Psy 524AinsworthS’ 04Final ExamPsy 524AinsworthPlease write on the exam. You have exactly 2 hours. You can use a calculator (I suggest it) and nothing else. Good Luck and have a great summer!!!!Name_______________________________________Psy 524AinsworthS’ 04Final Part IShort answer questionsDiscriminant Function Analysis (12 pts)1. Purpose of the method (give at least 2 research questions).2. How is discriminant function analysis different from MANOVA?3. Describe the assumption of homogeneity of variance/covariance matrices.4. Describe the use of classification scores in discriminant function analysis (basically, how do you get them and what are they used for, no equations).Logistic Regression (12 pts)5. Purpose of the method (give at least 2 research questions).Psy 524AinsworthS’ 046. How is logistic regression different from discriminant function analysis? When is it better to use discriminant function analysis over logistic regression?7. What is a logit? How can you transform a logit into a probability?8. What is the only real limitation with logistic regression? How does probit analysis differ from logistic regression?Factor Analysis (12 pts)9. Purpose of the method (give at least 2 research questions).10. What is the theoretical difference between principal components analysis and factor analysis?Psy 524AinsworthS’ 0411. What is the main purpose of factor rotation? Explain this. What is the difference between orthogonal and oblique rotation?12. Describe Kaiser’s criterion for picking the number of factors (hint: the default method in SPSS). What is this criterion based on (why do we do it?)?Structural Equation Modeling (14 pts)13. How is confirmatory factor analysis different than exploratory factor analysis (give me at least 2 ways)?14. What are the differences between the structural and measurement parts of an SEM model?Psy 524AinsworthS’ 0415. What do we use as data in an SEM model? What formula do we use to calculate the number of data points? How do we find the degrees of freedom for a model?16. Describe these matrices:a)  (Beta)b)  (Gamma) c)  (Phi)17. Define models that are:a) overidentifiedb) just identifiedc) under identifiedd) recursive.Psy 524AinsworthS’ 04Final Part IIOutput interpretationSection 1 (12.5 pts) – Discriminant Function Analysis Predicting Political Conservatism (Very Liberal, Liberal, Moderate, Conservative, Very Conservative)Analysis Case Processing Summary1194 91.125 1.986 6.65 .4116 8.91310 100.0Unweighted CasesValidMissing or out-of-range groupcodesAt least one missingdisc riminating variableBoth missing or out-of-rangegroup c odes and at least onemissing disc riminating variableTotalExcludedTotalN PercentTests of Equality of Group Means.935 20.718 4 1189 .000.873 43.244 4 1189 .000.837 57.813 4 1189 .000.910 29.448 4 1189 .000.946 17.068 4 1189 .000.954 14.267 4 1189 .000.898 33.732 4 1189 .000.938 19.518 4 1189 .000.754 96.801 4 1189 .000.969 9.605 4 1189 .000.954 14.288 4 1189 .000.927 23.552 4 1189 .000.886 38.362 4 1189 .000.899 33.457 4 1189 .000ITEM26 REAL MAN SHOULDAVOUD SHOWING WEAKNESSITEM29 FEMINISTS TOOAGGRESSIVEITEM31 TRADITIONAL SEXROLESITEM33 WOMEN NOT GOODENGINEERSITEM35 FEMALEAUTO-MECHANICSITEM37 ONE MEN SHOULDPLAY FOOTBALLITEM41 MALE NURSESITEM42 MEN WHO CRY INPUBLICITEM44 HOMOSEXUALITYITEM45 WOMEN IN COMBATITEM47 FEMALE FIGHTERPILOTITEM49 WOMENSUPERVISORSITEM51 MALE SECRETARIESITEM52 MAN WHO STAYSHOMEWilks' Lambda F df1 df2 Sig.1. What does the box above (tests of equality of group means) tell you?Psy 524AinsworthS’ 04Analysis 1Box's Test of Equality of Covariance MatricesLog Determinants14 -3.18514 -4.37614 -4.71014 -3.22114 -1.92714 -3.299PSD STUDENTS POLITICALSELF-DESCRIPTIONVERY LIBERALLIBERALMODERATECONSERVATIVEVERY CONSERVATIVEPooled within-groupsRank Log DeterminantThe ranks and natural logarithms of determinants printed arethose of the group c ovarianc e matrices.Test Results806.6011.80642099711.025.002Box's MApprox.df1df2Sig.FTests null hypothesis of equal population covariance matrices.1. What does Box’s M statistic test? Is it a problem here? Why or why not?Summary of Canonical Discriminant FunctionsEigenvalues.480a88.1 88.1 .570.036a6.6 94.7 .186.020a3.7 98.4 .140.009a1.6 100.0 .092Function1234Eigenvalue % of Variance Cumulative %CanonicalCorrelationFirst 4 canonical discriminant functions were used in the analysis.a. Wilks' Lambda.634 539.639 56 .000.938 75.581 39 .000.972 33.693 24 .090.991 10.126 11 .519Test of Function(s)1 through 42 through 43 through 44Wilks' Lambda Chi-square df Sig.2. What do the two boxes above (Eigenvalues and Wilk’s Lambda) tell you (annotate and interpret them)?Psy 524AinsworthS’ 04Structure Matrix.822* -.162 -.028 -.105.632* .041 .268 .410.543* -.175 -.366 .034.509* .313 -.242 .073.474* .203 .371 -.158.393* .313 .267 .159.372* .254 .221 .078.345* .032 -.091 .026.314* -.119 .001 .095.251* -.207 .056 .231.456 .587* -.188 -.289.436 .456* .123 -.147.348 .397* -.284 .150.294 .352 .037 .505*ITEM44 HOMOSEXUALITYITEM31 TRADITIONAL SEXROLESITEM29 FEMINISTS TOOAGGRESSIVEITEM51 MALE SECRETARIESITEM52 MAN WHO STAYSHOMEITEM49 WOMENSUPERVISORSITEM26 REAL MAN SHOULDAVOUD SHOWING WEAKNESSITEM35 FEMALEAUTO-MECHANICSITEM37 ONE MEN SHOULDPLAY FOOTBALLITEM45 WOMEN IN COMBATITEM41 MALE NURSESITEM33 WOMEN NOT GOODENGINEERSITEM42 MEN WHO CRY INPUBLICITEM47 FEMALE FIGHTERPILOT1 2 3 4FunctionPooled within-groups c orrelations between discriminating variables and standardizedcanonical discriminant func tions Variables ordered by absolute size of correlation within function.Largest absolute correlation between each variable and any di sc riminant func tion*. 3. What is the structure matrix? What does this tell you about the meaning of the discriminant functions (refer to your answer on number 3)?Classification StatisticsClassification Processing Summary13100911219ProcessedMissing or out-of-rangegroup codesAt least one missingdiscriminating variableExcludedUsed in OutputPsy 524AinsworthS’ 04Prior Probabilities for Groups.200 80 80.000.200 295 295.000.200 413 413.000.200 366 366.000.200 40 40.0001.000 1194 1194.000PSD STUDENTS POLIT ICALSELF-DESCRIPTIONVERY LIBERALLIBERALMODERATECONSERVATIVEVERY CONSERVATIVETotalPrior Specified Prior Effective Prior Unweighted WeightedCases Used in Analysis4. Where do the prior probabilities come from (“Prior” column from the table above)?5. What does function 2 tell you about separating the