You are on page 1of 4

Total: 28/30 Results 1) Question: Does teacher and peer math support predict students level of confidence to do math?

Multiple Regression In checking assumptions for the regression model it was shown that it has a normal distribution and has no systematic pattern or clustering of the residuals which would suggest model violation. Pearson correlations were computed to determine the relationship between subjects confidence to do math, the level of teacher support they receive, and the level of peer support they receive. Pearson correlations revealed positive correlation for the following variables: 1) confidence to do math and peer support (r = .141, p< .05); and 2) teacher math support and peer support (r = .290, p< .001). A multiple linear regression was completed to determine if teacher support and peer support accounted for a significant portion of variance in students confidence in their math ability. This will help determine if peer and teacher support predicts confidence in math ability. The standardized regression revealed a multiple correlation of .171 and R Square revealed that of the total variance in confidence in ability to do math, 29% was accounted for by teacher support and peer support. This is considered significant, F (2, 247) = 3.740, p<.05, therefore, rejecting the null hypothesis. However, standard linear regression indicated that only one independent variable (peer support) accounted for a significant portion of the variance (peer support t (249) = 2.063, p < .05; teacher support t (249) = 1.56, p > .05. Beta scores indicated that peer support accounts for significantly more of the variance than teacher support (peer = 17% and teacher 10.2%).

2) Question: Do those from different socioeconomic backgrounds differ in the amount they spend doing homework, the amount of teacher support they receive and their confidence in their math ability? MANOVA An ad hoc Helmert contrast was run to compare the means between each of the groups. Level 1 versus Later shows the comparison of the mean of group 1 to the means of group 2 and 3. Results indicated significant results for confidence in math ability (3.137, p < .05) and hours of math homework completed in a month (.932, p < .05); however, teacher support was found to be insignificant (-.930, p > .181). Level 2 versus Level 3 compares the means of group 2 and 3and indicated non-significance on all variables (confidence in math ability -.574, p = .737; hours of math homework in one month .178, p = .701; teacher support 1.348, p = .104). These results indicate that subjects from different SES backgrounds (grouping variable) differ on confidence in their math ability and the hours spent completing homework in one month. The Boxs M value was 18.142 indicating that the covariance matrices are equal (Boxs M = F (12, 26475.41) = 1.485, p = .121. Barletts test of sphericity tests the null hypothesis that the residual covariance matrix is proportional to an identity matrix. The test revealed significant results indicating that a multivariate analysis can be completed (Bartletts test (5) = 432.98, p < .001). Levenes test indicated non-significant result for the dependent variables, showing that the error variances are equal across the groups (confidence in math ability F (2, 247) = .728, p = .484; hours of math homework F (2, 247) = 1.331, p = .266; teacher support F (2, 247) = .849, p = .429). The SSCP Matrix indicated that the dependent variables have a weak correlation (ranging from -.023 to.417).

Wilks Lamda showed that there was a main effect of group on the dependent variables all together (F (6, 490) = 2.157, p = .046). This value is less than the significance value of .05, therefore it can be assumed that there are differences across the grouping variable (SES). The tests of subject effects identifies any main effect of group on each of the dependent variables. Results showed non-significant results (confidence in math ability F (2, 247) = 2.524, p = .082; hours of math homework F (2, 247) = 2.893, p =.057; teacher support F (2, 247) =2.388, p = .094). This indicates that there are no significant differences between the grouping variables (SES). Post hoc tests are used to identify which groups differed from one another on each of the dependent variables. The results of post hoc analysis were non-significant indicating that there are very little differences among those from different SES backgrounds on the dependent variables. Furthermore, homogeneous subsets are determined based on group differences and the Tukey test was used to assess the significance of those subsets. Results from this showed that subsets were found to be non-significant (confidence in math ability p = .105; hours completing math homework p = .067; teacher support p = .121). Discriminant Analysis The Boxs M value was non-significant (Boxs M = F (12, 269475.41) = 1.485, p = .121). This indicates homogeneity of covariance across groups. The tests of equality of group means indicates that there are no significant group differences across the variables (confidence in math ability F (2, 247) = 2.524, p = .082; hours of math homework in a month F (2, 247) = 2.893, p = .057; teacher support F ( 2, 247) = 2.388, p = . 094).

Only function 1 of the two discriminant functions revealed significance (CHI Square (6) = 12.823, p = .046). The eigen value was .043, thus accounting for 80.8% of the variance. However, function 1 only accounted for 4% of the variance. Standardized coefficients and structure matrices tables indicated that confidence in math ability and hours of math homework per month were moderate predictors of group membership, whereas teacher support was non-significant. It should be noted that confidence in math ability was more likely to predict group membership when compared to hours of math homework and was considered the most significant influence between groups.

You might also like