General Social Survey
In this assignment you will be examining some of the factors thought to influence trust in the USA, using survey data from the General Social Survey. You will be evaluated on your ability to carry out and understand the statistical procedures you are using and on your interpretation and presentation of the results.
The file gss2004.dta contains data from the 2004 United States General Social Survey, including almost 3000 respondents and about 1200 variables. Some relevant variables are mentioned in the “codebook” below. Further documentation for the GSS can be found at http://www3.norc.org/GSS+Website/Documentation/. In particular the codebook, which covers all waves from 1972-2010, is at http://publicdata.norc.org/GSS/DOCUMENTS/BOOK/GSS_Codebook.pdf. Alternatively, you may wish to look at the variable browser at http://www3.norc.org/GSS+Website/Browse+GSS+Variables/.
What you should hand in
A word-processed research report (in hard copy) of not more than 3000 words, including tables and charts, that addresses all the tasks below. All output – tables or graphics – from SPSS or STATA that you include in the report must be annotated, labelled clearly and integrated in the main text. Include the output that is necessary and not more. You will not get marks for including reams of output with no interpretation. Marks will be awarded for presentation of results. Please check that the online submission system does not adversely affect the formatting of your document.
You may include an appendix for tables that may be of supplementary interest but that are not essential. The word limit does not include material in the appendix.
It is not necessary to read any further material on the substantive topic than is included here. You are not expected to have any knowledge of the literature on social trust. The exercise merely uses the data as a vehicle for you to show your ability to carry out the statistical techniques learnt during the course.
1. Descriptive statistics
Before running statistical analyses, you should describe the data you are using. Briefly detail how it was generated, i.e. the GSS survey. Describe any variables you intend to use, and give appropriate descriptive statistics for them. Justify why these are the appropriate variables for your purposes.
2. Trust, community stability and demographics
One important theory of the determinants of social trust is that long-running membership of a small stable community increases trust, whereas members of larger, anonymous communities, or people with a shorter history of interaction with their community, may be less trusting.
Choose or create one or more appropriate independent variables to measure community membership. Write down hypotheses about how these independent variables might predict trust. Examine the relationship between trust and your independent variable, using contingency tables. Discuss your results carefully, paying attention to statistical significance, the meaning of the result, any potential problems with the method used, and questions of interpretation. Was your hypothesis confirmed or not?
Trust may also be affected by demographics such as age, gender, race and marital status. Again, examine the relationship between trust and these demographics, using appropriate bivariate statistical methods such as contingency tables or t tests.
3. Regressions with and without controls
We would like to know whether the effects of community stability are really caused by some other omitted variable. To find out, we can run multivariate regressions. For this we require a continuous dependent variable. One way to do this is to add different variables together. Create a new variable on the basis of CANTRUST and BEFAIR, measuring how much people trust others, and name it appropriately. Examine the new variable and comment on whether it is an appropriate dependent variable for regression analysis.
Next, run a bivariate regression examining the effect of community stability upon this new variable. Write up and interpret your findings, including formatted output from STATA or another program. Include a precise interpretation of the regression coefficient. What do the p-value and R2 tellyou?
Now run one or more multivariate regressions, including an appropriate set of demographic control variables. Again, write up and interpret your findings. Be sure to discuss any changes you observe in the effect of the independent variable. Does this change how you interpret your original bivariate regression? Also discuss your control variables: which of them appear to be important for explaining trust.
Discuss your results. What can you conclude about the relationship between community stability and trust? Identify any potential problems with your methods, and if possible test for them. Were your hypotheses confirmed? How might your research design be improved?
CANTRUST – PEOPLE CAN BE TRUSTED OR CAN’T BE TOO CAREFUL
BEFAIR – HOW OFTEN DO YOU THINK PEOPLE TAKE ADVANTAGEoften do you think people take advantage
MEMOTHER MEMBERSHIP IN FRATERNAL GROUP
MEMBERSHIP IN SERVICE GROUP
MEMBERSHIP IN VETERAN GROUP
MEMBERSHIP IN POLITICAL CLUB
MEMBERSHIP IN LABOR UNION
MEMBERSHIP IN SPORTS CLUB
MEMBERSHIP IN YOUTH GROUP
MEMBERSHIP IN SCHOOL SERVICE
MEMBERSHIP IN HOBBY CLUB
MEMBERSHIP IN SCHOOL FRATERNITY
MEMBERSHIP IN NATIONALITY GROUP
MEMBERSHIP IN FARM ORGANIZATION
MEMBERSHIP IN LITERARY OR ART GROUP
MEMBERSHIP IN PROFESSIONAL SOCIETY
MEMBERSHIP IN CHURCH GROUP
MEMBERSHIP IN ANY OTHER