Read the article by Perepiczka et al. in this week’s assigned readings. Choose one of the major sections of the article to evaluate.

Walden University ScholarWorks

School of Counseling Publications College of Social and Behavioral Sciences

2011

Relationship Between Graduate Students’ Statistics Self-Efficacy, Statistics Anxiety, Attitude Toward Statistics, and Social Support Michelle Perepiczka Walden University

Nichelle Chandler

Michael Becerra

Follow this and additional works at: http://scholarworks.waldenu.edu/sc_pubs

Part of the Counseling Commons, and the Mental and Social Health Commons

This Article is brought to you for free and open access by the College of Social and Behavioral Sciences at ScholarWorks. It has been accepted for inclusion in School of Counseling Publications by an authorized administrator of ScholarWorks. For more information, please contact [email protected].

99

Relationship Between Graduate Students’ Statistics Self-Efficacy, Statistics Anxiety, Attitude Toward Statistics, and Social Support Michelle Perepiczka Nichelle Chandler Michael Becerra

Statistics plays an integral role in graduate programs. However, numerous intra- and interpersonal factors may lead to successful completion of needed coursework in this area. The authors examined the extent of the relationship between self-efficacy to learn statistics and statistics anxiety, attitude towards statistics, and social support of 166 graduate students enrolled in master’s and doctoral programs within colleges of education. Results indicated that statistics anxiety and attitude towards statistics were statistically significant predictors of self-efficacy to learn statistics, yet social support was not a statistically significant predictor of self-efficacy. Insight into how this population responds to statistics courses and implications for educators as well as students are presented.

Keywords: graduate students, statistics, anxiety, self-efficacy, attitudes, social support

More graduate programs in various social science fields are requiring students to complete research methods including statistics courses or a blended combination thereof (Davis, 2003; Schau, Stevens, Dauphinee, & Del Vecchio, 1995). These course requirements pose a dilemma for educators and students because many students perceive statistics as difficult and unpleasant (Berk & Nanda, 1998). Some students can struggle in statistics courses as a related complication of this perception as well as other intrapersonal factors related to the course.

To investigate graduate students’ experiences in statistics courses, researchers studied different avenues to understand what occurs with students so steps can be taken to improve learning as well as satisfaction in college statistics courses. For instance, researchers suggested non-cognitive factors such as motivation for further learning (Gal & Ginsburg, 1994; Finney & Schraw, 2003), statistics self-efficacy (Onwuegbuzie & Wilson, 2003), and attitude toward statistics (Araki & Schultz, 1995; Elmore, Lewis, & Bay, 1993; Waters, Martelli, Zakrajsek & Popovich, 1988; Wise, 1985) should be assessed and addressed with students. Finney and Schraw theorized that the difficulty students experience with statistics is not necessarily due to lack of intelligence or poor aptitude, but may be a result of the above mentioned factors. Bonilla (1997), Cohen and McKay (1984), and Solberg and Villarreal (1997) hypothesized that social support may act as a buffer against the development of these psychological manifestations.

The purpose of this study was to examine the various factors that have been introduced in previous research in one comprehensive study. The goal was to determine how graduate student self-efficacy to learn statistics is predicted by statistics anxiety, attitude toward statistics, and social support (Gall, Gall, & Borg, 2007). The overarching intent was to document graduate student self-efficacy to learn statistics and identify how certain variables influence statistics self-efficacy (Pan & Tang, 2005).

Self-Efficacy to Learn Statistics

In order to understand the implications of this research, an explanation of the key variables found in the literature review must first be discussed. Self-efficacy to learn statistics is the dependent variable in this study. Bandura (1977) originally defined general self-efficacy as one’s judgments of his or her capabilities to organize and carry out courses of action required to attain specific types of performances. Bandura asserted that self-efficacy beliefs are manifested from four primary sources, which include the following: (a) personal accomplishments, (b) vicarious learning experiences, (c) verbal persuasion, and (d) emotional arousal. These primary sources lay the foundation for building the concept of self-efficacy to learn statistics. Finney and Schraw (2003) defined self-efficacy to learn statistics and developed an assessment to measure

The Professional Counselor Volume 1, Issue 2 | Pages 99–108

© 2011 NBCC, Inc. & Affiliates www.nbcc.org

http://tpcjournal.nbcc.org doi:10.15241/mpa.1.2.99

Michelle Perepiczka, NCC, and Nichelle Chandler, NCC, are professors at Walden University. Michael Becerra, NCC, is an Assistant Professor at the University of Alabama. Correspondence can be addressed to Michelle Perepiczka, Walden University, School of Social Work and Human Services, 100 Washington Avenue South, Suite 900, Minneapolis, MN, 55401, [email protected].

100

The Professional Counselor Volume 1, Issue 2

this phenomenon. Self-efficacy to learn statistics is an individual’s confidence in his or her ability to successfully learn statistical skills necessary in a statistics course.

A large amount of information is available on self-efficacy related to academic performance (Lent, Brown, & Larkin, 1984, 1986; Pajares, 1996; Pajares & Miller, 1995; Zimmerman, 2000; Zimmerman, Bandura, & Martin-Pons, 1992). However, little is known specifically about self-efficacy to learn statistics. Finney and Schraw (2003) investigated whether self-efficacy to learn statistics is related to performance in a statistics course and whether self-efficacy to learn statistics increased during a 12-week introductory statistics course. One hundred and three undergraduate students from a large Midwestern university participated in the survey. Finney and Schraw reported a positive relationship between statistics self-efficacy and academic performance as well as an increase in self-efficacy to learn statistics over the duration of the course. Onwuegbuzie (2000) also reported students with the lowest levels of perceived competence had the highest levels of statistics anxiety. Additionally, Pajares and Miller (1995) documented an inverse relationship between self-efficacy and math anxiety.

Statistics Anxiety

Statistics anxiety is one of the three independent variables in this study. Researchers have documented a large amount of information on statistics anxiety over the years. For instance, there are multiple definitions of statistics anxiety available in the literature. Onwuegbuzie, DaRos, and Ryan (1997) defined statistics anxiety as “a state-anxiety reaction to any situation in which a student is confronted with statistics in any form and at any time” (p. 28). Cruise, Cash, and Bolton (1985) defined statistics anxiety as “the feelings of anxiety encountered when taking a statistics course or doing statistical analyses: that is, gathering, processing, and interpret[ing]” (p. 92). The latter is the definition utilized for this study.

We know that instructors of research and statistics courses often encounter students with high levels of statistics anxiety upon their arrival to class (Perney & Ravid, 1991). According to Onwuegbuzie, Slate, Paterson, Watson, and Schwartz (2000), 75% to 80% of graduate students in the social sciences appeared to experience high levels of statistics anxiety. Statistics anxiety was found to be higher among female and minority graduate students in comparison to their male and Caucasian counterparts (Onwuegbuzie, 1999; Zeidner, 1991).

Researchers identified three categories of variables—situational, dispositional, and environmental—that are related to statistics anxiety (Onwuegbuzie & Wilson, 2003). Situational antecedents are factors that surround the student, including previous statistics experiences (Sutarso, 1992). Researchers found a negative connection between the number of completed mathematics courses and statistics anxiety (Auzmendi, 1991; Robert & Saxe, 1982; Zeidner, 1991). Forte (1995) found minimal previous math experience, late introduction to quantitative analysis, anti-quantitative bias, lack of appropriation for the significance of analytical models, and lack of mental imagery were factors contributing to statistics anxiety among social work students.

Dispositional antecedents are intrapersonal factors students bring to the classroom (Onwuegbuzie & Daly, 1999), which includes issues such as perfectionism and perception of abilities at developmental stages in life (Pan & Tang, 2004). Walsh and Ugumba-Agwunobi (2002) found evaluation concern, fear of failure, and perfectionism provoked statistics anxiety. Environmental antecedents are interpersonal factors related to the classroom experience (Onwuegbuzie & Daly, 1999), which can include the student’s experiences with the professor. Tomazie and Katz (1988) reported previous experiences in statistics courses have influenced learning in a current course. Moreover, the environmental antecedent has the least research available in the literature.

Attitude Toward Statistics

Attitude toward statistics is the second independent variable in this study. Attitude towards statistics is defined in this study as a combination of a students’ attitude toward the use of statistics in their field of study and the students’ attitudes towards the statistics course (Cashin & Elmore, 1997; Wise, 1985). Researchers explored this area; however, there are many gaps left to fulfill. Gal and Gingsburg (1994) reported students often enter statistics courses with negative views or later develop negative feelings regarding the subject matter of statistics. Researchers found no statistically significant differences

The Professional Counselor Volume 1, Issue 2

101

among females’ and males’ attitudes towards statistics (Araki & Schultz, 1995; Cashin & Elmore, 2005; Harvey, Plake, & Wise, 1985). However, conflictingly, Waters et al. (1988) and Roberts and Saxe (1982) found male students had more positive attitudes towards statistics than female students.

According to Perney and Ravid (1991), statistics courses are viewed by most college students as a road block to obtaining their degree. Students often delay taking their statistic courses until the end of their program. Researchers found students’ negative attitudes toward statistics is an influencing factor in low student performance in statistics courses (Araki & Schultz, 1995; Elmore et al., 1993; Harvey et al., 1985; Schulz & Koshino, 1998; Robert & Saxe, 1982; Waters et al., 1988; Wise, 1985).

Perceived Social Support

Perceived social support is the final independent variable in this study. Perceived social support for this study is defined as the level of support an individual self identifies as received from friends, family, and significant others (Zimet, Dahlem, Zimet, & Farley, 1988). This variable is influential in this study in terms of the potential buffering effect it may have on the other independent variables, statistics anxiety and attitude towards statistics.

According to Bonilla (1997), social support acts as a buffer to dysfunctional thoughts or attitudes. In 1985, Cohen and Wills investigated the process through which social support has a beneficial effect on well-being. The buffering model maintains that support is related to well-being primarily for persons under stress. Cohen and Wills identified four support resources, which include the following: (a) esteem support such as the person is valued and accepted, (b) informational support, (c) social companionship such as engaging in leisurely activities with others, and (d) instrumental support such as an individual providing a person with financial aid, material resources, or need-based services.

Solberg and Villarreal (1997) conducted a study to explore the interactions between social support and physical as well as psychological distress of Latino college students. The authors reported social support moderated the distress. Specifically, the Latino students who believed social support was available had lower psychological distress than students who believed that social support was less accessible.

Research Questions

Six research questions were included in this study. The first four focus on descriptive information from our sample and include the following: (a) what is the graduate student self-efficacy level, (b) what is the graduate student statistics anxiety level, (c) what is the graduate student attitude toward statistics, and (d) what is the graduate student level of perceived social support? The predominate research question driving this study is, what is the extent of the relationship, if any, between graduate students’ self-efficacy to learn statistics and statistics anxiety, attitude towards statistics, and social support? A supplemental research question was, what is the influence of social support on statistics anxiety and attitude towards statistics?

Method

Participants Participants were recruited by the researcher emailing faculty members of doctoral and master’s programs within colleges of education at 250 universities within the United States. The faculty members were asked to forward information about the opportunity to participate in the study to their students. One hundred sixty-six graduate students within colleges of education representing 27 states fully completed the online survey within the 8-week data collection timeframe. An a priori power analysis was conducted considering involvement of three predictors in the multiple regression equation and estimating a moderate effect size based on similar studies. It was determined that 119 participants are needed to achieve adequate power in the study (Faul, 2006); thus, an appropriate sample size was achieved to obtain adequate power in the analysis (Gall et al., 2007).

The sample was predominately female (N = 136, 81.9%) compared to males (N = 30, 18.1%). Participants’ age ranged

102

The Professional Counselor Volume 1, Issue 2

from 21 to 71 with 34.4 as the mean age. The cultural makeup of the sample consisted of 4 Native American (2.4%), 4 Asian/Pacific Islander (2.4%), 24 African American (14.5%), 124 Caucasian (74.7%), and 10 Latino participants (6%).

The academic level of the participants was close to evenly split with 92 master’s students (55.4%) and 74 doctoral students (44.5%). The majority of the sample (N = 144, 86.7%) were enrolled in counseling or related educational programs such as mental health counseling, school counseling, rehabilitation counseling, student affairs, and counselor education and supervision. Twenty-two (13.3%) participants were enrolled in education graduate programs such as educational leadership, curriculum and instruction, and educational technology. One hundred thirty-six participants (81.9%) were enrolled in programs that were accredited by at least one accreditation body appropriate to their program.

Participants had different backgrounds in terms of taking statistics courses. The mean number of completed graduate statistics classes at the time of participating in the study was 1.63 classes for the sample. The range of courses was 0 to 6, and the mode was 0 classes with 45 participants (27.1%) not having completed a single graduate level statistics course. Of the 121 who completed a statistics course previously, the mean final grade was 89.34% with the lowest grade earned reported as 70%.

Instruments A demographic questionnaire was used to collect information related to participants’ personal characteristics as well as previous experiences with graduate statistics classes. The Self-Efficacy to Learn Statistics (SELS) scale was used to measure the dependent variable (Finney & Schraw, 2003). The SELS measures confidence in one’s ability to learn necessary statistics while in a statistics course in order to successfully complete 14 specific tasks using a 1 (no confidence at all) to a 6 (complete confidence) response scale. Only a total score is obtained from the instrument. Internal consistency reliability was reported as .975 Cronbach’s alpha. Validity evidence of SELS to other variables was reported. The SELS was positively correlated with the Math Self-Efficacy scale and negatively correlated to the general and statistics Test Anxiety Inventory subscale providing evidence of concurrent validity. The norm group for the instrument was a total of 154 college students enrolled in an introductory statistical methods course.

The Statistics Anxiety Rating Scale (STAR) was used to measure the independent variable statistics anxiety (Baloglu, 2002; Cruise &Wilkins, 1980). The assessment is a 51-item Likert scale ranging from 1 (no anxiety) to 5 (very much anxiety) and measures anxiety in two parts. The first part includes 23 statements related to statistics anxiety and the second part has 28 items related to dealing with statistics. A total score as well as six subscores including the following are generated with this instrument: Worth of Statistics, Interpretation Anxiety, Test and Class Anxiety, Computation Self- Concept, Fear of Asking for Help, and Fear of Statistics Teacher. Reliability for each of the subscales ranged between .68 to .94 with a median of .88 (Worth of Statistics .94, Interpretation Anxiety .87, Test and Class Anxiety .69, Computational Self-Concept .88, Fear of Asking for Help .89, and Fear of Statistics Teachers .80). Validity evidence of STARS to other variables was reported. The STARS had a strong correlation (r = .76) to the Math Anxiety Scale (Roberts & Bilderback, 1980). The instrument was normed with 1,150 university students enrolled in statistics courses.

The independent variable, attitude toward statistics, was measured by the Attitude Toward Statistics (ATS) scale (Schultz & Koshino, 1998). This is a 29 item, 5-point Likert scale ranging from strongly disagree to strongly agree. A total score and two subscale scores, Attitudes Toward the Field and Attitudes Toward the Course, are obtained from the instrument. Both subscales were reported as reliable with Cronbach’s alpha at .92 for Attitudes Toward the Field and .91 for Attitudes Toward the Course (Wise, 1985). The ATS was reported to have strong concurrent validity with the Statistics Attitude Survey. The norm group consisted of 162 university students enrolled in an introductory educational statistics course.

The third independent variable, social support, was measured by the Multidimensional Scale of Perceived Social Support (MSPSS) (Zimet, Powell, Farley, Werkman, & Berkoff, 1990). The instrument has 12 items and utilized a 7-point Likert scale ranging from very strongly disagree to very strongly agree. A total score and three subscale scores (support from significant others, support from family, and support from friends) were obtained. The instrument was reported as reliable with Cronbach’s alpha coefficients reported as .85 to .91 for the three subscales. Test-retest values ranges from .72 to .85. Zimet et al. reported significant correlations between the MSPSS subscales and the Depression and Anxiety subscales of the Hopkins Symptom Checklist as evidence of construct validity for their instrument. The norm group consisted for 275 university students at Duke University.

The Professional Counselor Volume 1, Issue 2

103

Data Analysis A simultaneous multiple regression was analyzed to determine the extent of the relationship between graduate students’ self-efficacy to learn statistics and statistics anxiety, attitude towards statistics, and social support. Alpha level was set at .05 for the analysis and semipartial correlation coefficients were assessed for practical significance. The multiple regression was repeated, removing social support from the analysis to explore any moderating effects of social support on the model.

Results Descriptive statistics of the sample data are displayed in Table 1 and sample scores for the assessments with a comparison to the maximum and minimum scores for the instruments are included in Table 2. Self-efficacy to learn statistics scores were normally distributed (SW(173) = .986, p = .076) and the box plot for the criterion variable confirmed normality as well. Standardized residuals also were normally distributed (SW(173) = .988, p = .159) and the box plot for the standardized residuals and scatterplots confirmed normality of the error variance or homoscedasticity. Scatterplots were analyzed for linearity, and it was determined no curvilinear relationships between the criterion variable and predictor variables were evident. Statistics anxiety and attitude towards statistics were highly correlated (-0.83), indicating multicollinearity.

Table 1 Descriptive Statistics, Predictor Variable Correlations, Multiple Regression Results

Self-Efficacy 49.73 18.97 1 -0.679 0.708 -0.023 – – – – –

Statistics Anxiety 119.95 35.83 – 1 -0.832 0.006 -0.15 0.051 -0.292 -3 0.003

Attitude Toward Stats 106.73 18.91 – – 1 0.017 0.467 0.098 0.466 4.785 0.001

Social Support 5.69 1.04 – – – 1 -0.53 0.981 0.981 -0.54 0.593

Table 2 Assessment Scores

Scale M SD Min Max Lowest Possible Highest Possible

Statistics Self-Efficacya 49.73 18.97 14 84 14 84

Statistics Anxietyb 119.95 35.83 56 201 51 255 Worth of Statisticsc 32.2 12.68 16 75 16 80 Interpretation Anxietyc 26.99 9.31 11 53 11 55 Test and Class Anxietyb 24.48 8.12 8 40 8 40 Computation Self-Conceptc 15.42 6.32 7 30 7 35 Fear of Asking for Helpb 9.81 3.35 4 20 4 20 Fear of Statistics Teachersb 11.05 4.42 5 25 5 25

Attitude Toward Statisticsd 106.73 18.91 38 143 29 145 Attitude Toward the Fieldd 78.4 11.96 29 100 20 100 Attitude Toward the Classd 28.33 8.79 9 45 9 45

M SD Self- Efficacy

Statistics Anxiety

Attitude Toward Stats

Social Support B SE Beta t p

104

The Professional Counselor Volume 1, Issue 2

Perceived Social Supporte 5.69 1.04 1 7 1 7 Significant Other Supporte 5.83 1.14 1 7 1 7 Family Supporte 5.41 1.48 1 7 1 7 Friend Supporte 5.67 1.03 1 7 1 7

aLower scores indicate low self-efficacy. bLower scores indicate less reported anxiety. cLower scores indicate perceiving statistics as less useful. dLower scores indicate negative attitude. eLower scores indicate less support.

There was a statistically significant relationship between self-efficacy to learn statistics and statistics anxiety, attitude towards statistics, and social support: F(3, 162) = 60.489, p < .001. A moderate effect size was noted with 52.8% of the variance accounted for in the model, R2 = .528. Statistics anxiety and attitude towards statistics were statistically significant predictors of self-efficacy to learn statistics and accounted for 3% and 7% of the variance, respectively. Social support was not a statistically significant predictor of self-efficacy to learn statistics and accounted for .1% of the variance. When social support was removed from the analysis, there was no change in statistical or practical significance.

Discussion

This study sought to explore the relationships of graduate students’ self-efficacy to learn statistics, statistical anxiety, attitudes towards statistics, and social support. The scores from the various instruments identifying each of the aforementioned variables produced both negative and positive correlations among each other. A statistically significant relationship was found among self-efficacy and statistical anxiety, attitudes towards statistics and social support indicating the importance of the graduate students’ belief in their competence of facing the challenges of learning statistics. However, there was no change in the relationship when social support was removed from the analysis; thus, it was not a contributing variable. Statistics self-efficacy scores from participants indicated moderate responses which mirrored the prior studies involving undergraduate students (Pajares, 1996; Zimmerman, 2000). As this was the first study that investigated graduate students, these results create a path for future research.

There was a negative correlation between self-efficacy to learn statistics and statistical anxiety of the graduate students. The negative correlation is consistent with Onwuegbuzie’s (2000) findings. Participants reported the lowest responses in the Fear of Asking for Help and Worth of Statistics subscales, signaling graduate students reluctance for asking for assistance from the professor and peers as well as a low belief in the applicability and purpose of statistics. Overall, these results and the negative correlation between self-efficacy and anxiety seem to depict a kind of self-fulfilling prophecy that graduate students assume when faced with taking statistics which is similar to Perney and Ravid’s (1991) report.

A positive correlation was found between self-efficacy to learn statistics and attitudes towards statistics. This results indicated that the better the attitude of the graduate students towards statistics, the higher self-efficacy beliefs to learn the subject. Results indicated a more moderate response to attitudes not found in other studies where students were coming in with a negative attitude or were developing negative attitudes towards the end of the course (Gal & Gingsburg, 1994). It may be considered that graduate students in this study were neutral in their attitudes towards learning statistics without extreme reactions.

Participants reported a high level of social support, which indicates that most of the graduate students believed they had adequate support. The sample perceived social support as an influential factor in their lives, which is similar to most college student population reports (Solberg & Villarreal, 1997). However, social support was not a statistically significant predictor of self-efficacy to learn statistics. Also, when this variable was removed from the multiple regression analysis, there was no statistical or practical change in the regression. The insignificant result implies that social support was present for students, but it did not interact as a buffer between variables and possibly decrease anxiety or increase positive attitudes as indicated by Bonilla (1997), Cohen and McKay (1984), and Solberg and Villarreal (1997). Thus, social support may possibly help one cope but not necessarily remove the problem, change attitudes, or change thinking.

The Professional Counselor Volume 1, Issue 2

105

Multicollinearity be