Comparing Means between Groups using T Tests

Overview:

This assignment uses the data file GSS2004Intro.sav , a subset of variables from the General Social Survey (GSS) 2004, conducted by the National Opinion Research Center at the University of Chicago. The GSS is a survey involving demographic, attitudinal, and behavioral items that include views on government and satisfaction with various facets of social life.

In these exercises, we want to see whether men and women differ in the average number of children, their use of email, and their age.

Exercises and Questions:

1. Run an exploratory data analysis on the variables CHILDS (number of children), EMAILHR (number of hours spent on email per week), and AGE by GENDER (hint: don’t forget to select the "Pairwise Deletion" option). Are any of the variables normally distributed? What differences between males and females do you notice from the boxplot of CHILDS?
2. Use Chart Builder to produce a paneled histogram for CHILDS by GENDER.
3. Run t tests looking at mean difference by gender for the three variables CHILDS, AGE, and EMAILHR. Interpret the results. Which variables met the homogeneity of variance assumption? Are the means of any of the variables significantly different for males and females? Can you explain why?
4. Use Chart Builder to display an error bar chart of number of children by gender. The analysis will suggest that women have a greater number of children than men. Can you suggest reasons for this seemingly odd result?

Correlation: Understanding Linear Association between Variables

Overview:

The first half of this assignment uses the data file Bank.sav , a personnel file containing salary, demographic and work experience and compensation-related data from 474 employees of a bank in the early 1970s. The compensation data has not been converted to current dollars.

The second half of the assignment uses the data file GSS2004Intro.sav , a subset of variables from the General Social Survey (GSS) 2004, conducted by the National Opinion Research Center at the University of Chicago. The GSS is a survey of U.S. citizens involving demographic, attitudinal, and behavioral items that include views on government and satisfaction with various facets of social life.

Exercises and Questions:

1. In regards to the data file Bank.sav , suppose you are interested in predicting current salary (SALNOW), based on age (AGE), education in years (EDLEVEL), minority status (MINORITY), beginning salary (SALBEG), gender (SEX), and work experience before coming to the bank (WORK).
   1. Run frequencies on MINORITY so you understand its distribution. Run descriptive statistics on the other variables for the same reason.
   2. Produce correlations with each of these variables and current salary (SALNOW).
   3. Create scatterplots of the variables and current salary. Which variables have strong correlations with current salary? Did you find any potential problems with using linear regression? Did you find any potential outliers?
2. Looking now at the dataset GSS2004Intro.sav , suppose you are interested in predicting parents’ age when their first child is born. This information may be of interest, for example, if you were looking at programs targeted toward teenage parents. The outcome variable is AGEKDBRN. Consider AGE, education (EDUC), spouse’s education (SPEDUC), household size (HHSIZE), number of children (CHILDS), and SEX (a numerically-coded version of the variable GENDER). Note that by considering SPEDUC, you are limiting the analysis to those parents who are currently married.
   1. Create a numeric version of GENDER using the "Recode into Different Variable" procedure. Name the new variable SEX and recode "M" to 0 and "F" to 1. Assign value labels to the new variable.
   2. Run frequencies on SEX so you understand its distribution. Run descriptive statistics on the other variables for the same reason.
   3. Product correlation coefficients for all variables with AGEKDBRN.
   4. Create scatterplots of the variables and AGEKDBRN. Which variables have strong correlations with AGEKDBRN? Do you find any potential problems with using linear regression? Did you find any potential outliers?

Comparing Means between Groups Using One-Factor ANOVA

Overview:

This assignment uses the data file GSS2004Intro.sav , a subset of variables from the General Social Survey (GSS) 2004, conducted by the National Opinion Research Center at the University of Chicago. The GSS is a survey involving demographic, attitudinal, and behavioral items that include views on government and satisfaction with various facets of social life.

In this set of exercises, we will investigate the number of hours spent watching TV (TVHOURS) versus the number of hours using the Web (WWWHR). We want to see whether the average number of hours spent watching TV or the number of hours spent using the Web differ by marital status (MARITAL).

Exercises and Questions:

1. Run an exploratory data analysis on TVHOURS and WWHR by MARITAL (don’t forget to select the "Pairwise Deletion" option).

1. Are any of the variables normally distributed?
2. What differences do you notice in the means and standard deviations for each group?
3. Is the homogeneity of variance assumption likely to be met?

2. Run One-way ANOVA looking at the mean difference by marital status for these two variables. Request the test for homogeneity and the robust analysis of variance. Interpret the results.

1. Which variables met the homogeneity of variance assumption?
2. Are the means of any of the variables significantly different for marital status groups?

3. From the One-way ANOVA dialog box, click on "Post-Hoc Tests" and select a liberal test such as LSD, or a more conservative test such as Scheffe or the Games-Howell if the variables did not meet the homogeneity of variance criteria.

1. Which groups are significantly different from each other?
2. Do the tests agree? If not, how might you summarize the results?

4. Use Chart Builder to display an error bar chart for each of these analyses.

.

Using the Chi-Square Test of Independence to Compare Categorical Variables

Overview:

This assignment uses the data file GSS2004Intro.sav , a subset of variables from the General Social Survey (GSS) 2004, conducted by the National Opinion Research Center at the University of Chicago. The GSS is a survey involving demographic, attitudinal, and behavioral items that include views on government and satisfaction with various facets of social life.

In the following exercises, we want to study the relationship of two demographic variables, race (RACE) and gender (GENDER), with three behavior variables: HLTH1 (whether you were ill enough to go to the doctor last year), NATAID (your attitude toward spending on foreign aid), and NEWS (how frequently you read the newspaper).

Exercises and Questions:

1. Before running the analysis, think about the variables involved in these tables. What relations would you expect to find here and why?

2. Run cross tabulations of race (RACE) against the variables HLTH1, NATAID, and NEWS. Request the appropriate percentage within racial categories and run the chi-square test of independence.

3. Repeat the analysis after substituting GENDER in place of RACE. How would you summarize each finding in a paragraph?

4. Run a three-way table of HLTH1 by GENDER by RACE and of NATAID by RACE by GENDER. Request the chi-square test of independence. Do these findings affect your summaries from above? If so, how?

5. Now, request appropriate measures of association for a HLTH1 by GENDER table and a NEWS by RACE by GENDER table. Are the results consistent with your interpretation up to this point? Based on either the association measures, or percentage differences, would you say the results have practical (or ecological) significance?

6. Create a clustered bar chart displaying the results of one of your tables.

Price: $49.99

Solution: The downloadable solution consists of 40 pages, 2400 words and 34 charts.
Deliverable: Word Document