SOC 300--Fall, 2001
M. Senter
Reading: Applied Social Research, Chapters 14 (excluding final section on
multivariate analysis), 15, 5, 13
Readings in Social Research, Chapter 5 and 6
Homework (on the web page): 5, 6, 7, 8
Consult Web Page: Introduction to
SPSS; Understanding Chi-Square; Example of Bivariate Table under Tables
The format of this exam will be the same as Exam #1--multiple-choice, short-answer, and computer question. (This exam focuses a great deal on "doing" analyses.)
1. Be able to create--by hand and with SPSS--univariate tables (frequency distributions, percentage distributions, cumulative percentage distributions) and graphs (bar graphs, line graphs, pie graphs, and histograms). Make sure that you know how to interpret (put into words) what these tables and graphs tell you.
2. Be able to compute--by hand and with SPSS--means, medians, and modes. Make sure that you can compute by hand the measures of central tendency from raw data (that is, from the actual data cases before they have been summarized in tables) and from data that have already been summarized in tables. Make sure that you know when each statistic is appropriate to use and that you know how to make sense of these statistics (in words).
3. Be able to compute by hand the following measures of dispersion: the range and the standard deviation. Know the special properties of the standard deviation when your data are normally distributed. Know the proportions of the area under the normal curve between selected points.
4. Be able to summarize data using rates and ratios. Be able to transform percentages into rates per 1,000 and rates per 100,000 and to transform rates per 100,000 into rates per 1,000 and percentages.
5. Be able to compute the percentage change and percentage-point change (epsilon) in a variable over time.
6. Be able to create, percentagize, and make sense of a cross-tabulation to test a hypothesis. Be able to do this by hand and with SPSS. (Make sure that you can distinguish between an independent and a dependent variable.) Make sure you know how to RECODE variables with SPSS and how to eliminate MISSING DATA as appropriate to the variable you are analyzing.
7. Be able to use SPSS to compute a chi-square statistic from a cross-tabulation and know the interpretation (meaning) of the statistic and its significance level (p).
8. Be able to create by hand and with SPSS a clustered bar graph and a scattergram (scatterplot). Be able to discuss what these bivariate graphs tell you.
9. Be able to compute--by hand and with SPSS--the lambda statistic from a cross-tabulation and know how to interpret it. Know the interpretation of gamma and the correlation coefficient. You do NOT have to know how to compute gamma and the correlation coefficient by hand, but you do want to know how to compute all of these statistics with SPSS.
10. Be able to discuss the differences between reliability and validity and know, generally, how one would assess the reliability and validity of measures. Be able to use SPSS to compute Cronbach's alpha and know what it means.
11. Make sure that you know how to create with SPSS and interpret (make sense of) a correlation matrix.
12. Know how to compute an index or scale with SPSS. Know when it is appropriate to do so.
Other Concepts and Terms You Want to Understand:
pre-coding, coding, codebook, coding, recoding, raw data, optical scan (opscan) form (sheet), direct data (terminal) entry, CATI, system file, missing data, valid data, variable name (mnemonic), variable label, value label, primary data analysis, secondary data analysis, quantitative data, qualitative data, data cleaning, wild code check, consistency check, mutually exclusive response options, exhaustive response options, level of precision, item, level of measurement, nominal variables, ordinal variables, interval variables, ratio variables, univariate tables, frequency distribution, percentage distribution, cumulative percentage distribution, bar graph, line graph (frequency polygon), pie graph or chart, histogram, proportion, percent, rates per 1000, rates per 100,000, ratios, percentage change vs. percentage-point change, epsilon, univariate statistics, measures of central tendency, mean (average), median, mode, measures of dispersion, range, standard deviation, normal (bell) curve, description vs. explanation, bivariate tables, cross-tabulation, cell of a table, percentagizing a cross-tabulation to test a hypothesis, independent variable, dependent variable, measure of independence, inferential statistics, chi-square statistic, observed frequencies, expected frequencies, the marginals of a crosstabulation (they're just the row and column totals), significance level (p), confidence level (1-p x 100), a "statistically significant" relationship, "statistical significance" vs. "substantive importance," inferential statistics, measures of association, proportionate reduction in error (PRE) measure, lambda statistic, gamma statistic, correlation coefficient, positive correlation (direct relationship), negative correlation (inverse relationship), zero (no) correlation, strong relationship, weak relationship, moderate relationship, scattergram or scatterplot, index or scale, Likert scale, semantic differential scale, error in measurement, random error, systematic error, test-retest reliability, split half reliability, Cronbach's alpha, face validity, criterion validity, concurrent validity, predictive validity, construct validity
• Bring a calculator if it would
be helpful. (I'll bring one, too, and the computers contain a calculator as well.)
• I will give you the formulae for
the standard deviation, and lambda statistic on the exam.
• The data analysis using SPSS will
involve the data from the 1998 General Social Survey. This is the data
(GSS1998.sav) that
you access with the FILE/OPEN on the file server CHSBS1.