Presentation is loading. Please wait.

Presentation is loading. Please wait.

Using the SDA on the Web Ed Nelson, CSU Fresno Social Science Research and Instructional Council.

Similar presentations


Presentation on theme: "Using the SDA on the Web Ed Nelson, CSU Fresno Social Science Research and Instructional Council."— Presentation transcript:

1 Using the SDA on the Web Ed Nelson, CSU Fresno Social Science Research and Instructional Council

2 Survey Documentation and Analysis (SDA) Program u Written at UC Berkeley u Used by ICPSR and others-- referred to as DAS (Data Analysis System) u Data files must be converted to SDA format before use. ICPSR has converted a number of data sets in their topical archives into SDA format and are converting more.

3 Sources of Data at ICPSR (http://www.icpsr.umich.edu) u ICPSR topical archives –National Archive of Computerized Data on Aging (NACDA) –National Archive of Criminal Justice Data (NACJD) –International Archive of Education Data –Substance Abuse and Mental Health Data Archive (SAMHSA) u General Social Survey u National Election Study

4 General Procedure u Select study u Open window to browse codebook u Select what you want to do u Click on START

5 What Can You Do? u Browse codebook u Subset data u Download data and documentation u Run statistical procedures

6 Statistical Procedures u Frequencies u Crosstabs u Comparison of means u Comparison of correlations

7 What Else Can You Do? u Recode (temporarily) u Use control variables u Use filter variables u Use weight variable

8 Documentation and Data u Codebook (ASCII/PDF) u SPSS/SAS/Stata syntax u Data file

9 Using Statistical Programs u Specify variables u Select display options (e.g., statistics, text to display) u Select action (run, clear)

10 Frequencies Program -- Specify Variables u Row variable (required) u Filter variables u Weight variable

11 Frequencies Program -- Select Statistics u Percents u Central tendency -- mean, median, mode u Variability -- standard deviation, variance u Coefficient of Variation u Standard error of the mean

12 Example: Monitoring the Future u Explores values, behavior, and lifestyles of American youth u Focus on drug use u 1975 to present u Investigators: Jerald G. Bachman, Lloyd D. Johnson, and Patrick M. O’Malley, University of Michigan, Institute for Social Research

13 Monitoring the Future -- Study Design u Self-administered questionnaire u 8th, 10th, and 12th graders u Multistage area probability sample u Students randomly assigned to one of six questionnaires u Core questions -- demographics and drug use

14 Select Study -- 1998 Monitoring the Future u ICPSR study number 2751 u 12 graders u Year: 1998

15 Monitoring the Future -- Variables of Interest u Demographics: V150 (sex), V151 (race) V163 (father’s educational level), V164 (mother’s educational level) u Religious variables: V169 (attend religious services), V170 (importance of religion) u Educational aspirations: V183 (attend four-year college) u Recreation: V194 (# of times go out per week), V195 (# of dates per week) u Drug use: V103 to V108 (alcohol), V112 to V114 (Marijuana), V124 to V126 (Cocaine)

16 Monitoring the Future -- Frequencies u Alcohol use (V107--number of times drank alcohol enough to feel pretty high) u Importance of religion in life (V170)

17 Crosstabs Program -- Specify Variables u Dependent variable -- row variable (required) u Independent variable -- column variable (required) u Control variables u Filter variables u Weight variable

18 Crosstabs Program -- Select Statistics u Percents -- vertical (row), horizontal (column), total u Chi square (Pearson’s, Likelihood Ratio) u Eta u Gamma u Tau-b and Tau-c u Somer’s d

19 Monitoring the Future -- Crosstabs (Bivariate) u Row (dependent) variable -- V107, number of times drank alcohol enough to feel pretty high u Column (independent) variable -- V170, importance of religion

20 Recoding (temporarily) u Let’s start by recoding the number of times the respondent drank alcohol enough to feel pretty high into two categories--none or few (1-2) and half or more (3-5) u V107 (r: 1-2 “few or none”; 3-5 “half or more”) –Semicolon separates recodes –Assigns values of 1, 2, etc. –Value labels can be inserted within quotes u Missing data -- anything not recoded is treated as missing data

21 Monitoring the Future -- Crosstabs (Multivariate) u Now that we have run the two-variable crosstab, let’s add a control variable. u We’ll add the variable sex (V150) as the control variable.

22 Comparison of Means Program -- Specify Variables u Dependent variable (required) u Row (independent) variable (required) u Column (control) variable u Control (additional) variable u Filter variables u Weight variable

23 Comparison of Means Program -- Select Statistics u Mean of dependent variable u Difference from overall mean u Standard deviation u Number of cases, weighted number of cases u Standard errors and confidence intervals

24 Comparison of Means Program -- Select Statistics (Advanced) u Complex samples –Standard errors –Design effect –RHO statistic u ANOVA

25 Monitoring the Future -- Comparison of Means u Compute the mean use of Marijuana over the respondent’s lifetime by the number of times the respondent goes out in a week u Dependent variable is V112 (use of Marijuana over one’s lifetime) u Row (independent) variable is V194 (number of times goes out in a week) u Column (control) variable is V150 (sex)

26 Comparison of Correlations Program -- Specify Variables u Variables to be correlated (required) u Row variable (required) u Column (control) variable u Control (additional) variable u Filter variables u Weight variable

27 Comparison of Correlations Program -- Select Statistics u Correlation –Pearson’s –Log of odds-ratio (for dichotomies) u Difference from overall correlation u Standard errors

28 Filter Variables u Can also use filter variables to select particular cases u Variable name (____; ____; ___) –Where _____ stands for a range of values or a particular value –E.g., sex (1) –E.g., age (65-89) u Using more than one filter variable –E.g., sex (1), age (65-89) to select all those who are 1 on sex and age 65 to 89 –Joins the two variables with an AND

29 Subsetting Data Sets u Select the files you want to construct –Data file (ASCII) –Codebook (ASCII) –Data definitions for SPSS or STATA or SAS u Select the cases to include (leave blank if you want all the cases) u Select the variables to include


Download ppt "Using the SDA on the Web Ed Nelson, CSU Fresno Social Science Research and Instructional Council."

Similar presentations


Ads by Google