not complete the parent questionnaire will be examined to determine non-response biases. Categories of parent involvement will be collapsed into parent involvement--low, moderate, and high groups. Both the Parent questionnaire and the Student questionnaire will be completed by marking one of six ordinal categories: false, mostly false, more false than true, more true than false, mostly true, and true. Demographic variables are extremely vital. Because gender and SES are associated with family variables and educational expectations, I would include these variables in my analysis. SES in particular has been reported as having a strong influence on educational expectations (e.g., Wilson & Wilson, 1992). Design This study will be a multifactoral design and will have both repeated measures and within subjects. Educational achievement, the dependent variable, a non-manipulated independent variable would be parental involvement. A chi-square would be used as well as a cluster analysis and multiple regression Because of the large sample size subgroups would be developed. A minimum subgroup (sample segment) size will be set at n = 100 after categories of a variable are merged, and at n = 200 before categories were merged. No group of less than 200 would be split, and no subgroup smaller than 100 would be created. Expected Results There are many expected results for this study. It appears there will be external and internal validity. Correlations will be consistently positive. That is, a relatively low amount of parental involvement will relate to a relatively low amount of academic achievement and a relatively high amount of parental involvement will relate to a relatively high amount of academic achievement. Also, adolescents with higher parent involvement will also be higher in their educational expectations. SES will correlate most highly with educational expectations. Gender will only weakly rel...
