Deck 10: Discriminant Analysis

Question 1

Discriminant analysis seeks to identify a combination of IVs, measured at the nominal level, which best predicts membership in a particular group, as measured by an interval DV.

Answer 1

False

Question 2

Descriptive discriminant analysis refers to the process of classifying participants into groups based on a combination of measures.

Answer 2

False

Question 3

The correlated linear combinations of original variables identified in descriptive discriminant analysis are called discriminant functions.

Answer 3

False

Question 4

If the discriminant analysis is being used for purposes of prediction or classification, the goal of the interpretation of discriminant functions is to determine dimensions that serve as the basis for reliably classifying participants into groups.

Question 5

Discriminant analysis is often seen as the reverse of MANOVA.

Question 6

In discriminant analysis, the IVs serve as the grouping variables and the DVs are the predictors.

Question 7

If the goal of analysis is prediction, the researcher would determine the number of dimensions that maximize the differences among the groups in question.

Question 8

If prediction is the goal of analysis, the researcher might use discriminant function scores in order to describe group differences.

Question 9

The main analysis obtained from a discriminant analysis is the summary of the discriminant functions.

Question 10

A discriminant score is analogous to a factor score.

Question 11

A canonical correlation is a value that is equivalent to the correlation between the discriminant scores and the levels of the independent variables.

Question 12

A high value for this correlation indicates a function that does not discriminate well between participants.

Question 13

Standardized coefficients are the basis for the calculation of discriminant scores in discriminant analysis.

Question 14

Standardized coefficients are used to assess the relative contributions of individual variables to the discriminant function(s).

Question 15

The hit rate refers to the number of correct classifications.

Question 16

Assessing the accuracy of the hit rates cannot be accomplished through the use of the jackknife procedure.

Question 17

An important consideration when assessing a classification procedure is the cost of classification.

Question 18

Misclassification is best explained through the statistical analysis rather than being assessed and explicated by the knowledge and intuition of the researcher.

Question 19

In standard, or direct, discriminant analysis, each predictor is entered into the equation simultaneously and is assigned only its unique association with the groups, as defined by the DV.

Question 20

In sequential, or hierarchical, discriminant analysis, the predictors are entered into the analysis in an order specified by the researcher.

Question 21

The assumptions for discriminant analysis are basically the same as those for MANOVA, and there is no need for adjustments when classification is the goal of analysis.

Question 22

Discriminant analysis is not sensitive to sample size.

Question 23

Multivariate normality in discriminant analysis implies that the sampling distributions of the linear combinations of dependent variables are normally distributed.

Question 24

There exists no test of the normality of all linear combinations of sampling distributions of means of predictor variables in discriminant analysis.

Question 25

Checking for univariate normality is a good starting point for assessing possible violations of homoscedasticity.

Question 26

Discriminant functions are:

A) Linear combinations of the original (predictor) variables that maximize the between-to-within association, as measured by the sum-of-squares and cross-products (SSCP) matrices.
B) Uncorrelated linear combinations of the original (predictor) variables that maximize the between-to-within association, as measured by the sum-of-squares and cross-products (SSCP) matrices.
C) Correlated linear combinations of the original (predictor) variables that maximize the between-to-within association, as measured by the sum-of-squares and cross-products (SSCP) matrices.
D) Uncorrelated linear combinations of the original dependent variables that maximize the between-to-within association, as measured by the sum-of-squares and cross-products (SSCP) matrices.

Question 27

The logic behind discriminant analysis involves finding:

A) The function with the smallest eigenvalue.
B) The function with the largest eigenvalue.
C) Discriminant functions that are related.
D) None of the above.

Question 28

A discriminant analysis output typically has four parts. Which one of the four described below is incorrect?

A) Preliminary statistics that describe group membership and covariances.
B) Significance tests and strength of relationship statistics for each discriminant function.
C) Discriminant function coefficients.
D) Group classification.

Question 29

Discriminant analysis produces a series of preliminary statistics that assist in interpreting the overall analysis results, which include all but which of the following?

A) Table of means and standard deviations for each DV by group.
B) ANOVA analysis testing for group differences among the IVs.
C) Covariance matrices.
D) Box's test.

Question 30

The final step in the interpretation of discriminant analysis results is to:

A) Determine the extent to which group differences support the functions generated by reviewing group variance for each function as presented in the table of Functions at Group centroids.
B) Determine the extent to which group differences support the functions generated by reviewing group means for each function as presented in the table of Functions at Group centroids.
C) Determine the extent to which groups support the functions generated by reviewing group means for each function as presented in the table of Functions at Group centroids.
D) Determine the extent to which groups support the functions generated by reviewing group variance for each function as presented in the table of Functions at Group centroids.