Deck 12: Statistics in Perspective

Question 1

The first step in analyzing data in a comparison group study is to:

A) calculate averages
B) construct frequency polygons
C) calculate measures of variability
D) determine statistical significance

Answer 1

construct frequency polygons

Question 2

A useful means of evaluating obtained differences among averages is by:

A) comparison to averages of previously identified groups
B) examining the size of the groups
C) converting them to percentiles
D) calculating the magnitude of t

Answer 2

comparison to averages of previously identified groups

Question 3

Which of the following is not a danger in using inferential statistics to evaluate differences in averages?

A) Important differences may be ignored.
B) Resulting probabilities are likely to be in error.
C) Complex calculations are likely to contain errors.
D) Large samples can make trivial differences appear important.

Answer 3

Complex calculations are likely to contain errors.

Question 4

One commonly used index of effect size is obtained by:

A) dividing the difference between means by the standard error of measurement
B) dividing the difference between means by the standard error of the mean
C) dividing the difference between means by the standard deviation of the control group
D) none of the above

Question 5

In a study comparing class size to mathematics achievement, the unit of analysis is:

A) the individual student
B) the class
C) the teacher
D) none of the above

Question 6

It is generally recommended that an effect size (E.S.) be of what magnitude before concluding that a difference in means is important?

A) .25
B) .50
C) 1.00
D) 2.00

Question 7

Scatterplots are recommended in addition to correlation coefficients because they:

A) can correct errors in calculating
B) provide a pictorial rather than a numerical assessment
C) indicate which average should be used
D) provide more information

Question 8

Which of the following statements is correct?

A) More than one Pearson correlation coefficient can be calculated from the same scatterplot.
B) Only one Pearson correlation coefficient can be calculated from one scatterplot.
C) A scatterplot will show a higher relationship than the correlation coefficient calculated from it.
D) A scatterplot will show a lower relationship than the correlation coefficient calculated from it.

Question 9

In practice, most useful correlations between different variables are between:

A) .00 and .30
B) .40 and .70
C) .70 and .90
D) over .90

Question 10

If eta and r are in substantial disagreement:

A) eta should be used
B) r should be used
C) neither should be used
D) an error in calculation has been made

Question 11

When in doubt as to whether parametric or nonparametric statistics are appropriate, a researcher should:

A) report only parametric statistics
B) report only nonparametric statistics
C) report both
D) use neither

Question 12

The most appropriate indicator of the importance of a relationship between two categorical variables is:

A) the crossbreak table
B) Chi-Square
C) a comparison of actual and expected frequencies
D) the contingency coefficient

Question 13

In a study in which all subjects receive both treatments, the appropriate descriptive statistic to use is:

A) the Pearson correlation coefficient
B) the mean
C) the contingency coefficient
D) Chi-Square

Question 14

The importance of a 15-point difference in means between two groups:

A) depends on the spread of scores in each group
B) depends on the shape of the frequency polygons
C) depends on the number of cases in each group
D) all of the above

Question 15

Calculation of Pearson r is most justified when the points on a scatterplot:

A) approximate a triangular shape
B) contain a few points that do not fit the overall pattern
C) fall into two distinct groups
D) cluster around a diagonal line

Question 16

Confidence intervals are recommended rather than significance tests because they:

A) provide more information
B) are easier to calculate
C) require fewer mathematical assumptions
D) do not require as large a sample

Question 17

When random sampling was not used, probabilities obtained from inferential statistical procedures should:

A) not be reported
B) be treated only as crude indices
C) be reported only if less than .01
D) be used as the primary index of the importance of results

Question 18

If inferential statistics are calculated, they should be obtained:

A) as the first step in data analysis
B) as part of a comprehensive assessment of results
C) only after demonstration that the magnitude of the obtained relationship is worthy of attention
D) as part of the determination of effect size

Question 19

In a comparison group study using two randomly selected groups, effect size = .24, and p<.01. This means the:

A) difference in means is small and probably due to sampling error
B) difference in means is small but probably is not due to sampling error
C) difference in means is large enough to be important, but may well be due to sampling error
D) difference in means is large and probably not due to sampling error

Question 20

A researcher discussing the information in item 19 could be criticized for:

A) making too much of a trivial relationship
B) incorrect use of inferential statistics
C) an error in computation
D) all of the above

Question 21

Matching

-Categorical variable

A) delta
B) Pearson r
C) eta.
D) IQ scores
E) gender

Question 22

Matching

-Straight-line relationship

A) delta
B) Pearson r
C) eta.
D) IQ scores
E) gender

Question 23

Matching

-Curvilinear relationship

A) delta
B) Pearson r
C) eta.
D) IQ scores
E) gender

Question 24

Matching

-Quantitative variable

A) delta
B) Pearson r
C) eta.
D) IQ scores
E) gender

Question 25

Matching

-Effect size

A) delta
B) Pearson r
C) eta.
D) IQ scores
E) gender

Question 26

Gain scores can be obtained by subtracting pretest scores from posttest scores.

Question 27

Most researchers consider that any delta of .50 or larger is an important finding.

Question 28

A Pearson r of -.90 indicates a very weak relationship.

Question 29

Scatterplots provide information on which correlation coefficient to calculate.

Question 30

A correlation coefficient of r = .05 can be interpreted as having both practical and statistical significance.