Advanced Methods of Social Research Assignment -
This assignment is based on the ideas for research projects included in Appendix E of the course textbook (Healey, Joseph F. 2015. Statistics: A Tool for Social Research. Stamford: Cengage Learning. 10th edition).
Exercise -
Select variables that are available in the 2004, 2010, and 2016 General Social Survey (GSS) to estimate characteristics of the U.S. adult population. You will use Stata to estimate and analyze sample statistics and confidence intervals. These variables can be the same as those used in the classroom or in the previous assignment. See the GSS codebook or the GSS Data Explorer website for a list of variables available in GSS.
A. Estimating means
Include at least 2 tables (1 for each variable).
1. There are relatively few interval-ratio variables in GSS, and for this part of the project you may use ordinal variables that have at least three categories or scores. Choose two variables that fit this description.
2. Estimate means, standard errors, sample size, and construct 95% confidence intervals for the mean of each of your variables for each year. When computing the standard error, consider the effect of clustering and stratification, as well as the effect of sampling weights (i.e. complex survey design), using the "svyset" command. The command "svy: mean" provides an estimate of the population mean and an estimate of its standard error. For example, svy: mean conrinc.
3. For each variable, report and explain the results, the confidence interval, the confidence level, and the sample size.
4. Include in your analysis a brief explanation of the role of these concepts and terms in the estimation: sample, population, statistic, parameter, equal probability of selection method (EPSEM), representative, and confidence level. This can be part of the overall data and methods paragraph.
B. Estimating proportions
Include at least 2 tables (1 for each variable).
1. Choose two nominal- or ordinal-level variables. These variables should be different than the variables used for estimating means.
2. Estimate proportions, standard errors, sample size, and construct 95% confidence intervals for the proportions of each category of your variables for each year. When computing the standard error, consider the effect of clustering and stratification, as well as the effect of sampling weights (i.e. complex survey design), using the "svyset" command. The command "svy: prop" provides an estimate of the population proportion and an estimate of its standard error. For example, svy: prop letin1.
3. For each variable, report and explain the results, the confidence interval, the confidence level, and the sample size.
4. Include in your analysis a brief explanation of the role of these concepts and terms in the estimation: sample, population, statistic, parameter, equal probability of selection method (EPSEM), representative, and confidence level. This can be part of the overall data and methods paragraph.
C. Two-sample t-test
Include at least 2 tables (1 for each dependent variable).
1. Use the two variables from part A of this assignment (interval-ratio or ordinal-level variables that have three or more scores). They will be the dependent variables.
2. Choose independent variables that might logically be a cause of your dependent variables. These independent variables should have only two categories. You can still use independent variables with more than two categories by collapsing their scores with the "generate" and "replace" commands in Stata. Independent variables can be at any level of measurement. You may use the same independent variable for both dependent variables.
3. Estimate two-sample t-tests with equal variances for each of your dependent variables for each year. You do not have to consider the complex survey design or the sampling weight for these estimations. Use the command "ttest" with a specific dependent variable and the option "by" to indicate the independent variable. For example, ttest conrinc, by(sex).
4. Report and explain the results of the tests. At a minimum, your analysis should clearly identify the independent and dependent variables, the sample statistics, the value of the test statistic, the results of the test, and the confidence level.
D. Analysis of variance
Include at least 2 tables (1 for each dependent variable).
1. Use the two variables from part A of this assignment (interval-ratio or ordinal-level variables that have three or more scores). They will be the dependent variables.
2. Choose independent variables that might logically be a cause of your dependent variables and that have between three and five categories. You may use the same independent variable for both dependent variables.
3. Estimate one-way analysis of variance for each of your dependent variables for each year. You should consider the effect of sampling weights, using the "aweight" command. The "oneway" command reports one-way analysis-of-variance (ANOVA) models. For example, oneway conrinc raceeth [aweight=wtssall].
4. Report and explain the results of the tests. At a minimum, your analysis should clearly identify the independent and dependent variables, the sample statistics, the value of the test statistic, the results of the test, the degrees of freedom, and the confidence level.
E. Chi square
Include at least 2 tables (1 for each dependent variable).
1. Use the two variables from part B of this assignment (nominal- or ordinal-level variables). They will be the dependent variables.
2. Choose independent variables that might logically be a cause of your dependent variables. Independent variables can be at any level of measurement as long as they have five or fewer (preferably two or three) categories. Output will be easier to analyze if you use variables with fewer categories. You may use the same independent variable for both dependent variables.
3. Estimate the chi square tests for each of your dependent variables for each year. You do not have to consider the complex survey design or the sampling weight for these estimations. Use the command "tab" with the option "chi" to indicate the Pearson's chi-square test. It is almost always desirable to report the column percentages as well. The dependent variable should have the categories listed on the rows and the independent variable on the columns. For example, tab letin1 sex, chi col.
4. Report and explain the results of the tests. At a minimum, your analysis should clearly identify the independent and dependent variables, the sample statistics, the value of the test statistic, the results of the test, the degrees of freedom, and the confidence level.
Attachment:- Assignment Files.rar