The Role of Affective and Motivational Factors in Statistics Performance in University Students

When faced with the prospect of having to complete a statistics course at university, students either fall into a state of anxiety about failing the course or they form a belief that they will do well. This is due to their perception of self-efficacy. Perceived self-efficacy is, as stated by Bandura (1994) & Pajares (2002), the construct of a person's beliefs in their ability to perform in certain tasks including academic tasks. It has been shown that people who were confident in their abilities to achieve well were found to be more successful in their achievement level which in this case was achievement in a statistics course (Schutz, Drogosz, White, & DiStefano, 1998). For self-efficacy to be effective there must be more than a perceived result but also a perceived ability to master the particular task faced. While self-efficacy is formed by previous experiences of similar tasks and their respective results, it is also affected by anxiety and stress and thus lowers one's perceived ability to achieve a certain outcome (Bandalos, Yates, & Thorndlike-Christ 1995).

Past research has presented evidence to show that there are strong negative correlations between test anxiety and self efficacy (Bandalos et al., 1995; Bandalos, Finney, & Geske 2003; Schutz et al., 1998). Bandalos et al. (1995) also found that test anxiety negatively correlated with statistics achievement. On the contrary, Bandalos et al. (2003) found that there wasn't a relationship between test anxiety and statistics achievement. Gal & Ginsburg (1994) present evidence that shows that affective states like test anxiety cause a loss of confidence in the students and have a tendency to pull them into a "cycle of failure" where they continually fail because they have a belief that they can't succeed. Bandalos et al. (2003) found that statistics achievement had a relationship with past achievement and self efficacy.

This current study aimed to examine how self-efficacy, test anxiety and past achievement affected the expected outcomes of second year university students in a statistics course. It was hypothesised that outcome expectancy for performance on the statistics mid-semester exam would be affected by statistics self-efficacy, statistics test anxiety and past achievement.

Method

Participants

This study consisted of 161 undergraduate psychology students undertaking a second year statistics course. They were asked to complete a series of questions relating to their beliefs and expectations of statistics. One of the questions asked if they were repeating the course or not. Those who were repeating were excluded in the overall study as they represented a different population and could thus hinder the validity of the results. Due to this there were a resulting 143 participants when the people repeating were removed from the study. This group consisted of 118 females and 25 males with the age ranging from 18 to 55 with a mean age of 23.45 years (SD = 7.92).

Materials

A questionnaire was used to collect all the data for this study. The first section consisted of basic demographics which were age, gender and whether or not the participant was repeating the course. A question also asked to give an expected outcome for the course defined in terms of expected percentage for the future mid-semester exam. The next section consisted of 10 items with each responded to on a 7-point Likert scale from 1 strongly disagree to 7 strongly agree. These questions formed a statistics self-efficacy score as on Statistics Self-Efficacy Scale from Bandalos et al. (2003). Examples of these items were "I'm certain I will be able to understand the ideas taught in this research methods and statistics class" and "Compared with other students in this research methods and statistics class, I expect to do well". The reliability of the items was tested and a Cronbach's alpha of 0.94 was obtained. Participants were categorised into two groups with the use of a median split. The two groups were low SSE (statistics self-efficacy) (coded as 1) and high SSE (coded as 2).

The following 20 items were adapted from the Test Anxiety Inventory (TAI; Speilberger, 1980) and measured a participant's Statistical Test Anxiety. The adaptation consisted of changing the items to specifically refer to test anxiety on statistics exams. For example, "I feel confident and relaxed while taking research methods and statistics tests" (reverse scored) and "During research methods and statistics tests I am so tense that my stomach gets upset". Each of the items or statements were responded to on a 7-point Likert scale as with the SE scale. The reliability of the items was found to be a Cronbach's alpha of 0.95. A median split was again used to assign the participants into two groups. The first being low statistics test anxiety (coded as 1) and second being high statistics test anxiety (coded as 2).

Each participant has their previous Quantitative Analysis 1 results were obtained and formed into two groups of fail, pass conceded and pass (coded as 1) and credit, distinction and high distinction (coded as 2).

Procedure

The participants were asked to fill out the questionnaire during their tutorial class for the statistics course. They were also given the opportunity to not complete the questionnaire if they so wished.

Design

A within-subjects experimental design was used to determine whether self-efficacy, test anxiety and past statistics achievement had an effect on expected statistics outcome. There were three independent variables; statistics self-efficacy, test anxiety and past achievement. A median split was performed on self-efficacy and test anxiety to separate the participants into low and high levels of each IV. Past achievement was separated into low and high with the low achievers receiving a mark of fail, pass conceded or pass in QA1 with the high achievers receiving credit, distinction or high distinction. Each IV was separately tested on the dependent variable, expected outcome in the statistics course. A questionnaire was used to determine each participant's score for the three independent variables and also to find out their expected percentage on the mid semester exam.

Results

Each analysis was conducted with an independent groups t-test with an alpha level of .05. Normality and homogeneity of variance assumptions were met. A significant effect of self-efficacy on expected outcome was found, t (128) = -4.079, p < .0001, d = -0.72. The high self-efficacy group reported a significantly higher expected outcome percentage (CI0.95: 75.38%< &#956;