4: Association claims

Question 1

interrogating construct validity

Answer 1

-measure’s reliability:
• Test-retest reliability
• Inter-rater
• Internal reliability

-Subjective evidence for the measure’s construct validity:
:face or content validity

-Empirical evidence of the measure’s construct validity:
.criterion or
.convergent and divergent

Question 2

interrogating stat validity

Answer 2

1. What is the effect size?
2. Is the correlation statistically significant?
3. Could outliers be affecting the association?
4. Is there restriction of range?
5. Is the association curvilinear?

Question 3

effect size

exceptions to large effect sizes

Answer 3

-Stronger associations = strong effect
sizes (larger r)
...permit more accurate predictions
...more important
...more stat sig

small effect sizes can translate to saving lives in a med context

Question 4

stat siginifact

Answer 4

-probability of obtaining an
association of that strength by chance alone
-p-value

-less than alpha (usually α = .05) is considered to be
statistically significant

-dependent upon both effect size and sample size

Question 5

distort the strength
of association and might result in false positives or
false negatives:

Answer 5

• Outliers
• Restriction of range
• Curvilinear relationship

Question 6

false positives

Answer 6

-Value of r without outlier
at BAS-RR of 11 is
much lower

• The r-value is sensitive to
  the presence of an outlier
• Original r-value largely
  driven by the outlier

Question 7

false negtives

Answer 7

• Value of r for complete dataset: r = .40

- Value of r with outlier removed is much higher: r = .70

Question 8

restriction of range

Answer 8

-might make a sample’s correlation

appear smaller than it really is in the population

Question 9

curvilinear

Answer 9

-inverted U-shape

-Pearson r will yield a small,
nonsignificant value in this case (may
be a false negative)

-the r will be close to zero, even though there is a relationship

-Must use other techniques to fit
nonlinear data

Question 10

interrogating external validity

Answer 10

-Does the association generalize to other people, places and times?
• How the sample was selected

-Nonrandom sampling may be sufficient if population
validity is not a priority

Question 11

Why is it invalid to draw casual conclusions from

correlational research designs?

Answer 11

-Directionality problem (no temporal precedence)
• Even if causal, not always possible to know the direction of causation (A may cause B, or B may cause A, or reciprocal)

-Third-variable problem (poor internal validity)
• Changes in another unmeasured third variable may
actually cause the values of A and B to co-vary

Question 12

matching characteristics of pre-existing groups

Answer 12

-compares a pre-existing group to a control group on some measure

-Measure data defining group membership
• Group membership may be defined based on some threshold value of
collected quantitative data
• Risk for many extraneous variables differing between pre-existing groups

-Matched control group reduces possible third-variable explanations by roughly equating groups on some extraneous variables
•reduce influences of extraneous variables by using a control group with certain characteristics matched to the pre-existing group

Question 13

when to use correlational research

Answer 13

-When gathering data in the early stages of research

-When manipulating an independent variable is impossible, impractical, or unethical
• When you are relating two or more naturally occurring variables
• Differences between pre-existing groups
• Manipulation is unduly intrusive, difficult, or otherwise impractical
• Manipulation is unfair or puts participants at undue risk of harm

Question 14

why non-experimental

Answer 14

• Many candidate predictor and/or criterion variables to explore
• Predictor variable is impossible to manipulate
• Predictor variable is difficult or impractical to manipulate
• Predictor variable is unethical to manipulate
• May suggest future experimental studies to establish causation