Attitudes of Medical Students Toward Statistics in Medical Research: Evidence From Saudi Arabia

Abstract

Today, most medical research depends on statistics, and interpreting these is crucial as the world tries to improve health outcomes. Medical students are at the front line in this effort but many have negative attitudes toward statistics. To assess this and offer recommendations, this study evaluates the attitudes of medical students in Riyadh, Saudi Arabia, toward statistics in medical research. The factors of sex, age, and program year are analyzed. In total, 327 medical students completed the Attitudes Toward Statistics in Medical Research survey. The results show that students somewhat appreciate the value of statistics in their professional careers but have negative to neutral feelings toward statistics, their own intellectual knowledge and skills in statistics, and the difficulty of the subject. Males show more positive attitudes than females, while students 23 or older perceive statistics as more difficult, and have more negative feelings toward statistics than younger students. No effect is found for the program year. To address this situation, medical educators should design courses in medical statistics that stimulate student understanding in this field, emphasizing its importance. Innovative methods of delivering course material should be explored to improve students’ sense of competency. Supplementary materials for this article are available online.

Keywords:

• Medical educators
• Medical student competency
• Statistics course material

1 Introduction

Medical statistics is a branch of applied statistics involving the application of statistical techniques in the design, analysis, and interpretation of public health and medical research data. The interpretation of a large amount of data in the health field depends, to a large degree, on the statistical methods chosen and how the data are used to test hypotheses and estimate associations. The role of statistics in the medical field is well known and most medical education curricula incorporate statistics courses at both the undergraduate and postgraduate levels (Sami 2010). However, in today’s environment, the interpretation and understanding of such statistics are crucial as the world tries to improve health outcomes. Medical students are the front line in this effort but these students differ in their attitudes toward statistics, with many experiencing statistics anxiety (Onwuegbuzie and Seaman 1995; West and Ficalora 2007; Beurze et al. 2013; Hannigan, Hegarty, and McGrath 2014).

Attitudes can be defined as “affective responses that involve negative or positive feelings of moderate intensity” (McLeod 1992, p. 581), whereas statistics anxiety can be defined as “the apprehension that occurs when an individual is exposed to statistics content or problems and instructional situations, or evaluative contexts that deal with statistics” (Macher et al. 2015, p. 1). Unfortunately, the relationship between students’ attitudes toward statistics and statistics anxiety has been found to be very strong (Mji and Onwuegbuzie 2004). Consequently, many students believe that statistical courses are a major barrier to getting their degrees, and low performance in these courses has become a problem in educational institutions worldwide. Students with negative attitudes toward statistics and high anxiety are often more likely to postpone or avoid learning data analysis, taking statistics courses, or completing statistical assignments. This has a direct effect on their performance as “Positive attitudes keep us using what we have learned. They also encourage us to seek opportunities to learn more” (Ramirez, Schau, and Emmioğlu 2012, p. 67). A study of 520 first and second year medical students enrolled in a research methodology class revealed a somewhat different relationship between statistical anxiety and course performance (Beurze et al. 2013). The study concluded that when the statistical anxiety was moderate, there was only a relatively small effect on student performance (Beurze et al. 2013).

Several studies have attempted to identify the factors that influence students’ attitudes toward statistics, however, there is a lack of agreement among these studies. Some studies have found that student performance in previous mathematics courses was a strong predictor of student attitudes toward and anxiety about statistics (Baloğlu 2003; Carmona, Martínez, and Sánchez 2005; Beurze et al. 2013; Hannigan, Hegarty, and McGrath 2014). Additionally, older students have been shown to have more negative attitudes toward statistics than younger students, suggesting that age is a predictor of attitude (Baloğlu 2003). Another study comparing attitudes toward biostatistics between students in clinical and nonclinical fields found that students with clinical academic specialties reflected more negative attitudes (Zhang et al. 2012). Moreover, female students reported a discernibly (i.e., significantly) higher level of anxiety (Beurze et al. 2013) and more negative attitudes toward statistics (Mills 2004) than their male counterparts. In contrast, other scholars have found no difference in attitudes between males and females (Cherian and Glencross 1997; Carnell 2008; Mji 2009).

Since negative attitudes are associated with low academic performance in statistics courses (Zhang et al. 2012), evaluating attitudes toward a course can help instructors provide students with better guidance, thereby improving student success. As such, it is important to examine student attitudes toward statistics as these attitudes can affect their future ability to utilize these tools, as well as the future success of the entire healthcare field (Hilton, Schau, and Olsen 2004; Beurze et al. 2013).

Although studies have evaluated student attitudes toward statistics, attitudes toward statistics in specific research methodology classes, such as medical research programs (MRPs), have not been thoroughly investigated. Most studies have explored attitudes toward statistics among college students in different academic disciplines, with only a few evaluating student attitudes in the health profession (Zhang et al. 2012; Beurze et al. 2013; Hannigan, Hegarty, and McGrath 2014). Additionally, no study has pursued this among students in Saudi Arabia. Furthermore, studies assessing the relationships between student attitudes toward statistics and their demographic and clinical characteristics have yielded conflicting results. Therefore, this study is designed to evaluate the attitudes of medical students toward statistics in medical research and to assess three factors that may affect these attitudes among students in Saudi Arabia.

2 Methods

The study uses a cross-sectional survey design and was conducted between December 2017 and April 2018. Participants were undergraduate medical students enrolled in either the first or second year of the MRP at a large health sciences university in Riyadh, Saudi Arabia. The total number of undergraduate medical students enrolled in the program during the academic year 2017–2018 was 600. The choice to include all medical students enrolled in the MRP in the survey adds to the novelty of this study. While previous research evaluated student attitudes among those enrolled in statistics courses, ours evaluates a larger population size of students enrolled in the health-related research program.

The MRP is a required component of the curriculum where students have to conduct one research project running for a maximum of two years (initially started as a four-year program and then shortened to a two-year program) (Althubaiti 2015) (see Appendix A in the supplementary materials for program details). In year one of the MRP, students identify their area of interest, choose a medical research supervisor, complete a literature review, plan a methodology, and submit a research proposal to the College of Medicine Research Committee. A series of research-based and introductory sessions to the principles of statistics are given throughout the year (e.g., frequency distribution, descriptive statistics, correlation, t-tests, and sample size calculations). In year two, several additional sessions are offered to students to support them in implementing their research projects. These include a series of computer-lab sessions using statistical programs such as SPSS and JMP and hands-on sessions to teach students how to develop descriptive and inferential statistics. The instructors for these sessions are from different disciplines (e.g., statistics, epidemiology, and medicine).

2.1 The Attitudes Toward Statistics in Medical Research Survey

The Attitudes Toward Statistics in Medical Research (ATSMR) survey (Dexter 2000) was developed to explore ATSMR among healthcare students in medical, nursing, and physical and occupational therapy programs at the University of New Mexico (see Table 1). It includes 32 items measured on a 7-point Likert-type scale (strongly disagree = 1, neither agree nor disagree = 4, strongly agree = 7). The survey covers four attitude components: Affect (positive and negative feelings concerning statistics: nine items), (e.g., I like statistics); Cognitive Competence (students’ attitudes about their own intellectual knowledge and skills applied to statistics: seven items), (e.g., I have no trouble interpreting statistical information in health science journals); Value (attitudes about the usefulness, relevance, and worth of statistics in one’s personal and professional life: eight items), (e.g., statistical thinking will not be necessary in my professional life); and Difficulty (attitudes about the difficulty of statistics as a subject: eight items), (e.g., statistical analyses are extremely complicated). Higher scores on the ATSMR subscales correspond to more positive attitudes toward statistics and vice versa. The ATSMR was developed based on the well-known and widely used Survey of Attitudes Toward Statistics (SATS) (Schau et al. 1995). The SATS measures attitudes relating to the actual use of statistical computation. However, ATSMR measures attitudes relating to the use and interpretation of statistics specifically in the medical literature (Dexter 2000).

Table 1 Subscales and items in ATSMR for health professions instrument.

Subscale	Description	Item
Affect	Positive and negative feelings concerning statistics	I like statistics. (+)
		I get frustrated when I read statistical results in medical journals. (-)
		Statistics does not scare me at all. (+)
		I enjoy reading statistical results in health science journals. (+)
		I feel intimidated by the statistical information I read in journal articles. (-)
		I dread having to read statistical results in journals. (-)
		I feel self-confident when reading statistical information in journals. (+)
		Trying to understand statistical information in journals is stimulating to me. (+)
		Reading statistical results in journals makes me feel uncomfortable. (-)
Cognitive Competence	Students’ attitudes about their own intellectual knowledge and skills when applied to statistics	I have no trouble with interpreting statistical information in health science journals. (+)
		I have difficulty understanding statistical results because of how I think. (-)
		Statistics does not make sense to me. (-)
		I have no difficulty understanding the statistical results in the studies I read. (+)
		There is too much math in statistics for me to understand it. (-)
		If I apply myself, I can understand the statistical results I read in journals. (+)
		I find it easy to understand statistics concepts. (+)
Value	Attitudes about the usefulness, relevance and worth of statistics in one’s personal and professional life	The statistical information in the studies I read is not that important. (-)
		Statistics are a necessary part of quantitative medical research. (+)
		Interpreting statistical results in journals should be a required part of my professional training. (+)
		Understanding statistical concepts will make me more successful as a clinician. (+)
		Statistics is not useful for most health care professionals. (-)
		Statistical thinking will not be necessary in my professional life. (-)
		Statistical analysis of clinical data has been an important factor in the evolution of modern health care. (+)
		As a professional, I will have no use for statistics. (-)
Difficulty	Attitudes about the difficulty of statistics as a subject	Learning to interpret statistical information requires a great deal of discipline. (-)
		The statistical results in health science journals are easy to understand. (+)
		Most people can learn to interpret statistical results quickly. (+)
		Statistics involves massive computations. (-)
		Statistical results are highly technical. (-)
		Most people have a hard time understanding the statistical information presented in health science journals. (-)
		Statistical analyses are extremely complicated. (-)
		One would have to be a statistician to really understand the statistical information in most medical journals. (-)

NOTE: + indicates positively worded items; - indicates negatively worded items.

Since our subjects were medical students, we chose the ATSMR survey designed specifically to measure the attitudes of such students toward statistics in medical research. Moreover, as these medical students were enrolled in an MRP where performance is not measured via a written statistics exam nor is statistics a topic presented in depth, the commonly used SATS was not appropriate.

The ATSMR survey was distributed during normal class activities to all 600 students enrolled in the program. Data were collected approximately 10 weeks into the semester for both first year and second year students to ensure that all students had some exposure to the statistics sessions.

2.2 Ethical Consideration

Students were informed (using consent forms and cover letter) about the aim of the study and were ensured anonymity and confidentiality. In addition, they were assured that participation would not affect their academic progress. Participation in the study was voluntary. Data were kept confidential and were used only for the study’s aim. After the ethical review, the study was approved by the Institutional Review Board at the King Abdullah International Medical Research Center, Riyadh, Saudi Arabia.

2.3 Internal Reliability

The reliability of the measuring instrument was determined for the specific sample used in this study. The reliability scores (Cronbach’s alpha) for the overall ATSMR scale and its subscales ranged between 0.72 (Difficulty) and 0.92 (overall scale) (Dexter 2000). During the piloting of the survey, subscales demonstrated two-week test–retest reliabilities between 0.69 (Cognitive Competence) and 0.89 (Affect) and the language used for the items was deemed appropriate for the medical students. Hence, the alpha values were sufficiently high to suggest scale reliability; thus, no items were deleted from the subscales.

2.4 Influential Factors

Based on the literature (Baloğlu 2003; Mills 2004; Beurze et al. 2013), three factors that could influence students’ attitudes toward statistics were analyzed: age, sex, and research program year. Additional reasons for this were as follows.

In terms of age, in addition to enrolling high school graduates, the university offers the opportunity to study for a degree in medicine to those with bachelor’s degrees in scientific fields (e.g., science, applied medical sciences, pharmacy) and holders of such degrees are generally from an older age group ($\ge$23 years old). As the university requires that all students take an undergraduate statistics course, the surveyed medical students would have some prior exposure to statistics. However, students from the older age group would also have applied statistics experience during their professional careers and/or in their research submitted to fulfill the requirements of their previous bachelor’s degree.

In terms of the program research year, this could be a factor as students in year one would be enrolled in preliminary statistics courses while those in year two would be taking practical statistical analysis and hands-on sessions. Finally, in terms of sex, this is a common factor of interest when exploring student attitudes toward statistics.

The study explored the medical students’ attitudes toward statistics and investigated the differences in the ATSMR attitude subscales, Affect, Cognitive Competency, Difficulty, and Value, by sex (male vs. female), age (younger vs. older students), and program year (year one vs. year two of the MRP). The interaction among sex, age, and year of program in relation to the attitude subscales was also examined.

2.5 Statistical Tests

Prior to the analysis, responses to negatively worded items were reversed and item scores summed within each subscale and then divided by the number of actual items in the subscale to compute subscale scores. The overall percentage of cases with missing data was low (<5%), and no specific pattern was discerned among the missing values analyzed in terms of the assessed factors or outcome subscales. Hence, cases with missing values were eliminated from the data using the listwise deletion method. The data were viewed in terms of the mean and standard deviation (SD) for continuous variables, and percentages and frequencies for categorical variables. Based on the skewness and kurtosis values that were used to determine if data were normally distributed, parametric inferences were used. Data were screened for univariate and multivariate outliers.

The bivariate associations between all variables in the study were examined. Pearson’s correlation coefficient was used to analyze correlations between the ATSMR subscales. Associations between the factors were tested using chi-square tests. A split-plot analysis of variance (ANOVA) design was applied using the independent variables sex, age (<23, $\ge$23), and year of MRP (year one and year two) as the between-subjects variables and the attitude toward statistics subscale scores as the within-subject variable or the repeated factor. The full model was fitted allowing for the evaluation of all main effects and interactions (Forstmeier and Schielzeth 2011). Simple effects tests were conducted when the interactions effects were statistically discernible (significant) using a series of one-way repeated ANOVA measures followed by Tukey’s HSD tests and Bonferroni adjustments. Partial η², which represents the proportion of the total variance attributable to the effect, was used as an estimate of effect size for the split-plot ANOVA (partial ${\eta }^2$: small = 0.01, medium = 0.06, and large = 0.14) (Cohen 1988). Cohen’s d was used to estimate the magnitude of difference between the comparative groups, where effect sizes of 0.8 or greater were considered large, those of 0.5 medium, and less than 0.2 small (Cohen 1988). Mauchly’s test of sphericity was performed to test variance heterogeneity among the attitude subscales. When sphericity was violated, the Greenhouse–Geisser correction was used. A p-value < 0.05 was considered statistically discernible. Data were analyzed using the SPSS statistical package version 25 (SPSS Inc., Chicago, IL, USA).

3 Results

3.1 Student Characteristics

A total of 327 students completed the questionnaire, with an overall response rate of 54.5%. The mean age was 22.5 years (SD = 1.18), ranging from 21 to 27. The majority of the participants were male (n = 201; 61.5%) and 23 years or older (69.7%); 65.7% were in their first year and 34.3% were in their second year (Table 2). Moreover, of those in first year, 200 (or 93%) were younger than 23, and of those in second year, 84 (or 75%) were 23 years or older.

Table 2 Survey respondents by age, gender, and year of medical research program.

Student characteristics	Frequency	Percentage
Age
<23 years	228	69.7
$\ge$23 years	99	30.3
Sex
Male	201	61.5
Female	126	38.5
Year of medical research program
Year one	215	65.7
Year two	112	34.3

NOTE: Total number of participants = 327.

3.2 ATSMR Scores

The descriptive statistics reporting the attitudes of students toward statistics are shown in Table 3 looking at the three influential factors. The mean score of the subscale Value was 4.67 (0.41), with the scores ranging from 3.75 to 5.38. Although the Value scores were the highest among the four components, looking at the mean score, most students neither agreed nor disagreed (or slightly agreed) on the usefulness, relevance, and worth of statistics in their personal and professional life. Their attitudes toward statistics as measured by the Affect subscale scored the lowest, reflecting a significant and statistically discernible negative attitude (mean = 2.44, SD = 0.52, range 1.33–4.33). In addition, they tended to consider statistics difficult (mean = 2.82, SD = 0.44, range 1.88–5). Students also had negative to neutral attitudes regarding their intellectual knowledge and skills in statistics (mean = 3.07, SD = 0.32, range 2.57–4).

Table 3 Mean (standard deviation) of students’ attitudes toward statistics by sex, age, and year of medical research program.

	All students (n = 327)	Males (n = 201)	Females (n = 126)	Age < 23 years (n = 228)	Age $\ge$ 23 years (n = 99)	Year 1 (n = 215)	Year 2 (n = 112)
Affect	2.44	2.53	2.29	2.50	2.29	2.39	2.33
	(0.52)	(0.54)	(0.44)	(0.53)	(0.46)	(0.43)	(0.38)
Cognitive	3.07	3.08	3.05	3.07	3.06	3.07	3.03
Competence	(0.32)	(0.32)	(0.31)	(0.31)	(0.33)	(0.29)	(0.26)
Value	4.67	4.68	4.66	4.66	4.71	4.65	4.71
	(0.41)	(0.38)	(0.38)	(0.38)	(0.39)	(0.37)	(0.38)
Difficulty	2.82	2.80	2.84	2.92	2.59	2.88	2.69
	(0.44)	(0.46)	(0.41)	(0.42)	(0.41)	(0.43)	(0.37)

NOTE: Total number of participants = 327.

3.3 Relationship Among Subscales

Table 4 displays the correlations among the ATSMR subscales. Cognitive Competence was moderately positively correlated to the Affect subscale (r = 0.32, p-value < 0.001) and weakly positively correlated to the Difficulty subscale (r = 0.23, p-value = 0.03). The Affect subscale was moderately positively correlated to the Value subscale (r = 0.3, p-value = 0.01). Correlations between other combinations of subscales were not statistically discernible.

Table 4 Correlations among the ATSMR subscales.

Subscales	Affect	Cognitive Competence	Value
Affect
Cognitive Competence	0.32${}^{\ast }$
Value	0.30${}^{\ast }$	0.10
Difficulty	0.12	0.23${}^{\ast }$	0.01

NOTE: Total number of participants = 327.

* Pearson product-moment correlation coefficients discernible at p < 0.05.

3.4 Sex, Age, and Year of Research Program

Mauchly’s test of sphericity indicated that the assumption of sphericity had been violated (χ²(5) = 59.6, p-value < 0.001). Therefore, the Greenhouse–Geisser epsilon adjustment to the degrees of freedom was used ($ϵ=0.89$). We found a statistically discernible interaction between sex and attitude [F(2.7, 859.9) = 4.3, p-value = 0.007, partial ${\eta }^2$ = 0.013]. Table 5 shows the results of the ANOVA. The simple effects analysis using one-way repeated ANOVA measures showed that attitude differed discernibly among the subscales for both male and female students ([F(3, 200) = 1027.24, p < 0.001] and [F(3, 200) = 1067.48, p < 0.001], respectively). Separate Tukey post-hoc analyses for males and females revealed discernible differences in all pairwise comparisons for both male and female students. Simple effects analyses also revealed that male and female students differed in their attitudes on the Affect subscale [F(1, 319) = 8.45, p-value = 0.004]. Males demonstrated discernibly more positive attitudes in their feelings toward statistics (mean = 2.53, SD = 0.54) than females did (mean = 2.29, SD = 0.44). The difference had a medium effect (p-value < 0.001, d = 0.49).

Table 5 ANOVA results for medical student data.

Source	df	F	p	Partial-${\eta }^2$
Within-subjects effects
Attitude	2.7	902.419	<0.001	0.739
Sex*attitude	2.7	4.336	0.007	0.013
Age*attitude	2.7	9.417	<0.001	0.029
Year*attitude	2.7	1.482	0.22	0.005
Sex*age*attitude	2.7	0.123	0.73	0.001
Sex*year*attitude	2.7	1.376	0.25	0.004
Age*year*attitude	2.7	0.613	0.59	0.002
Sex*year*age*attitude	2.7	0.709	0.53	0.002
Error	859.9
Between-subjects effects
Sex	1	1.810	0.18	0.006
Age	1	16.872	<0.001	0.050
Year	1	2.027	0.16	0.006
Error	319

NOTE: Greenhouse–Geisser values as sphericity cannot be assumed (p < 0.001).

In addition, there was a discernible interaction between age and attitude [F(2.7, 859.9) = 9.42, p < 0.001, partial $\eta$² = 0.03]. Tukey’s post-hoc analysis for younger (<23 years) and older students ($\ge$23) separately revealed discernible differences in all pairwise comparisons (all p-values < 0.001) (Table 3). The youngest students (<23 years of age) differed discernibly (mean = 2.50, SD = 0.53) with regard to their feelings toward statistics from the oldest students ($\ge$23 years of age) (mean = 2.29, SD = 0.46). In general, they had more positive attitudes on the Affect subscale (p-value = 0.001, d = 0.42). The youngest students also regarded statistics as less difficult (mean = 2.92, SD = 0.42) than the older students did (mean = 2.59, SD = 0.41). The difference had a large effect (p-value < 0.001, d = 0.8). The main effects of program year and all other interactions of sex, age, and program year were not statistically discernible.

4 Discussion

The aim of this study was to assess current attitudes among medical students toward statistics in medical research as the importance of healthcare statistics continues to grow. Specifically, we explored the influence of the variables sex, age, and year of enrollment in the relevant program on medical students’ attitudes. Most studies have investigated this topic using the traditional SATS-36 to gauge student attitudes among those enrolled in statistics courses rather than in health-related research programs. The ATSMR survey used here has an advantage in that it evaluates attitudes as they relate to statistics but specifically in the context of medical research rather than actual statistical computation (Dexter 2000). As yet, the ATSMR as an instrument has not been widely used, but the results here provide some insight into how it could be used in similar research in other countries.

Overall, medical undergraduate students in this study had negative to neutral attitudes toward statistics. Moreover, the mean scores on the Affect and Difficulty subscales fell below the median score of 4 (neutral attitude) on the Likert scale (2.44 and 2.82, respectively). In similar studies that have included a sample of medical or nursing students in a statistics course, Affect and Difficulty subscales also have reflected negative attitudes from these students (Williams 2010; Zhang et al. 2012; Hannigan, Hegarty, and McGrath 2014). The Cognitive Competence subscale mean score was 3.07, slightly below the scale’s average. A similar result was reported in Dexter (2000), although most students indicated more positive attitudes toward Cognitive Competence compared with the Affect and Difficulty components. In comparison with the Affect, Cognitive Competence, and Difficulty subscales, the Value subscale had the highest mean score (mean = 4.67, SD = 0.41), a result similar to that reported by Zhang et al. (2012) and Dexter (2000). The mean score (4.67) for the Value subscale reflects that students chose a middle ground, not reflecting either real positive or negative attitudes toward the value of statistics. This suggests that students may value statistics somewhat in their personal and professional lives. Miles et al.’s (2010) results indicated that physicians did not understand the value and relevance of statistics in their undergraduate training years. Greater exposure to the importance of statistics in research and in treatment decisions as well as their role in the evaluation of research during one’s career (Miles et al. 2010; Williams 2010) is needed. One approach could be to present examples from scientific papers in different domains during the students’ regular, evidence-based, and problem-based learning classes.

All the subscales had no discernible or low/moderate correlation. Therefore, a combined single summated score for the ATSMR survey would not be recommended. In the existing literature on attitudes toward statistics, the magnitude and pattern of the correlations between the Affect, Cognitive Competence, Value, and Difficulty subscales are neither consistent (Vanhoof et al. 2011), nor commonly reported. The Cognitive Competence subscale was correlated to the Affect and Difficulty subscales. These findings show that students reporting better positive attitudes about their own intellectual knowledge and skills in statistics also reported better positive feelings about statistics, tending to view statistics as less difficult. Intervention methods to enhance overall student attitudes toward statistics should aim at changing their perceptions of their own competency to give students greater confidence in their abilities in this area, which could then translate to more positive attitudes and value. Moreover, it is worthwhile exploring the reasons for the negative feelings about statistics to determine the role course instructors can play in affecting these feelings.

No discernible effect was found for the program year on the attitudes toward statistics. Since students in first year of program were younger than students in second year of program, this suggests that when controlling for age, the program year had little effect on student attitudes. Older students had more negative attitudes regarding Affect and Difficulty than younger students did. This finding is consistent with that of previous studies (Koller, Baumert, and Schnabel 2001; Baloğlu 2003) where students 23 years or older perceived statistics as more complicated than younger students. The older students in this study had degrees in science, or a science-related subject, and had statistical experience in previous research. The implication is that experience does not necessarily improve student attitudes, contradicting the findings of Zhang et al. (2012) who stated that medical students with research experience tended to have more positive attitudes. Male students were noted to have discernibly and significantly more positive feelings (Affect subscale) toward statistics than females. This is similar to Dexter’s results (2000) that indicated that male attitudes were discernibly more positive in terms of Affect than female attitudes. However, in most previous research on students in other disciplines, no discernible difference in statistics attitudes based on sex has been found (Hilton, Schau, and Olsen 2004; Carnell 2008; Mji 2009; Kiekkas et al. 2015). In a study examining 121 first-year entry medical students, males revealed more positive attitudes about statistics, but without any statistically discernible sex differences (Hannigan, Hegarty, and McGrath 2014). The sex differences in this study may not be based on sex but on instruction. In this university in Riyadh, Saudi Arabia, the MRP is implemented separately for males and females using different instructors and sessions. As a result, differences in attitudes may also indicate differences in the implementation of the program. In further studies, student attitudes and any correlation between specific views and sex should be explored in more detail.

Based on the results of this study, it appears that a number of methods can be adopted to improve student attitudes, motivation, and perceived difficulties in applying and understanding statistics in health research. For example, increasing student confidence in their competency in the subject matter, improving course content and design (DeVaney 2009; Gundlach et al. 2015; Hund and Getrich 2015), and offering special assistance (e.g., additional one-on-one statistical consultation sessions) to the neediest subgroups (i.e., women and older students) to ultimately change the trajectory of student attitudes toward statistics.

Today’s environment emphasizes the important role that statistics plays in solving global health crises, where statistical figures and predictions are used frequently in the media and the community. Our hope is that this can trigger future medical student interest in and understanding of the importance of statistics to the healthcare environment.

5 Limitations and Future Study

As with all studies, this study has some limitations that should be mentioned. First, the study used a survey approach to collect data, which could have introduced response bias (Althubaiti 2016). The nonresponse rate was high as completing the survey was not mandatory. Moreover, medical students are recognized as a group with low survey response rates due to their demanding studies and schedule (Grava-Gubins and Scott 2008); therefore, the results should be interpreted with caution. Second, although not explored here, there are conceivably many other possible factors affecting attitudes toward statistics such as socio-economic status, success in prior statistics coursework, or GPA scores. This study was designed to explore how sex, age, and year of MRP might influence attitudes after filtering out other possible factors of interest in the pilot. For example, when collecting previous grades in statistics courses, the average grade in undergraduate statistics courses was an A- as reported by these students. This could mean that these students had more positive attitudes toward statistics than others would and that their undergraduate experience may not have influenced their attitude. Thus, future studies need to explore other variables that may affect student attitudes. Finally, this was the first study to explore attitudes of medical students in this region using the ATSMR survey. Data were collected at a single university, such that the results could not be generalized to other student samples in Saudi Arabia. Future research could examine possible differences in attitudes across different universities in the region. Moreover, longitudinal research could be crucial in assessing medical students’ attitudes toward statistics as well as their evidence-based medical performance going forward.

Acknowledgments

The author would like to thank Dr. Candace Schau and Dr. James G. Dexter for their invaluable assistance in providing the ATSMR survey.

Supplementary Materials

Medical research program component.