GRE Subject Test: Psychology : Measurement & Methodology
Study concepts, example questions & explanations for GRE Subject Test: Psychology
All GRE Subject Test: Psychology Resources
Example Questions
Example Question #41 : Measurement & Methodology
When conducting a two-way analysis of variance, the degrees of freedom (df) for each factor is best expressed as __________.
one less than the number of levels for all factors
one less than the number of levels for that factor
one less than the number of trials for that factor
None of these
one less than the number of participants (n) in each trial
one less than the number of levels for that factor
The df for each independent variable or factor is determined by how many different levels of that factor are included in the analysis. This can create a problem for calculating sums of squares and ultimately for calculating post hoc analyses, as the number of levels in a factor can be very large.
Example Question #42 : Measurement & Methodology
Which statistical procedure is best used when there are two or more independent variables, the variables can be held fixed with respect to each other, and a conditional expectation of dependent variable given the independent variables is required?
Spearman's rank-correlation coefficient
Two-way ANOVA
Analysis of covariance (ANCOVA)
Regression analysis
All of these
Regression analysis
A regression analysis, so long as all but one independent variable at a time can be held fixed, is able to produce a predictive window called a conditional expectation that allows for prediction of dependent variable values given set independent variable levels.
Example Question #43 : Measurement & Methodology
The primary difference between a sequential sum of squares and an adjusted sum of squares is that the adjusted sum of squares __________.
only works for models with small values for n
only works for models with large values for n
does not depend on the order in which factors are entered into the model
depends on the order in which factors are entered into the model
None of these
does not depend on the order in which factors are entered into the model
An adjusted sum of squares is a model that determines the amount of remaining variance in a model is explained by a factor given that all the other factors are already in the model. For a sequential sum of squares, the amount of variance in a model is explained only given the factors which are already in the model. Thus, the first factor inserted in a sequential sum of squares carries the most weight.
Example Question #44 : Measurement & Methodology
All of the following are true about the General Linear Model (GLM) except which of the following?
It generalizes multiple linear regression to more than one dependent variable
It works irrespective of the error terms in each matrix
All of these
It incorporates F-testing, t-testing, linear regression, and ANOVA models
It can test hypotheses using either multivariate or multiple independent univariate tests
It works irrespective of the error terms in each matrix
The general linear model, expressed as 
Example Question #45 : Measurement & Methodology
One appropriate use of a biased ANOVA term for expected mean squares might be which of the following?
If the factors are expected to have an interaction
If the population mean cannot be estimated
If the population means are not equal to one another (the null hypothesis is false)
If one or more terms are nonparametric in the model
None of these
If the population means are not equal to one another (the null hypothesis is false)
A biased EMS is most commonly used when random terms are part of the factors in the model (and thus, when the population means cannot be reasonably assumed to be equal irrespective of treatment effect).
Example Question #46 : Measurement & Methodology
All of the following are true about Pearson's r except which of the following?
Strength of correlation ranges from -1 to 1
All of these
Confidence intervals cannot be established for weak values of r
Sample r is extremely accurate at estimating population correlation coefficients if sample size is large
Sampling distribution can be made to fit Student's t-distribution
Confidence intervals cannot be established for weak values of r
Constructing a confidence interval for Pearson's r can be done through a process called "bootstrapping", which involves re-sampling with replacement a certain number of values in each correlation a large number of times.
Example Question #47 : Measurement & Methodology
Which of the following is a situation where you might prefer a one-tailed test to a two-tailed test?
When the sample size is excessively low
When the sample size is excessively high
When only deviations above a particular score are of research interest
When comparisons between groups are more important then comparisons across groups
None of these
When only deviations above a particular score are of research interest
A two-tailed test differs from a one-tailed test primarily in that it examines deviations both above and below a particular mean. Thus, a one-tailed test is most appropriate when only one direction of deviation is of relevant interest.
Example Question #48 : Measurement & Methodology
Each of the following is true about homoscedasticity except which of the following?
It largely determines goodness of fit (as defined by Pearson's r)
None of these
It follows the assumption that standard deviations of the error terms are constant within measures
Meeting the assumption of homoscedasticity is necessary to demonstrate an unbiased estimate
It can be tested by regression of squared residuals
Meeting the assumption of homoscedasticity is necessary to demonstrate an unbiased estimate
Homoscedasticity is explicitly not required to ensure unbiased, asymptotically normal, or reliable results. They are, however, required to show goodness of fit.
Example Question #49 : Measurement & Methodology
When testing for non-groupwise homogeneity of variance, the most sensitive test to departures from normality is generally which of the following?
None of these
Goldfeld-Quandt test
Levene's test
Bartlett's test
Brown-Forsythe test
Bartlett's test
Levene's test and the Brown-Forsythe test for homogeneity of variance are not very sensitive to departures from normality, which makes them more useful as tests for nonstandard data sets but less able to predict goodness of fit. Goldfeld-Quandt is a test for groupwise homogeneity of variance, and is useless for the stated purpose.
Example Question #50 : Measurement & Methodology
Dr. Crawford believes she has invented a pill to help with student memorization. For a trial experiment, she gathered two groups of twenty students. She gave the twenty students in Group A the “smart pill”, and the twenty students in Group B a sugar pill. After waiting five minutes, both groups of students were given a list of forty words, and were instructed to memorize the entire list in any order. The students were given five minutes to memorize the list. The students were then asked to verbally recite all of the words they could remember in any order within three minutes.
Group A recited an average of fifteen words, while Group B recited an average of ten words.
Which statistical analysis should Dr. Crawford use?
Cannonical correlation
T-test
Chi-square
ANOVA
Pearson-R
T-test
A t-test is most appropriate when only analyzing two variables. Dr. Crawford is analyzing the means between Group A, who received the smart pill, and Group B, who did not. This test will determine whether the differences in means is due to chance or if there is a statistically sound significant difference.
All GRE Subject Test: Psychology Resources
