Predicting physics achievement: attitude towards physics, self-efficacy of learning physics, and mathematics achievement

Sample

A total of 301 Turkish high school students (male=143, female=158) participated in this study. Their grades were ranged from 9 to 11. Data were collected from three different high schools in one of the cities that located in Eastern region of Turkey. Purposive sampling was used (Fraenkel & Wallen, 2005). The reason of this is to reach a variety of students with different characteristics. In each school only one classroom for each grade was randomly chosen. While also choosing the schools to be included in the study, students’ achievement scores in nationwide exams TEOGs (i.e., the exams that were applied to the students during their elementary education) were taken into consideration. Students are placed in high schools according to their scores in these exams in Turkey. For example, the students who take high scores in the exams have a right to enroll in the schools that composed of mostly high-achieving students. Similarly, the students who take low scores in the exams have a right to enroll in the schools that composed of mostly low-achieving students. Examining cut off scores of the schools in the city center according to the TEOG exam results (TEOG Lise Taban Puanları 2014 2015 MEB, 2015) three schools composed of high-achieving (approximate 440 TEOG base scores), moderate-achieving (approximate 360 TEOG base scores) and low-achieving (approximate 300 TEOG base scores) students were chosen. By this way, the students with various achievement levels were reached.

Data Collection

A cross-sectional survey was used in data collection (Fraenkel & Wallen, 2005). Data were collected through a questionnaire. In it students were first required to respond the questions about their demographic information (gender, grade level), and mathematics and physics grade point average (GPA) scores on a 100 point-scale. These GPA scores were considered as the students’ mathematics and physics achievement scores in this study. In addition, the eight-item scale self-efficacy of learning that is one of the dimensions of Motivated Learning Strategies Questionnaire (MSLQ) (Pintrich, Smith, Garcia & McKeachie, 1993) and the scale measuring Attitude towards Physics (ATP) with 30 items (Tekbıyık & Akdeniz, 2010) was included in the questionnaire.

MSLQ was adapted into Turkish by Büyüköztürk, Akgün, Özkahveci and Demirel (2004). Only the dimension self-efficacy of learning in this adapted version was used in this study. The eight-item dimension’s Cronbach’s alpha reliability coefficient was 0.86 (Büyüköztürk et al., 2004). This scale was a two-sided seven-point scale and the participants were required to indicate how much the items reflect their ideas (Pintrich et al., 1993). The scale measuring students’ self-efficacy of learning physics in this study was also modified from self-efficacy of learning dimension of MSLQ.

ATP scale with 30 items, ranging from 1 (strongly disagree) to 5 (strongly agree), had four dimensions that are importance (10 items), comprehension (7 items), requirement (7 items), and interest (6 items) (Tekbıyık & Akdeniz, 2010). This scale’s dimensions’ (importance, comprehension, requirement, and interest) Cronbach’s alpha reliability coefficients were 0.838, 0.795, 0.749, and 0.717, respectively and ATP scale’s overall alpha was 0.873 (Tekbıyık & Akdeniz, 2010). Consequently, a unique questionnaire that explores students’ demographic background, mathematics and physics achievements, self-efficacy of learning physics, and attitude towards physics were distributed to the students.

Data Analysis

Data was analyzed using hierarchical regression analysis to elicit the significant predictors of physics achievement. In the hierarchical regression analysis, the independent variables are entered into the regression model by the researcher according to an order. This order of entry of the variables into the hierarchical model could be determined according to the logical or theoretical considerations (Tabachnick & Fidell, 2007). In this study, psychological constructs attitude towards physics and self-efficacy of learning physics were first considered as more powerful predictors of physics achievement because these variables measure something about physics. Moreover, some theorists’ (e.g., Bandura, 1993; Fishbein & Ajzen, 1975) ideas about the strong relationship between behavior and psychological constructs influenced the determination of the order of entry of the variables in this study. For example, Fishbein and Ajzen (1975) tried to explain individuals’ behavioral intentions and behaviors considering attitudes, and Bandura (1993) indicated the strong relationships between individuals’ self-efficacy and performances. In this regard, first of all, students’ attitude towards physics and self-efficacy of learning physics were entered into the regression equation to predict their physics achievement. Then the mathematics achievement that was considered as having lesser priority to predict the physics achievement was forced into the equation. However, before starting to perform the regression analysis, whether the data of this study meet the assumptions of regression analysis was examined. At the beginning, three participants’ data detected as outliers were removed from the data; and therefore, the analysis was done with the remaining 298 participants’ data. At the same time validity and reliability of the scales were tested. Reliability of each scale was tested by calculating Cronbach’s alpha reliability coefficient. To test the construct validity of the scales confirmatory factor analysis (CFA) was carried out on AMOS program. Means and standard deviations of each variable were also calculated. Finally, the hierarchical regression analysis was performed