Choose the alternative that best completes the stem of each question.

1		An advantage of using an experimental multivariate design over separate univariate designs is that using the multivariate analysis
			A)	allows you to look at more complex relationships than does the univariate strategy.
			B)	provides a more powerful test of your hypotheses.
			C)	allows you not to worry about meeting restrictive assumptions characteristic of univariate statistics.
			D)	both a and b
2		Correlational multivariate analyses include
			A)	diriminant analysis.
			B)	multiple regression.
			C)	canonical correlation.
			D)	all of the above
			E)	both a and b only.
3		The presence of outliers in your data
			A)	affects the magnitude of the correlations calculated but not the slope of the regression line.
			B)	affects the slope of the regression line but not the magnitude of the correlations calculated.
			C)	affects both the slope of the regression line and the magnitude of the correlations calculated.
			D)	is less of a problem for multivariate statistics than it is for bivariate or univariate statistics.
4		An effective way of detecting outliers in a multivariate data set is to
			A)	convert raw scores to z scores and evaluate the degree of deviance of the z scores.
			B)	conduct individual Pearson correlations on your data before conducting any multivariate test.
			C)	do nothing; outliers do not significantly affect multivariate statistics.
			D)	both a and b
5		__________ occurs when variables in your analysis are highly correlated.
			A)	Heteroscedasticity
			B)	Multicollinearity
			C)	Reflecting
			D)	Outlier bias
6		____________ causes the observed value of a variable to differ to some extent from its true value.
			A)	Homoscedasticity
			B)	An outlier
			C)	Error of measurement
			D)	Multicollinearity
7		Generally speaking, multivariate analysis requires
			A)	fairly large samples.
			B)	small samples.
			C)	less concern over meeting assumptions than do univariate tests.
			D)	sampling from a population that is not normally distributed.
8		In a factor analysis, the correlation between an individual variable and an underlying dimension is a
			A)	discriminant function.
			B)	factor loading.
			C)	squared semipartial correlation.
			D)	canonical function.
9		The factors extracted in a factor analysis are made more clear and interpretable by
			A)	converting raw scores to z scores prior to analysis.
			B)	eliminating variables that have low correlations with other variables.
			C)	applying a square root transformation to the raw data prior to analysis.
			D)	statistically rotating factors.
10		According to Tabachnick and Fidell (2001), principal components analysis could be used to
			A)	help infer causality from correlational data.
			B)	extract as many factors as possible from your data prior to a factor analysis.
			C)	experiment with different communality values after an exploratory factor analysis.
			D)	determine the degree of contribution of a variable in a multiple regression analysis.
11		____________ is a statistical technique used to evaluate the relationship between two variables statistically controlling the effects of a third.
			A)	Discriminant analysis
			B)	Canonical correlation
			C)	Partial correlation
			D)	Factor analysis
12		A statistical technique that involves entering multiple predictor variables into an equation according to a specified order determined by theory is
			A)	hierarchical regression.
			B)	simple regression.
			C)	stepwise regression.
			D)	none of the above
13		The use of stepwise regression techniques is frowned on because
			A)	only three predictor variables can be entered at a time.
			B)	it tends to be too sensitive to causal relationships among variables.
			C)	it tends to capitalize on chance and may be limited to a particular sample.
			D)	all of the above
14		If you have multiple predictor variables and a dichotomous dependent variable, the most appropriate multivariate test is
			A)	stepwise regression.
			B)	canonical correlation.
			C)	factor analysis.
			D)	discriminant analysis.
15		Loglinear analysis
			A)	is a nonparametric statistic.
			B)	works much like chi-square.
			C)	can be used in place of ANOVA, MANOVA, or multiple regression where your data are categorical.
			D)	all of the above
16		If you have two sets of variables to correlate, the most appropriate multivariate test is
			A)	stepwise regression.
			B)	canonical correlation.
			C)	factor analysis.
			D)	discriminant analysis.
17		Using a MANOVA in place of a univariate analysis for a within-subjects experiment is advantageous because MANOVA
			A)	allows you to circumvent some of the restrictive assumptions of the univariate within-subjects ANOVA.
			B)	allows you to include more than two independent variables in your analysis.
			C)	uses separate error terms to test effects rather than a pooled error term.
			D)	none of the above
18		In research situations in which you want to measure or manipulate categorical variables, an appropriate alternative to statistics such as ANOVA, MANOVA, or multiple regression would be
			A)	canonical correlation.
			B)	multiple t tests.
			C)	path analysis.
			D)	multiway frequency analysis.
19		According to your text, the statistic used to evaluate data in a loglinear analysis is
			A)	G².
			B)	F.
			C)	d.
			D)	Chi-square.
20		Path analysis is
			A)	a unique statistical test, allowing you to evaluate multiple dependent variables in one test.
			B)	an application of multiple regression to investigating causal relationships among variables.
			C)	not used to investigate causal relationships, but is a multivariate statistic.
			D)	an extension of the Pearson r to multivariate designs.