EPPP Practice Questions: Statistics & Research Design

Question 1

In a positively skewed distribution, from smallest to largest in magnitude, the measures
of central tendency are:
a. mean, median, mode.
b. mean, mode, median.
c. mode, median, mean.
d. mode, mean, median.

Answer 1

c. mode, median, mean.
Remembering that the “tail tells the tale” and that the median is always the middle score
would have helped you identify the correct answer to this question: In a positively skewed
distribution, the mode has the smallest value, the median has the middle value, and the mean
(which is affected most by the extreme scores in the positive tail) has the largest value.

Question 2

The Central Limit Theorem predicts that a sampling distribution of the mean will
increasingly approach a normal shape:
a. regardless of the shape of the population distribution as the sample size increases.
b. regardless of the shape of the population distribution as the number of samples
increases.
c. only when the population distribution does not deviate from the normal.
d. only when the sample distributions do not deviate significantly from the normal.

Answer 2

a. regardless of the shape of the population distribution as the sample size increases
This is the correct description of the prediction of the Central Limit Theorem. Note that it is
the size of the sample that is the critical factor. (The Central Limit Theorem always assumes
an infinite number of samples.)

Question 3

The probability of making a Type I error is increased by:

a. conducting a single multivariate test rather than several univariate tests.
b. changing the level of significance from .01 to .05.
c. changing beta from .01 to .05.
d. conducting a two-tailed (rather than a one-tailed) test.

Answer 3

b. changing the level of significance from .01 to .05
The probability of making a Type I error is directly affected by the level of significance (alpha).
The larger the level of significance, the greater the likelihood of a Type I error.

Question 4

“Sampling error” is due to:

a. the unreliability of the test.
b. the invalidity of the test.
c. random factors that produce a nonrepresentative sample.
d. non-random factors that produce a nonrepresentative sample.

Answer 4

c. random factors that produce a nonrepresentative sample
As the result of random (chance) factors, the sample may not “mirror” the population in
important characteristics. This is referred to as sampling error.

Question 5

To increase power, you would:

a. reduce alpha.
b. increase beta.
c. reduce sample size.
d. increase sample size.

Answer 5

d. increase sample size
Increasing sample size makes it easier to statistically detect the effects of the independent
variable; i.e., it increases power. Other ways to increase power include increasing the
magnitude of alpha, using a reliable dependent variable measure, and using a parametric
statistical test.

Question 6

To calculate the F-ratio for a one-way ANOVA:

a. MSB is multiplied by MSW.
b. MSB is divided by MSW.
c. MSW is divided by MSB.
d. MSW is subtracted from MSB.

Answer 6

b. MSB is divided by MSW.
The F-ratio is calculated by dividing a measure of treatment plus error by a measure of error
only in order to obtain an estimate of treatment effects. MSB (mean square between) is a
measure of treatment plus error, while MSW (mean square within) is a measure of error only,
and the F-ratio is calculated by dividing MSB by MSW.

Question 7

For her math class assignment, a high school student will compare the average number
of hours that students in her school spend playing computer games each day to the
national average. To analyze the data she collects, the student will use which of the
following statistical tests?
a. single-sample t-test
b. independent samples t-test
c. single-sample chi-square test
d. multiple-sample chi-square test

Answer 7

a. single-sample t-test
The t-test is used to compare two means, and when one of the means is a sample mean and
the other is a known population mean, the appropriate t-test is the single-sample t-test.

Question 8

An investigator wants to test the hypothesis that the average number of aggressive acts
that children exhibit in an unfamiliar situation is related to gender and sociability
(“sociable” or “shy”). He obtains a sample of 30 boys and 30 girls who have been rated as
either sociable or shy and then has observers count the number of aggressive acts each
child exhibits in an unfamiliar situation during a 30-minute play period. The best
statistical test to analyze the data the investigator collects in this study is which of the
following?
a. t-test for independent samples
b. chi-square test
c. one-way ANOVA
d. two-way ANOVA

Answer 8

d. two-way ANOVA
The research study involves two independent variables (sociability and gender) and a single
dependent variable that is measured on a ratio scale (number of aggressive acts). The
appropriate statistical test is the two-way analysis of variance.

Question 9

A researcher uses a factorial ANOVA to statistically analyze the effects of four types of
training strategies and three levels of self-efficacy on a measure of job performance. The
results indicate that there are significant main effects of each independent variable and a
significant interaction. The researcher will conclude that:
a. training is effective regardless of level of self-efficacy.
b. each type of training is equally effective for each level of self-efficacy.
c. the most effective type of training depends on level of self-efficacy.
d. overall, training is effective only for people with a certain level of self-efficacy.

Answer 9

c. the most effective type of training depends on level of self-efficacy
Because the interaction is significant, this means that the effects of the types of training may
differ for different levels of self-efficacy – e.g., training method #1 might be most effective for
people with low self-efficacy, while training method #2 might be most effective for people with
high self-efficacy.

Question 10

Which of the following provides a measure of shared variability and is calculated by
squaring the correlation coefficient?
a. coefficient of concordance
b. kappa statistic
c. coefficient of determination
d. eta

Answer 10

c. coefficient of determination
The coefficient of determination is calculated by squaring the correlation coefficient (r) and
provides a measure of shared variability, or the amount of variability in one variable that is
explained by variability in another variable

Question 11

Which of the following correlation coefficients is useful for determining the degree of
association between two nominal variables?
a. biserial
b. contingency
c. Spearman
d. Pearson

Answer 11

b. contingency
The contingency coefficient is used to determine the correlation between two nominal
(categorical) variables – for example, between a parent’s parenting style (authoritative,
authoritarian, permissive, or rejecting/neglecting) and his/her educational achievement (high
school graduate, undergraduate degree, master’s degree, doctoral degree).

Question 12

An organizational psychologist is hired by a company to help identify job applicants
who should be considered for managerial level positions. In this situation, the psychologist
wants to develop a selection test battery that will provide the following information: “Which
group (manager or non-manager) does a job applicant most closely resemble?” Which of
the following methods for combining selection test scores will be most useful for answering
this question?
a. multiple regression
b. discriminant function analysis
c. factor analysis
d. canonical correlation

Answer 12

b. discriminant function analysis
There are several methods for combining predictor scores to estimate performance on a
criterion. When the criterion is group membership (i.e., when the criterion is measured on a
nominal scale), discriminant function analysis is an appropriate method.

Question 13

External validity is likely to be a problem when:
a. the results of a research study indicate that a test’s validity coefficient differs for
members of different groups.
b. an investigator uses a nonparametric statistical test to analyze the data she collected
in a research study.
c. the research study conducted by an investigator does not include a no-treatment
control (comparison) group.
d. the research design used by an investigator involves sequentially administering the
different levels of the IV to the same group of participants.

Answer 13

d. the research design used by an investigator involves sequentially administering the
different levels of the IV to the same group of participants
External validity refers to the generalizability of research results to different participants,
settings, and conditions. This external validity of the research study described in this answer
is threatened by multiple treatment interference because it involves exposing all participants
different levels of the IV. Consequently, it will not be possible to generalize the results to
situations in which individuals receive only one level of the IV.

Question 14

When using an ABAB design, you are:

a. administering two different treatments at two different times.
b. administering one treatment at two different times.
c. administering one treatment to two different groups.
d. administering one treatment to two different behaviors.

Answer 14

b. administering one treatment at two different times
The ABAB design has two no treatment (A) phases and two treatment (B) phases. The same
treatment is administered during the B phases.