Foundations of
Research
13. Survey Research Part 2
This is a PowerPoint Show

Open it as a show by going to “slide show”.

Click through it by pressing any key.

Focus & think about each point; do not just
passively click.
© Dr. David J. McKirnan, 2014
The University of Illinois Chicago
McKirnanUIC@gmail.com
Do not use or reproduce without
permission
1
Foundations of
Research
13. Survey Research Part 2
Testing hypotheses
with surveys.
2
Foundations of
Research
3
Alcohol: Knowledge 1
What is the chemical in
alcohol?
Let’s do our own practice survey!
a) ethanol
b) zymase
c) dehydrogenase
d) nicotine
Shutterstock
Foundations of
Research
4
Alcohol: Knowledge 2
Which of these drinks
contains the least
alcohol?
A = 12 ounces of beer
B = 5 ounces of wine
C = 1 1/2 ounces of vodka
D = 8 ounces of malt liquor
E = All are equal
Shutterstock
Foundations of
Research
5
Alcohol: Knowledge 3
How long does it take for
alcohol to affect the
brain?
a) 10 seconds
b) 90 seconds
c) 10 minutes
d) 90 minutes
Shutterstock
Foundations of
Research
Alcohol attitudes, 1
6
Drinking helps me feel
whatever way I want to
feel.
A = agree strongly
B = agree
C = somewhat agree
D = slightly agree
E = do not agree
Shutterstock
Foundations of
Research
Alcohol attitudes, 2
7
Alcohol makes me feel
happy.
A = agree strongly
B = agree
C = somewhat agree
D = slightly agree
E = do not agree
Shutterstock
Foundations of
Research
Alcohol attitudes, 3
8
Drinking adds a certain
warmth to social
occasions.
A = agree strongly
B = agree
C = somewhat agree
D = slightly agree
E = do not agree
Shutterstock
Foundations of
Research
Alcohol behavior, 1
9
How often do you have a
drink containing
alcohol?
A = never
B = monthly or less
C = 2 to 4 times a month
D = 2 to 3 times a week
E = 4 or more times a week
Shutterstock
Foundations of
Research
Alcohol behavior, 2
10
How many drinks
containing alcohol do
you have on a typical
day when you are
drinking?
A=1
B=2
C=3
D = 4 or 5
E = 6 or more
Shutterstock
Foundations of
Research
Alcohol behavior, 3
11
How often during the last
year have you found that
you were not able to stop
drinking once you had
started?
A = never
B = less than monthly
C = monthly
D = weekly
E = daily or almost every day
Shutterstock
Foundations of
Research
Alcohol behavior, 4
12
How often during the last
year have you failed to do
what was normally
expected from you
because of drinking?
A = never
B = less than monthly
C = monthly
D = weekly
E = daily or almost every day
Shutterstock
Foundations of
Research
Knowledge / attitude / behavior composites
Knowledge
 What is the chemical in alcohol
Here is our survey.
13
 Several items assess each
Hypothetical Construct we are
studying.
 Which of these drinks contains the least alcohol?
 How long does it take for alcohol to affect the brain?
Attitudes and beliefs
 Drinking helps me feel whatever way I want to feel.
 Alcohol makes me feel happy.
 Drinking adds a certain warmth to social occasions.
Behavior
 How often do you have a drink containing alcohol?
 How many drinks containing alcohol do you have on a typical
day when you are drinking?
 How often during the last year have you found that you were
not able to stop drinking once you had started?
 How often during the last year have you failed to do what was
normally expected from you because of drinking?
Foundations of
Research
Knowledge / attitude / behavior composites
Knowledge
Here is our survey.
 Several items assess each
Hypothetical Construct.
 What is the chemical in alcohol
 Typically we also collect
 Which of these drinks contains the leastdemographics
alcohol?
(age, gender,
ethnicity…)
and other Ψ
 How long does it take for alcohol to affect
the brain?
variables.
Attitudes and beliefs
 We choose items based on our
 Drinking helps me feel whatever way I want
to feel.
Theory
about alcohol use and
problems.
 Alcohol makes me feel happy.
 We
can simply Describe our
 Drinking adds a certain warmth to social
occasions.
data in each category.
Behavior
 We might Block the data by
 How often do you have a drink containing
alcohol?
important
categories of
 How many drinks containing alcohol doparticipants
you have on a typical day
How when
manyyou
students
use alcohol at
are drinking?
 Or we can Test Hypotheses
all?
 men
How have
often aduring
last year have youabout
foundcorrelations
that you were
not
among
Do
more the
positive
able totoward
stop drinking
variables
attitude
alcoholonce
than you had started?
women?
How often during the last year have you failed to do what was
normally expected from you because of drinking?
14
Foundations of
Research
Knowledge
Knowledge / attitude / behavior composites
15
Research questions:
 How well do the items within each category
 What is the chemical correlate?
in alcohol
 Which of these drinks
least alcohol?
• contains
Are they the
internally
Reliable?
 How long does it take for alcohol to affect the brain?
• Can we construct a Composite Variable from
each item set?
Attitudes and beliefs
Blocking
: Gender?
 Drinking helps me 
feel
whateverVariables
way I want
to feel. Ethnicity? Age?...
 Alcohol makes mefeel
happy.
What
are the Correlations among composite
variables?
 Drinking adds a certain warmth to social occasions.
• How much are knowledge or attitudes & beliefs
Behavior
associated with alcohol use?
 How often do you have
a drink
containing
• Are
alcohol
use andalcohol?
problems correlated?
 How many drinks containing
alcohol
do you
have on
typical
day
• Are these
effects
Mediated
byaother
variables?
when you are drinking?
 Do depression, isolation, or stress make people
 How often during the last year
have you
founduse
thatoryou
were not
vulnerable
to alcohol
problems?
able to stop drinking once you had started?
 How often during the last year have you failed to do what was
normally expected from you because of drinking?
Foundations of
Research
Class data, n = 97
Multiple items assessed the Hypothetical Constructs of
Alcohol- related information, attitudes, use, and problems.
 Information: # of correct answers for items 1  3.
 Attitudes: items 4, 5 & 6.
 Use: drinking frequency * # drinks / occasion (items 7 & 8).
Example
 Problems: Items 9 & 10.
Information:
Mode = 2 items correct.
All figures & tables: David McKirnan
16
Foundations of
Research
Class survey data, Attitudes
 To test weather the 3 attitude items comprised a single
Hypothetical Construct, we examine their
intercorrelations
Example
 All the correlations and high and statistically significant,
so we can combine them into a single composite variable.
17
Foundations of
Research
Class survey data, Attitudes
 The frequency distribution for the composite attitude
variable shows an approximate normal distribution.
Example
 The Mean and Median are
equal, near the center of the
distribution
 The mode reflects a larger
block of participants with
moderately strong attitudes.
18
Foundations of
Research
Class survey data, Alcohol problems
 We perform the same operations for items 9 & 10, the alcohol
problems measures.
 The correlation between these items is very high.
Example
 We can validly combine them into one variable.
19
Foundations of
Research
Class survey data, Alcohol problems
 Most people report no problems, as shown in the Mean and
Median
 The rest of the distribution shows a strong positive skew.
Example
 This skew pulls the Mean higher than the Mode & Median.
20
Foundations of
Research
Class survey data, Correlation results
 The amount of information people have about alcohol is
unrelated to their alcohol use or problems
 The correlations are low
 Neither is statistically significant.
Example
 Positive attitudes toward alcohol are strongly associated with
alcohol use.
 … and moderately (statistically significantly) with alcohol problems
21
Foundations of
Research
22
Correlations and mediators
With more data we could use mediating analyses to develop
to test theories about alcohol use / problems

Say we find that people who are lonely or depressed (negative
affect) have more alcohol use & problems.

Why might this be?

People with negative affect may come to expect (hope?) that
alcohol will make them feel better.

The attitude may be what actually leads to alcohol use.

Here attitudes / beliefs would mediate the effect of negative
affect on alcohol use & problems.
Depression
loneliness
stress
Positive
attitudes
toward
alcohol
Alcohol
use &
Problems
Foundations of
Research
23
Correlations and mediators
Using mediating analyses to develop to test theories about
alcohol use / problems

Say we find that people who
are lonely
or depressed
Using
a mediating
analysis(negative
to ask
affect) have more alcohol use
& problems.
“why”
is the core step in Theory

Why might this be?
Development.

People with negative affect may come to expect (hope?) that
alcohol will make them feel better.

The attitude may be what actually leads to alcohol use.

Here attitudes / beliefs would mediate the effect of negative
affect on alcohol use & problems.
Depression
loneliness
stress
Positive
attitudes
toward
alcohol
Alcohol
use &
Problems
Foundations of
Research
24
Correlations and mediators
Survey data are important for developing or testing new
theories of behavior

Can have high external validity

Can model different relations among variables.
Problems?

The data are correlational and cross-sectional.

The causal arrows may go in different directions.
Depression
loneliness
stress
Positive
expectation
of alcohol
Alcohol
use &
Problems
Foundations of
Research
25
Correlations and mediators
Survey data are important for developing or testing new
theories of behavior

Can have high externalGenetics:
validity

Can model different relations
Positiveamong variables.
alcohol effects
Problems?
& negative
affectivity

The data are correlational
and cross-sectional.

The causal arrows may go in different directions.

There may be a third variable that we did not measure.
Depression
loneliness
stress
Positive
expectation
of alcohol
Alcohol
use &
Problems
Foundations of
Research
cures
Cures for correlation problems?

Longitudinal studies


Examine changes over time to
better determine cause and effect
Experimental studies

Simulate and manipulate the
predictors, e.g.;
 Induce temporary stress /
negative affect
 Assess attitudes
 Assess alcohol use
26
Foundations of
Research
27
We Use Longitudinal data
 To examine change
 To approximate “cause and effect”

Research has shown that
depression, loneliness &
stress are correlated.

So, we combine them into a
single composite variable.

This is our Predictor
Depression
Loneliness
Negative
Affect
Stress
(Measurement version of IV)


The same for alcohol use
and problems.
This is our outcome
(Measurement version of DV)
Alcohol
use
Alcohol
problems
Alcohol
abuse
Foundations of
Research


28
Longitudinal data: Path analysis
We test the hypothesis that Negative Affect
correlate with alcohol abuse & problems…

…at baseline

…and at follow-up
Data at each time frame
are cross-sectional:
 We cannot determine if
emotions “cause” the
alcohol outcome, or
visa versa.
Time 1
Time 2
(baseline)
(follow-up)
Negative
Affect
Negative
Affect
Alcohol
abuse
Alcohol
abuse
Foundations of
Research

29
Longitudinal data: Path analysis
Looking at patterns across time lets us make more sensitive
statements.
Time 1
(baseline)
If the measures are reliable
Negative Affect at Time 1 will
correlate with NA at Time 2.
Time 2
(follow-up)
Negative
Affect
Negative
Affect
And the same for the alcohol
variable
Alcohol
abuse
Alcohol
abuse
Foundations of
Research

30
Longitudinal data: Path analysis
To test the hypothesis we look at the “cross” correlations…
Time 1
(baseline)
How strongly does
Affect at baseline
correlate with Alcohol at
follow-up?
Negative
Affect
Alcohol
abuse
Time 2
(follow-up)
Negative
Affect
A strong correlation would
tell us that Affect does
lead to more Alcohol
abuse over time.
Alcohol
abuse
Foundations of
Research

31
Longitudinal data: Path analysis
To test the hypothesis we look at the “cross” correlations…
Time 1
(baseline)
Similarly, how strongly
does Alcohol at T1
correlate with changes
in Affect at T2?
Negative
Affect
Alcohol
abuse
Time 2
(follow-up)
Negative
Affect
This correlation tells us
how much alcohol use
leads to negative affect
over time.
Alcohol
abuse
Foundations of
Research

32
Longitudinal data: Path analysis
Testing the hypothesis: does Negative Affect ‘cause’ Alcohol
use & problems
Time 1
(baseline)
Negative
Affect
Alcohol
abuse
If the correlation
between Affect at Time 1
and Alcohol at follow-up
is very strong…
Time 2
(follow-up)
Negative
Affect
Alcohol
abuse
Foundations of
Research

33
Longitudinal data: Path analysis
Testing the hypothesis: does Negative Affect ‘cause’ Alcohol
use & problems
Time 1
(baseline)
Negative
Affect
Alcohol
abuse
…and the correlation
between Alcohol at T1
and Affect at T2 is
relatively weak…
Time 2
(follow-up)
Negative
Affect
Alcohol
abuse
Foundations of
Research

34
Longitudinal data: Path analysis
Testing the hypothesis

Time 1
(baseline)
Negative
Affect
Alcohol
abuse
…we can conclude
that Affect may be a
more important
“cause” of alcohol
use / problems than
the other way around.
Time 2
(follow-up)
Negative
Affect
Alcohol
abuse
Foundations of
Research

35
Longitudinal data: Path analysis
Testing the hypothesis
Time 1
(baseline)
Of course we may find
that Alcohol “causes”
Affect, not the other way
around…
Negative
Affect
Alcohol
abuse
Time 2
(follow-up)
Negative
Affect
Either way, we have a
better sense of cause
and effect than with
cross-sectional data.
Alcohol
abuse
Foundations of
Research

Surveys typically use multiple items to measure
each hypothetical construct

Summary



Summary: Testing Hypotheses
Correlations among items tell us if they are reliable in
measuring the same construct.
We use Mediating Analyses to

Test hypotheses about correlations between constructs

Build or test theory
Cross-sectional analyses are difficult to interpret

Causal direction?

3rd variable problem
Longitudinal analyses help us determine causal
direction
36