Statistics Used in Scholarly Projects of Doctor of Nursing Practice Graduates

Saturday, 28 October 2017: 3:35 PM

Aliya Kuerban, PhD
School of Nursing, Molloy College, Rockville Centre, NY, USA




Doctor of nursing practice (DNP) students’ scholarly projects focus on practical application of existing knowledge (AACN, 2006a and AACN, 2015a). DNP students are required to complete a scholarly project upon completion of their program, which is equivalent to a dissertation requirement for students of doctors of philosophy (PhD). This project is considered as the synthesis of their DNP education process (Nelson et al., 2013). In order to complete DNP project, biostatistics skills are foundational. Without appropriate training in biostatistics, with limited literacy and reasoning ability in quantitative study, one will not be able to consume current research critically, which will then affect the accurate application of the knowledge. Most of the DNP programs only offer one semester of three credits biostatistics course (Kuerban, 2016). With a limited time, faculty members face a substantial challenge of covering vast topics on biostatistics. At the same time, the depth and width of the teaching is also needed to be balanced carefully within a one course time frame. Currently, there is no study done to explore statistics methods used commonly in DNP projects. A study that surveys DNP projects to find out the most often used statistical methods will deliver the most relevant information to both DNP faculty and students in terms of teaching and learning statistical skills within a short allowable time.


This study utilized the method of quantitative content analysis to review DNP projects completed recently. Projects submitted voluntarily to two main repositories were selected. These two repositories were Doctor of Nursing Practice Inc. website and Digital Commons by Bepress. Doctor of Nursing Practice Inc. is a not for profit organization and it maintains a list of DNP projects submitted since 2007. This list of scholarly projects served as one sampling frame of the study. However, due to only a small amount of projects submitted, only 50 projects with sufficient information on statistics methods used were selectively included in this study. Most of these projects were submitted from 2009 to 2013. Another repository used was Digital Commons by Bepress, which has a much larger list of projects. Fifty projects submitted in 2016 were randomly selected by using a random number generator. Projects from Digital Commons were more recent than the projects selected from Doctor of Nursing Practice Inc. Although this is not a longitudinal study, the differences in the time frame of the two groups of projects selected can provide insights into whether there are changes over time. With two lists of 50 projects from two repositories, content analysis was conducted to retrieve information regarding the subjects of the study, sample size, nature of the study, sampling strategies, and statistics methods used. After the completion of data collection, Statistics Package for Social Study version 23.0 software (IBM Corp. Released 2015, Armonk, NY) was used to analyze the data.

Results: Twenty out of 50 projects from the Doctor of Nursing Practice Inc. website specified their sample size. Forty-five out of 50 projects from Digital Commons specified the sample size. When combining these two groups together, 46.2% of studies’ sample size is less than 50. 90% of the projects from both sources are quantitative in nature. The rest of the 10% of the projects are either literature review or using qualitative methods to conduct the study. In terms of study setting, 90% of the projects from both sources were conducted either in a hospital or at a clinic. In addition, 7.5% of the projects were conducted on online. Among the projects examined, very few studies used random sample selection and the majority of the studies used convenience or purposive sampling strategies. The results from both samples are comparable in terms of the subjects being studied. About 30% of the projects studied patients directly. And more than two fifths of the projects from both samples focused on the behavior of either nurses or providers (including medical doctors, physician’s assistants, and nurse practitioners). The most often used method is descriptive analysis, 67% of the projects utilized this method. t Tests and Chi square analysis were also used often. Other methods used were Fisher exact test, Mann-Whitney test, McNemar test, logistic regression, and odds ratio.


Cohen (1965, 1992) proposed and argued a theory regarding the relationship of power of the effect of the study and its sample size, which was well accepted and applied in academia research. Based on his theory, the sample size of a study influences the power of the study; a small sample size directly affects the outcome of the study and its generalizability. Button and his colleagues (2013) confirmed the theory that a study with low statistical power would generate more unreliable variation of a true effect and the power of the study was limited by the size of the sample. Among the projects examined, two issues were detected. First of all, among the projects submitted 2009-2013, 60% of them did not report the actual sample size, which was improved to only 10% among the projects submitted in 2016. Secondly, among the projects that did specify their sample size, almost half of them had a size of less than 50. DNP students had a very limited time and financial resources to conduct larger studies. Small sample size and low power will continue to be an issue of most DNP projects. However, the problem of small sample size does not limited to DNP projects; many translational researches and new clinical trials have the same issue. Bacchetti, Deeks, & McCune (2011) proposed a new perspective on the limitation of small sample size and argued about the flaws in the conventional requirement of Cohen’s theory of using 80% power for sample size calculation. With reasonable argument, the value of the DNP projects should not be negated by their smaller sample size. Broome, Riner & Allam (2013) recommended to DNP programs to use multiple sites and involve multiple students to increase sample size, which could also be an innovative way to explore and overcome the intrinsic issue of small sample size that most DNP projects face.

Across two time frames, the subjects of DNP projects were comparable. Although relatively higher percentages of projects focused directly on patients, a similar amount of attention was paid to the other parts of the health care team, which included nurses and health care providers (medical doctors, nurse practitioners, and physician’s assistant). This diverse focus should be encouraged. At the same time, DNP students should also venture out to conduct projects on informatics interventions, which was lacking among the projects reviewed.

A panel of statistics experts suggested to use descriptive statistics to summarize DNP project results (Hayat et al., 2013), which was verified in the DNP projects explored in this study. Descriptive method was used most often in the 100 DNP projects examined. This study focused on the statistics methods used in DNP projects, which are the methods the DNP students not only need to understand how to interpret but also how to use proficiently. The final list of the top statistics methods provided here should not be considered as the only methods to be taught in a biostatistics course for DNP students. Instead, it gives the instructor some thoughts in terms of how to spend the time wisely with the students: give more hands on assignments for students to practice to fully master the methods included in the list, and give a broader review of other less frequently used methods.