Abstract
Abstract: Some factors that the researcher believed to affect the decline in academic achievement were studied on students' performance in statistics courses at the faculty of Education Janzour, University of Tripoli. For the purpose of studying the impact of various factors on academic success and failure in statistics courses. A sample of 111 female students was studied. Data was collected on their performance in the following courses: Basic Statistics, Introduction to Statistics, and Probability, from the departments of Mathematics, Physics, and Chemistry. The factors analyzed included pass/fail status, final grades in the courses, department, course type, library study habits, number of attended lectures, and number of missed lectures. The analysis revealed that the department had no significant impact on achievement, with a Chi-Square p-value of 0.312. However, the subject had a significant effect on student performance, with a Chi-Square p-value less than 0.05. Library use was not a significant predictor of achievement, as indicated by a Chi-Square p-value of 0.545. Conversely, lecture absences had a significant negative impact on performance, with a Chi-Square p-value of 0.006. Lecture attendance showed a significant relationship with achievement, supported by a Gamma test p-value of 0.025. Pearson correlation revealed a positive link between lecture attendance and final exam grades (0.409), indicating that more attendance correlates with higher grades. Conversely, missed lectures had a negative relationship with final exam grades (-0.556), meaning more absences are linked to lower grades. Cronbach's Alpha for the measurement was 0.705, indicating acceptable reliability. Removing "Study in library" significantly increased the Alpha to 0.913, suggesting it is crucial for reliability. In contrast, removing "The department" or "The subject" did not improve the tool's reliability. However, the binary logistic regression analysis handled the impact of variables slightly differently, in accordance with the rules specific to regression analysis. In the classification results, the model failed to correctly predict any "Fail" cases (0.0% accuracy) but accurately identified all "Pass" cases (100.0% accuracy), resulting in an overall accuracy of 77.5%. At Step 2, the model showed a -2 Log likelihood of 11.88, with Cox & Snell R² at 0.617 and Nagelkerke R² at 0.941, indicating a good fit with varying degrees of explained variance. The Hosmer and Lemeshow test revealed generally minor discrepancies between observed and expected values, suggesting a good model fit. At Step 1, the logistic regression model achieved high accuracy, predicting 96.0% of failures and 98.8% of successes, leading to an overall accuracy of 98.2%. *The logistic regression analysis reveals that all variables are not statistically significant at the 0.05 level. The coefficients and p-values are as follows: Depart(1) (-36.920, p=0.077), Depart(2) (-2.597, p=0.562), Subject(1) (4.590, p=0.424), stu_lib(1) (2.395, p=0.407), Pres (3.033, p=0.062), Abs (3.839, p=0.099), Exam (0.545, p=0.068), and the constant (-56.966, p=0.079). The logistic regression equation is: Logit(p)= -56.966-(36.920×Depart(1))-(2.597×Depart(2))+(4.590×Subject(1))+(2.395×stu_lib(1))+(3.033×Pres)+(3.839×Abs)+(0.545×Exam) This equation describes the relationship between the variables and the outcome, allowing for the assessment of the impact of each variable on the probability of Students' achievement as (Success and Failure). Keywords: Student Performance, Factors Impact, Lecture Attendance, Library Study Habits. Binary Logistic Regression