Assignment 5 2005-2006: Statistical Inference

Want to print this assignment? Click here

Date Assigned: November 15/17
Title of Assignment: Statistical Inference
Due Date: November 22/24
Points: 10
 

This exercise has several purposes. First, it will give you an opportunity to work with the idea of statistical significance. Second, it will offer you a hands-on experience with three widely used procedures for testing significance. Third, it will allow you to explore differences using either your index developed for the last assignment or a new index you may prefer. The procedures to be used are the T-Test, the One-Way ANOVA and Chi-square. The T-Test procedure is used to examine differences in the mean scores on a dependent variable with many categories, such as your un-recoded index,  for two groups, such as men and women. One Way Analysis of Variance tests for significant differences across more than two groups, also on dependent variables with numerous categories. Chi-square tests for significant differences in crosstabulations. All three procedures enable you to see whether groups differ "significantly".

For the assignment:

  1. Consider whether you expect gender, age and some other difference rooted in your previous work  to appear in working with your un-recoded index (either demographic or non demographic variables may be used).
  2. Identify indicators for each of the three independent variables and write hypotheses regarding  gender, age and  your third independent variable of theoretical  interest in terms of their affect on your index.
  3. Recode age into three meaningful categories. Recode the third independent variable into an appropriate number of categories.
  4. Test the gender hypothesis using T-Test, the age hypothesis using One-Way Anova and the third hypothesis using the more appropriate of the two techniques.
  5. Recode your index into 2 categories (high versus low scores) and crosstabulate it with gender, age and the third variable (also recoded into about three categories or so). In addition to calculating the appropriate measures of association, test each relationship for statistical significance using Chi-square. Recode your index once more, this time into three categories. Crosstabulate your recoded index with your three independent variables and test again for statistically significant differences.
  6. Where you find significant Chi-square differences in crosstabs with more than two columns, run an ANOVA to determine more precisely where the significant differences lie.

In conducting all your analyses be sure to eliminate missing values for all variables you use. Otherwise your results may be inaccurate due to the codes for missing values being  included in the calculation of group means.

When you have "clean" output, print out your results. Briefly discuss whether gender, age and the third hypothesized gap appears. Then put the results of your analyses into words with particular reference to whether any observed differences are due to chance . Finally, you should comment on the relevance of statistical significance testing for your work.


Assignments