PSC 4710: Data Analysis Workshop - PowerPoint Presentation

Slide1

PSC 4710: Data Analysis Workshop

What’s the purpose of this exercise?The workshop’s research questions:Who supports war in America?How consistent is support for conflict across different types of war?Should we be considering other, related questions? If this project were to be expanded, what could else we look at… how different groups change their support over time, support for other types of conflict, varying patterns in other societiesWill you want to do your research paper on work that builds on this workshop? You research paper assignment will ask you to do a more complex paper on public opinion and some type of conflictHow would a research paper using Pew data be different than this assignment with respect to its literature review and what methods you would need to use?

PSC 4000-level Stats WorkshopSlide2

Where Do You Get Survey Data?

Why are we looking at survey data from the Pew Foundation? How do we download a dataset from Pew?Why do we have to read the methodology sheet first (it’s boring, but critical)?What do we do with the questionnaire?How does scientific method work?: In theory and in the real worldCan we operationalize our dependent variable(s)? Are there some questions that we can use to identify who supports war?Do we have the independent variable(s) we need? Are there any questions that identify characteristics (e.g., gender) that we expect to be correlated with whether a person supports war?What specific hypotheses can we explore with our survey? What specific relationships do we expect to find between our independent and dependent variables?

PSC 4000-level Stats WorkshopSlide3

What Do You Need to Turn In?

Your assignment has two components: The first asks you to complete several data assignments The second is an 1000-word analytical essayThe assignments related to SPSS ask you to:Recode and label both new variables and their response categoriesYou will be asked to create dummy (1-0 value) variables. For example, you can use a question asking about religion to create a new variable called “Catholic,” where all Catholic respondents are coded 1 and all other respondents are coded zeroCalculate descriptive statisticsMean (the average value of the response categories)Median (the middle value for all of your observations)

Mode (the most common response if there’s no natural order to variable’s response categories)Generate a bar graphDo a little bivariate

data analysis using SPSS’s

cross-tab

routine

PSC 4000-level Stats WorkshopSlide4

Getting Started with SPSS

Open your dataset with SPSSOpen a new syntax fileWhat are the four SPSS screens you will be working with?Data editor: Data viewData editor: Variable viewThe syntax file (Why is it WAY better than just pointing and clicking all of the time? Why is this the ONLY file that you will want to save in most cases? ) The output filePSC 4000-level Stats WorkshopSlide5

Recoding Data in SPSS

Why do we need to recode?Need to drop refusals (but we want to be careful to think about what “don’t know means… e.g. if someone doesn’t know that she’s a born-again Christian, she’s not one)Variables may not measure constructs correctly, so we need to create different categories that make senseWe’ll want to create dummy (0-1) variablesHow do we recode?Use the point and click menus at the top of the data editorUse the “Recode Into Different Variable” optionSelect variable you want to recodeTell SPSS how you want to recode the variableName and label the new variableDefine valuesDo not hit OK; instead, you want to paste command into syntaxHighlight the appropriate syntax and click the run (arrow) button

PSC 4000-level Stats WorkshopSlide6

Recoding Data: Step by Step Illustration

PSC 4000-level Stats WorkshopSlide7

Recoding Data: Step by Step Illustration

PSC 4000-level Stats WorkshopSlide8

Recoding Data: Step by Step Illustration

PSC 4000-level Stats WorkshopSlide9

Recoding Data: Step by Step Illustration

PSC 4000-level Stats WorkshopSlide10

Recoding Data: Step by Step Illustration

PSC 4000-level Stats WorkshopSlide11

Recoding Data: Step by Step Illustration

PSC 4000-level Stats WorkshopSlide12

Labeling Values in SPSS

Why do we need to label the values used for different response categories (e.g., 1=male, 0=woman)?You may not remember how you recoded valuesLabels will transfer into tables and graphsHow do we label values for our variable categories?Use “Define Variable Properties” optionSelect variable(s) with values you want to labelTell SPSS how you want to label the valuesPaste command into syntaxHighlight syntax and click runOnce you’ve got the syntax down, you may want to work much more quickly with other variables by copying and pasting that syntax and making appropriate substitutions

PSC 4000-level Stats WorkshopSlide13

Creating Value Labels for Response Categories:

Step by Step IllustrationPSC 4000-level Stats WorkshopSlide14

Creating Value Labels for Response Categories:

Step by Step IllustrationPSC 4000-level Stats WorkshopSlide15

Creating Value Labels for Response Categories:

Step by Step IllustrationPSC 4000-level Stats WorkshopSlide16

Creating Value Labels for Response Categories:

Step by Step IllustrationPSC 4000-level Stats WorkshopSlide17

Creating Value Labels for Response Categories:

Step by Step IllustrationPSC 4000-level Stats WorkshopSlide18

Descriptive Statistics in SPSS

What descriptive statistics may interest us?FrequenciesMean (the average), Medians (the middle observation), and Mode (rarely used) How do we compute descriptive statistics?Use “Descriptive Statistics” optionSelect “Frequencies”Select variable(s) that you want to analyzeSelect “Statistics”Select the statistics you want to viewOption 1 (if you will need the results later): click the paste command to create syntax; highlight the syntax and click runOption 2 (if you are using the results right now): click OK and the syntax won’t be pasted into your syntax file, but the output will be created

PSC 4000-level Stats WorkshopSlide19

Descriptive Statistics: Step by Step Illustration

PSC 4000-level Stats WorkshopSlide20

Descriptive Statistics

PSC 4000-level Stats WorkshopSlide21

Descriptive Statistics: Step by Step Illustration

PSC 4000-level Stats WorkshopSlide22

Descriptive Statistics: Step by Step Illustration

PSC 4000-level Stats WorkshopSlide23

Descriptive Statistics: Step by Step Illustration

PSC 4000-level Stats WorkshopSlide24

Descriptive Statistics: Step by Step Illustration

PSC 4000-level Stats WorkshopSlide25

Descriptive Statistics: Step by Step Illustration

PSC 4000-level Stats WorkshopSlide26

Graphing in SPSS

What kinds of graphs might we create?BarLineHistogramHow do we compute descriptive statistics?Use “Descriptive Statistics” optionSelect “Frequencies”Select variable(s) that you want to analyzeSelect “Charts”Select the charts you want to view; check the “percentage” option in most cases so that your charts will be more readablePaste command into syntaxHighlight syntax and click run

PSC 4000-level Stats WorkshopSlide27

Creating Graphs

PSC 4000-level Stats WorkshopSlide28

Creating Graphs

PSC 4000-level Stats WorkshopSlide29

Creating Graphs

PSC 4000-level Stats WorkshopSlide30

Creating Graphs

PSC 4000-level Stats WorkshopSlide31

Creating Graphs

PSC 4000-level Stats WorkshopSlide32

Creating Graphs

PSC 4000-level Stats WorkshopSlide33

Creating Graphs

PSC 4000-level Stats WorkshopSlide34

Bivariate Analysis in SPSS

What types of bivariate analysis might we conduct?CorrelationCrosstabs (this is your best option for tables that will make sense to a general audience)Significance testsHow do we conduct bivariate analysis?Use options in the “Analyze” menuSelect variable(s) that you want analyzeTell SPSS how you want to analyze the valuesPaste command into syntaxHighlight syntax and click runPSC 4000-level Stats WorkshopSlide35

Calculating Correlation

PSC 4000-level Stats WorkshopSlide36

Calculating Correlation

PSC 4000-level Stats WorkshopSlide37

Calculating Correlation

PSC 4000-level Stats WorkshopSlide38

Calculating Correlation

PSC 4000-level Stats WorkshopSlide39

Calculating Correlation

PSC 4000-level Stats WorkshopSlide40

Guidelines for Correlation

Less than .1: Very weak relationship.1 to .2: Weak relationship.2 to .3: Moderate relationshipAbove .4: Strong relationshipPSC 4000-level Stats WorkshopSlide41

Review: Variable Types

CategoricalRepresent categories or names (e.g. gender, religion)OrdinalRank ordered indicators (e.g. agree, strongly agree)IntervalOrdered values where all unit increases are roughly equivalent (e.g. age)PSC 4000-level Stats WorkshopSlide42

Calculating Significance

Crosstabs will allow you to see relationships between two variables in more detailPut dependent variable in the rowsPut independent variable in the columnsCan also test association:Categorical/categorical: Lambda, Cramer’s VCategorical/ordinal: Cramer’s VOrdinal/ordinal: Gamma, Tau-B (same number of categories)Categorical/Interval: EtaOrdinal/Interval: Spearman’s RhoInterval/Interval: Pearson’s RPSC 4000-level Stats WorkshopSlide43

Crosstabs and Tests of Significance

PSC 4000-level Stats WorkshopSlide44

Crosstabs and Tests of Significance

PSC 4000-level Stats WorkshopSlide45

Crosstabs and Tests of Significance

PSC 4000-level Stats WorkshopSlide46

Crosstabs and Tests of Significance

PSC 4000-level Stats WorkshopSlide47

Crosstabs and Tests of Significance

PSC 4000-level Stats WorkshopSlide48

Crosstabs and Tests of Significance

PSC 4000-level Stats WorkshopSlide49

Crosstabs and Tests of Significance

PSC 4000-level Stats WorkshopSlide50

Crosstabs and Tests of Significance

PSC 4000-level Stats Workshop