and recommendations for research content within occupational therapy practitioner curricula, we reported thematic content analysis and coding prevalence.

Conclusion

Occupational therapy practitioners often lack the knowledge, resources, and confidence to effectively implement statistical analysis into their everyday clinical practice. Therefore, in this study, we identified the most common statistical analyses found within the occupational therapy literature to better inform educational efforts in preparing occupational therapy practitioners for the use and interpretation of statistics in clinical practice. We also identified which statistical methods are the most critical to include in the occupational therapy practitioner curricula, which has the potential to guide occupational therapy educators in their course and curriculum designing process moving forward.