Instructional Technology
Conference 2005
Proposal #47
Title: Do gender and socioeconomic status matter?
Name: Jiye Ai
Audience Level: All
Audience: General
Length: 1 Hour
Abstract:
According to some recent papers, the student°Øs performance may have certain relationship with student°Øs social characters. The goal of this paper is to investigate whether there is a difference between reading scores (and math score) among students with different gender and ses (socioeconomic status). The different statistical methods are applied to analyze the dataset of an elementary school.
Description:
Introduction
The sample size is 200 including 91 female and 109 male. For the ses, 47 out of 200 are in low class, 95 out of 200 are in middle class and 58 are in high class.
Description of my data
The gender and ses are 2 categorical variables. They are both independent variables (IV). The reading score and math score are 2 continuous variables. The reading score is a dependent variable (DV) and the math score is a covariate.
According these results, the reading score and the math sore are meeting the normality linearity assumptions.
Conduct my ANOVA test
The hypothesis of the homogeneity variance assumption is as below:
Ho: there is no significant difference across the groups of reading scores divide by gender and ses.
Conduct my planned contrast and post hoc test
Planned contrast
The ses is selected to do planned contrast due to its 3 levels. There is statistically significant difference between the mean of reading score of level 1 and the pooled mean of level 2 and 3 in ses.
Post hoc
Tukey, Scheffe and Bonferroni are conducted in this project. For the Sig value, Tukey is the best in all three post hoc tests. there is no a significant difference between the °∞low°± and °∞middle°± pairwise comparison.
Perform ANCOVA
Homogeneity of regression test
Assumption: there is no significant difference in the slope between the levels of gender/ses.
ANCOVA test
We can find only the main effect for ses is not significant. Others are significant. Thus, math score is an effective covariate that makes main effects and interaction more salient.
Conclusions
1. ANOVA test
There is no statistically significant difference between the mean of reading score of 2 levels of gender. There is statistically significant difference between the mean of reading score of 3 levels of ses. There is no statistically significant difference between the mean of reading score of 2 levels of gender across levels in ses and of different levels of ses across levels in gender.
2. Planned contrast
There is no statistically significant difference between the mean of reading score of level 1 and level 2 in ses. There is statistically significant difference between the mean of reading score of level 1 and the pooled mean of level 2 and 3 in ses.
3. Post hoc
There is no a significant difference between the °∞low°± and °∞middle°± pairwise comparison. But for the remaining pairs, there is significant difference between them.
4. ANCOVA
There is no statistically significant difference between the mean of 3 levels of ses after controlling the effect of math score.
There is no statistically significant difference between the mean of 2 levels of gender after controlling the effect of math score.
There is no statistically significant difference between the mean of 2 levels of gender across 3 levels in ses and the mean of 3 levels of ses across 2 levels in gender after controlling the effect of math score.
Session Type: Lecture / Presentation
Contact information/affiliation:
Jiye Ai, University of Missouri-Columbia
111 London Hall Columbia, MO 65211 USA
projectwhistlestop@yahoo.com (573)882-2524
Equipment: computer and projector