AN SPSS PRIMER - Cal U

AN SPSS PRIMERPrepared by Dennis C. Sweeney, Ph.D.1

An SPSS PrimerENTERING DATASPSS Data Editor: (Data View)This screen is similar to a spread sheet. It gives you access to all of the procedures for entering,transforming, printing, and analyzing your data.OVERVIEW Each row contains the data for one person. Each column contains the data for one variable. Variable names are limited to eight characterso The names must start with a letter.o The names cannot contain a space.FILE This menu contains all of the information forSaving the dataPrinting the dataOpening a data file.Closing a data file.DATA This menu lets you Create variables. Transform variables. Sort the data file.ANALYZEThis menu lets you do all of the statistical analyses.GRAPHThis menu has all of the graph functions.2

An SPSS PrimerFORMATTING VARIABLESSPSS Data Editor: (Variable View)This screen lets you describe each variable found in the Data View. It is saved along with the data file.Each variable is represented by a row.NAMEThis is the short variable name seen in the Variable View.TYPEThis identifies the variable as being one of the followingNumberDateDollarString (Letter) WIDTHThis tells SPSS how many integers/letters wide to display the value. For numbers this includesthe number of decimal places.DECIMALS This tells SPSS how many decimals to display. It will save many more decimals in the DataFile.LABELThis is an extended label for the variable. This label can be almost any length.VALUESThis gives values for categorical each value. For example, Males are coded as 1 for Gender,Females are coded as 2.MISSINGWhenever a person has left something blank you should assign a missing value. Zero (0),unless you tell SPSS that this should be treated as missing, is treated as a number and willdistort any statistical calculations.3

An SPSS PrimerDOING ANALYSESSPSS Data Editor: Analyze MenuThe analyze menu has the following options that we will use in this class.DESCRIPTIVE STATISTICSThis has a sub-menu that contains all of the procedures for descriptivestatistics. Clicking on Descriptive Statistics will give the secondary menu shown on the next page.COMPARE MEANS:Variance.This has all of the procedures for doing t-tests and simple Analysis ofGENERAL LINEAR MODEL:This brings up a sub menu for conducting advanced Analysis ofVariance and Multivariate Analysis of Variance routines.CORRELATE:This brings up a sub-menu for simple and partial correlations.REGRESSION:This brings up a menu for conducting Multiple Regression.DATA REDUCTION:SCALE:This brings up a sub-menu for conducting Factor Analysis.This brings up a menu for conducting reliability (internal consistency) analyses.NONPARAMETRIC TESTS:This menu brings up sub-menus for conducting a variety of nonparametric tests such as chi-square.4

An SPSS PrimerDOING ANALYSESSPSS Data Editor: Analyze Descriptive StatisticsFREQUENCIES:This choice brings up a sub-menu forconducting frequency tables and graphs for the variables forwhich the frequencies are calculated. (See below)DESCRIPTIVEThis choice brings up a sub-menu forcalculating descriptive statistics. It should be used for onlyinterval and ratio data.EXPLORE: This choice brings up a sub-menu forconducting a variety of simple inferential analyses includingt-tests.CROSSTABS: This choice brings up a sub-menu forcreating crosstabulation tables and includes options forconducting chi-square analyses on the results.SPSS Data Editor: Analyze Descriptive Statistics FrequenciesSTATISTICS:This brings up asub-menu for calculating descriptivestatistics. Keep in mind that youwould select this option only forinterval and ratio data.CHARTS:This option givesaccess to preparing a chart for thevariables you for which you arecomputing frequency tables. Bar Charts should be selectedfor Nominal data.Histogram should be selectedfor Interval and Ratio data.FORMAT:This option lets you tell SPSS how to order the data in the frequency table. You can selecthighest to lowest (preferred) or lowest to highest.5

An SPSS PrimerSPSS OUTPUTSPSS ViewerAll SPSS output is contained in its own file that is separate from the data file. You can modify text in theoutput, copy tables and figures so you can place them in other documents, and print the whole file or partsof it.To modify the label for tables and figures.1. Click the table or figure to make it active. When a table or figure is active it is surrounded by a boxwith handles. The Gender bar graph is active.2. Click on the title to make it active.3. Type your text for the appropriate title.6

An SPSS Primert-testSPSS Analyze - Compare MeansThis screen allows you to select a variety ofstatistical tests that compare means.MEANS compares the means for variousbroken down by another categorical variable(Gender, Race, etc.). It SPSS calculates thedescriptive statistics but does not do astatistical test of the difference between themeans.ONE-SAMPLE T TEST compares the meansfor within group (repeated measures) studies.INDEPENDENT-SAMPLES T TESTcompares the means of two independentgroups.To do the t-test you need to do two things. First, identify the dependent variable. Second, identify theindependent variable. The screen for doing this is shown below.1. Test Variable(s): is the window forthe dependent variable. You can enterone or more variables here.2. Grouping Variable: This is thewindow for entering the independentvariable. You need to tell SPSS whatthe values are for the grouping variableso, click ‘Define Groups ’ This willbring up the Define Groups screen.Then you need to enter the values foreach group. For example, suppose thegrouping variable is Gender and thereare three values for Gender (1 for male,2 for female, and 3 for undecided) andyou want to compare males to females.Then enter 1 for Group 1 and 2 forGroup 2 then click Continue.3. Click OK to do the analysis.7

An SPSS Primert-test: Sample output.SPSS computes two tables for the t-test. The Group Statistics table is a breakdown of the descriptive statistics by Gender.Group StatisticsGenderMaleRestraint RawScore669Mean10.58Std. Deviation4.485Std. ErrorMean.1736839.314.622.177NFemaleThe second table is the t-test summary. Levene’s Test is an evaluation of the assumption that the variances of the two samples are equal(one of the t-test assumptions). If the differences are not statistically significant, shown in the Sig. column, then use the row labeled Equalvariances assumed. Otherwise use the row labeled Equal variances not assumed.The t is shown in the t column, and the significance level is shown in the Sig. (2-tailed) column. SPSS calculates the actual significancelevel. In the example the significance level would be p .0001. The Mean Difference column shows the difference between the Male and Femalemeans.Independent Samples TestLevene's Test forEquality ofVariancesFSig.t-test for Equality of MeanstdfSig. (2-tailed)MeanDifferenceStd. ErrorDifference95% Confidence Intervalof the DifferenceLowerRestraint RawScoreEqual variances assumedEqual variances not 565.1261349.878.0001.270.248.7841.7568

An SPSS PrimerONE-WAY ANALYSIS OF VARIANCESPSS Analyze- Compare Means - One-way ANOVAOne-Way ANOVA. This is used when thereis one independent variable with more thantwo levels and one dependent variable.This screen is used to select thedependent variables ‘DependentVariable List’ and the independentvariables: ‘Factor’Post Hoc tests can be selected usingthe Post Hoc button. Select thosethat are appropriate for the analysisyou are using: Either Tukey orScheffe.A sample of the One-way ANOVA output is shown on the next page.9

An SPSS PrimerONE-WAY ANALYSIS OF VARIANCESPSS OutputSPSS provides several tables summarizing the ANOVA analysis. The first is the ANOVA summary table shown below. As with the t-test SPSScalculates the actual significance level in the Sig. column.Conformity Raw ScoreBetween GroupsSum ofSquares467.299df3Mean Square155.76616.709Within Second major table is the summary of the pairwise comparison of the means for the independent variable. This table has a lot of redundancy. Forexample, in the second row, the Asian mean is compared to Black, White and Hispanic means. In the third major row the Black mean iscompared to the Asian, White, and Hispanic means. SPSS flags significant differences with an *.Dependent Variable: Conformity Raw ScoreTukey HSD(I) RaceAsianBlackWhite(J) RaceBlackWhiteStd. Error.612.518HispanicSig.1.000.855Lower Bound-1.62-.92Upper nBlackHispanicHispanic95% Confidence .561.283.883.573.7810

An SPSS PrimerCORRELATIONAL ANALYSES.SPSS Analyze - CorrelateThis screen lets you select three types of correlational analyses.BIVARIATE- This is the correlation between two variables.For regular correlations select ‘Bivariate’.PARTIAL -This is the analysis of the correlation between twovariables removing the influence of a third variable.Use this screen to select the variablesto analyze. You can pick as manyvariables as you want.Note: As shown here,-The normal Pearson correlation willbe calculated.- SPSS will calculate a two-tailedsignificance level.- Correlations that are statisticallysignificant will be flagged with anasterisk.- The Options button permits you toselect various options such ascalculation of the descriptive statistics.11

An SPSS PrimerCHI SQUARE ANALYSISSPSS Analyze - Descriptive Statistics - CrosstabsCROSSTABS - This selection gives a breakdown of two or more nominal variables.Rows. This selection identifies thevariable that will be listed in each row.Column(s) This selection identifies thevariable that will be listed in the tablecolumns.Statistics Use this button to accessthe Chi-square option that is located inthe upper left-hand corner of this optionbox.Cells: This option gives various options for displayingthe cross tab table. One useful option is the Percentagescheck box. This will tell SPSS to put the percent ofresponses for each Row or Column or the Totalpercentages. This is not always necessary but can behelpful for interpreting the data.A sample of the chi-square output is shown on the nextpage.NOTE: You can also compute Chi-square by going to the Analyze - Nonparametric Tests selection. Ifyou chose this option you do not get the Crosstabs table.12

An SPSS PrimerCHI-SQUARE ANALYSISSPSS OutputThis is a sample of a Crosstab table. It shows the Row Percentages (% within Gender), Column percentage(% within Supervisory Level) and Total percent (% of Total). Sometimes these percent figures can beconfusing, so they should be used only when needed.Gender * Supervisory Level CrosstabulationGenderMaleCountSupervisor.279% withinGender% withinSupervisoryLevel% of TotalFemaleCount% withinGender% withinSupervisoryLevel% of TotalTotalCount% withinGender% withinSupervisoryLevel% of TotalSupervisory LevelFirst -LevelMid-LevelExecutiveManagerManageror V.P.262365212President %13.6%2.6%100.0%This table summarizes the Chi-Square analysis. The Pearson Chi-Square is the one used most frequently.Chi-Square TestsPearson Chi-SquareLikelihood RatioLinear-by-LinearAssociationN of Valid CasesValue218.761(a)235.176208.67244Asymp. Sig.(2-sided).000.0001.000df1798a 0 cells (.0%) have expected count less than 5. The minimum expected count is 16.27.13