Foundations of
Research
Developing research questions and hypotheses.
This is a PowerPoint Show

Click “slide show” to start it.

Click through it by pressing any key.

Focus & think about each point; do not
just passively click.

To print:

Click “File” then “Print…”.

Under “print what” click “handouts (6
slides per page)”.
Phenomenon
Theory
Hypothesis
Methods
& data
Results
© Dr. David J. McKirnan, 2014
The University of Illinois Chicago
McKirnanUIC@gmail.com
Do not use or reproduce without
permission
Discussion &
Conclusions
1
Developing research questions and hypotheses.
Foundations of
Research
 How do social values affect

science?

Where do research questions &
hypotheses come from.

Variables in research
2
Foundations of Social Values…
Research
3
Phenomenon
Theory
…affect what we choose
as our research question…
Hypothesis
Methods
& data
Results
Discussion &
Conclusions
Specific methods are
more standard and
objective
…and our conclusions
Foundations of Values, theory and data in the scientific process.
Research
4
 Social
Phenomenon
Theory
Hypothesis
Methods
& data
Results
Discussion &
Conclusions
values help define a
scientific “problem” or question.
 Norms,
values (& data) determine
what is credible / fundable.
 Theory
is influenced by norms +
empirical background of field.
 Science
hinges on clear,
objectively stated hypotheses.
 Clear
hypotheses lessen bias in
interpreting results.
 Methods
& analyses are most
objective, but fields vary in
methodological rigor.
 The
“meaning” of a finding is
influenced by cultural & social
values or concerns.
…for science and, particularly,
for society.
Foundations of
Research
Values and science: The internet and sexual risk.
5
 Social
Phenomenon
values help define a
scientific “problem” or question.
 Norms,
values (& data) determine
what is credible / fundable.
Theory
Hypothesis
Sexual internet use became a large
issue for gay/bisexual men in the
2000s
 Internet use exploded, and became a
Methods
& data
sexual venue.
 Gay men who found partners on the
internet were particularly risky.
Results
Discussion &
Conclusions
 The HIV prevention field needed to
understand how this worked.
 HIV made research on gay men and
the internet a legitimate and
important topic
Foundations of Values and science.
Research
6
Phenomenon
Theory
Hypothesis
Methods
& data
Results
Discussion &
Conclusions
 Theory
is influenced by values &
empirical background of field.
Several theories may help us explain
the effect of internet use on sexual
risk:
 Theories of depression: a Ψ factor
within the person may interfere with
decision making.
 Social structure: The social
environment of MSM may influence
internet use & risk.
Taking an individual v. social
perspective can be influenced by a
researcher’s values.
Foundations of Values and science
Research
Phenomenon
Theory
Hypothesis
Methods
& data
Results
Discussion &
Conclusions
7
 Science
hinges on clear,
objectively stated hypotheses.
 Clear
hypotheses lessen bias in
interpreting results.
We may test a depression theory by
hypothesizing that:
 People who are depressed are more
likely to find partners through the
internet than face-to-face, and
 Depressed people have more difficultly
negotiating for themselves.
 So, rather than using the internet to find
partners causing greater risk…
Internet
partners
Sexual
risk
Foundations of Values and science
Research
Phenomenon
Theory
Hypothesis
8
 Science
hinges on clear,
objectively stated hypotheses.
 Clear
hypotheses lessen bias in
interpreting results.
 Depression may lead people to find
partners on the internet and engage in
greater risk.
Methods
& data
Depression
Results
Discussion &
Conclusions
Internet
partners
Sexual
risk
Foundations of Values and science.
Research
Phenomenon
Theory
Hypothesis
Methods
& data
Results
Discussion &
Conclusions
9
Widely shared conventions make
value judgments less important.
However:
 Scientific fields – particularly
behavioral science – vary
considerably in their rigor.
 Choosing, e.g., quantitative vs.
qualitative research can be a value
choice.
 Fields such as literary criticism,
history or feminist studies may use
substantially different methods.
 Methods & analyses are most
objective, but fields vary in
methodological rigor.
Foundations of Values and science.
Research
10
No scientific finding can completely
answer a research question;
Phenomenon
 Its meaning is affected by existing
theory and empirical findings.
Theory
 Society may not be “ready” for
certain findings: e.g., data showing
that GLBT people are not “mentally ill”
were rejected for many years.
Hypothesis
Methods
& data
Results
Discussion &
Conclusions
 Many important findings
e.g. from economics
have little affect on social policy if
they contradict a widely shared
ideology.
e.g. austerity approaches to solving
recessions
 The
“meaning” of a finding is
influenced by cultural & social
values or concerns.
…for science and, particularly,
for society.
Developing research questions and hypotheses.
Foundations of
Research

How do social values affect science?

 Where do research
questions & hypotheses
come from.

Variables in research
11
Foundations of
Research
Research questions
Where do research questions come from?
 Practical questions
 Unanswered questions from previous research
 Testing theories.
12
Foundations of
Research
Sources of research questions
Practical / applied questions
Describe or explain an important physical or social
process
 Evaluate an intervention or policy change
EXAM PLE

Chicago closed a large
number of schools and
reassigned students.
This was predicted to
result in better scores.
Did it, over 5 years of
follow-up?
Click for Chicago Tribune article
13
Foundations of
Research
Example of a practical research question and design
HIV testing of infants plus early treatment for those found
to be infected can greatly decrease infant mortality.
EXAM PLE

Getting mothers or caregivers to test infants can be difficult,
particularly in developing counties.

Many such settings have…
 Poor medial infrastructure
 High stigma associated with HIV
 Fear / uncertainty regarding Western medicine.

Many mothers are, however, willing to attend immunization
clinics.

Could those clinics be useful to infant HIV testing?
14
Foundations of
Research
15
EXAM PLE
Universal HIV testing of infants at immunization clinics.

Objective: Determine the acceptability and feasibility of universal HIV
testing of 6-week-old infants attending immunization clinics.

Design: An observational cohort with intervention.

Methods: Infant HIV testing offered to all mothers bringing infants for
immunizations at clinics in KwaZulu Natal. Blood samples were collected
by heel prick. Exit interviews were requested irrespective of whether
mothers agreed to infant testing or not.

Results:
 Of 646 mothers, 584 (90.4%) agreed to HIV testing of their infant and 332
(56.8%) subsequently returned for results.
 332 of 646 (51.4%) mothers and infants thereby had their HIV status
confirmed or reaffirmed by the time the infant was 3 months of age.
 The majority of mothers said they were comfortable with testing of their
infant at immunization clinics and would recommend it to others.

Conclusion: Screening of all infants at immunization clinics is acceptable
and feasible as a means for early identification of HIV-infected infants and
referral for antiretroviral therapy.
Rollins, Nigel; Mzolo, Similo; Moodley, Tammy; Esterhuizen, Tonya; van Rooyen, Heidi. Universal HIV testing of infants at immunization clinics: an
acceptable and feasible approach for early infant diagnosis in high HIV prevalence settings. AIDS, Volume 23(14), 10 September 2009, p 1851–1857
Foundations of
Research
Sources of research questions
Practical / applied questions
Unanswered questions from previous research

Clarify conflicting / unclear findings


Who actually uses the internet to meet people…
Do previous findings generalize to…


…different groups

Many Psychology studies enroll middle class White female
undergraduates in research labs.

Do those findings generalize to general community samples?
…different research areas


Can interventions to increase healthy behavior generalize to
recycling and energy conservation?
…different research approaches

Do controlled lab studies generalize to less controlled field
research?
16
Foundations of
Research
Theories
Practical / applied questions
Unanswered questions from previous research
Testing theories

Use existing theory to explain a new phenomenon


Test contrasting theories of a phenomenon


“Sensation seeking” personality is associated with drug use. Might
it also explain “over-exercising”?
How much is adolescent problem behavior controlled by
psychological variables (depression) vs. peer influence?
Develop new / expanded theory

We discriminate among very subtle differences in smell. Might
olfactory cues affect who we are attracted to?
17
Foundations of
Research
Theories
18
How do we go from a research question to
an actual study?
 Phenomenon
 Theory
 Hypothesis
 Methods
Key Terms:
 Hypothetical
Construct
 Operational
Definition
Foundations of
Research
The research process
19
Phenomenon
Overall issue or question;
What controls emotional states?
Why are some people vulnerable to depression?
Theory
Possible explanation: “How it works” statement
Several theories may help explain the phenomenon
Theory 1
Theory 2
Emotional stability requires
secure emotional
attachments.
Some brains are genetically
disposed to serotonin
depletion during stress
Foundations of
Research
Research process, 2
20
Theory 1
Emotional attachment  emotional stability.
A theory can lead to several hypotheses
Hypothesis 1
Fewer parent – child interactions
 vulnerability to depression.
Hypothesis 2
Emotional support during stress
 less depression
A given hypotheses can be tested in several ways
Methods 1
Survey measurement: assess # of
“family meals” per week,
correlate it with self-reported
depression.
Methods 2
Experiment: Randomly assign
children to have / not have
structured parent & child
interactions. Induce stress to
both groups, assess depression
Foundations of
Research
Research process, 3
21
Theory 1
Emotional attachment  emotional stability.
Hypothesis 1
Family interactions  depression.
Hypothesis 2
(Non)Support + stress  depression
Some hypotheses are best tested in a
measurement approach, and some with
experimental designs
Best tested by a measurement
study
 Family interactions are
difficult to bring into the lab,
 Possible ethical problems.
Can be tested in an experiment:
Both support and stress can be
controlled or manipulated in
the lab.
Foundations of
Research
Research process:
The Big Picture
Phenomenon
Big picture question.
Theory 1
Possible explanation, invoking
one set of hypothetical
constructs.
Hypothesis 1
A prediction that logically flows
from – and tests – the theory.
Methods 1
Operationally define the variables
& test the hypothesis.
Theory 2
Alternate explanation, invoking
other hypothetical constructs.
Hypothesis 2
Another prediction that tests the
same theory.
Methods 2
An alternate operational definition
& way of testing the hypothesis.
22
Foundations of
Research
Question 1
A hypothetical
construct is…
A = A specific prediction about the
outcome of an experiment
B = A little known band from
Muncie Indiana
C = A general ψ process that
underlies our observations
D = A central element in a theory
23
Foundations of
Research
Question 2
To be testable, a
hypothesis…
Must rest on operational
definitions.
A = true
B = False
C = I don’t know
24
Foundations of
Research
Question 2
An operational
definition is
A = The procedure(s) we use to
measure a study variable
B = The way we define our theory
C = The procedure(s) we use to
manipulate a study variable
D = What we use to derive our
hypothesis
25
Foundations of
Research
Question 3
To be testable, a
hypothesis…
Must potentially be found to
be false.
A = true
B = false
C = I don’t know
26
Foundations of
Research
Question 1
A theory…
A = Leads to one specific
hypothesis
B = May be one of several ways to
explain something
C = Is not as important as simply
collecting data
D = Is what you make up to
explain why you forgot your
boy/girl friend’s birthday
E = Is not really affected by social
or personal values
27
Developing research questions and hypotheses.
Foundations of
Research

How do social values affect science?

Where do research questions &
hypotheses come from.

 Variables in research
28
Foundations of
Research
Theories
Types and uses of variables in research:
 Experiments

Independent vs. Dependent variables
 Measurement / field studies
 “Predictor” vs. “Criterion”
 Forming variables:


Direct manipulation

Measurement
Indirect manipulation
29
Foundations of
Research
Types of Variables:
Experimental designs
Independent Variable
30
Dependent Variable
Defines the “contrast space”
Models the phenomenon
 What is compared to what
e.g., drug v. placebo
 What is being explained;
e.g., task performance

Hypothetical “cause”
 “Effect”

Imposed by researcher
 Measured
as experimental conditions

Categorical (e.g., control
v. experimental group).
as the outcome
 Continuous (e.g., speed
or accuracy…)
Foundations of
Research
Types of Variables:
Measurement Studies
Predictor
31
Criterion or Outcome
Defines the “contrast space”
Models the phenomenon
 What predicts the outcome?
e.g., age, education, gender
 What is being explained;
e.g., political attitude

Hypothetical “Cause”
 “Effect”

Measured
 Measured
not imposed by researcher

Continuous
as the outcome
 Continuous
Foundations of
Research

Experimental v. Measurement designs
Experimental designs
 Manipulating the Independent Variable:
 Enhances internal validity
 May lessen external validity
 Participants randomly assigned to experimental v.
control groups

Measurement (or correlational) designs
 Measurement “in the field”
 May enhance external validity
 Typically lessens internal validity
 Sampling very important
32
Foundations of
Research
Creating independent variables [IVs]
1. Direct experimental manipulation

Most typical of “true” experiments

Maximum control over IV
2. Indirect manipulation via experimental or
research conditions

Less direct control over IV
3. Quasi-Independent variables: forming groups
using a measured variable.

Experiments without complete control over
variables

Used in measurement studies
33
Foundations of
Research
Forming Variables
1. Direct experimental manipulations
 Drug, biomedical or behavioral intervention or
treatment
 Lab study of, e.g., chemical processes under
different conditions.
 Focused experimental study of specific stimuli.
 System-wide “treatment” (e.g., policy change,
school-based program)
 Structure the IV
vis-à-vis:
 Simple presence v. absence of
the treatment or stimulus
 Single v. multiple treatment doses
 Type of treatment or stimulus
34
Foundations of
Research
Direct experimental manipulation
 Hypothesis: words presented in a semantic context are
recalled better than when presented randomly.
Independent
Variable
We manipulate this; it
defines our ‘contrast
space’ (what is being
compared to what…).
The IV is the putative
‘cause’ of the
dependent variable
Dependent
Variable
We measure this; it is
the phenomenon we
are trying to explain.
35
Foundations of
Research
Direct experimental manipulation
36
 Hypothesis: words presented in a semantic context are
recalled better than when presented randomly.
Independent
Variable
Experimental
group
Control
group
Dependent
Variable
Of course this group receives the experimental
condition or manipulation.
We think carefully about how we induce the
experimental condition, the appropriate
‘dose’, etc.
Foundations of
Research
Direct experimental manipulation
37
 Hypothesis: words presented in a semantic context are
recalled better than when presented randomly.
Independent
Variable
Dependent
Variable
Experimental
group
Control
group
This is the comparison or contrast group.
It may represent an alternative condition (e.g.,
comparing two drugs).
…or the simple absence of the experimental
condition (e.g., using the “standard of care” as a
control group in evaluating a biomedical or
behavioral intervention.)
Foundations of
Research
Direct experimental manipulation
 Hypothesis: words presented in a semantic context are
recalled better than when presented randomly.
Experimental
group
Control
group
Independent
Variable
Dependent
Variable
Target words
presented
within complete
sentences
Word
recognition task
Target words
presented
randomly
Word
recognition task
Completely controlled
by the experimenter
Experimental
manipulation same as
Independent Var.
38
Foundations of
Research
Forming Variables
39
2. Indirect experimental manipulations

Experimental “induction” of a mood or state…

“Stage manage” a social event

Induce mood via description of the experiment
EXAM PLE
Stress  require presentation in front of peers
Depression  Write about worst mistake you ever made
Stereotype threat  “This test reflects on your group”
Anxiety  Stage a robbery or fight
Happiness  Imagine you just won the lottery...
Relaxation  Meditation training
Foundations of
Research
Forming Variables
2. Indirect experimental manipulations


Experimental “induction” of a mood or state…

“Stage manage” a social event

Induce mood via description of the experiment
Do a manipulation check to see if you actually
manipulated your Independent Variable
 Self-report using a standard scale
 …did stress actually increase / decrease in the experimental
group?
 Observer ratings
40
Foundations of
Research
Indirect experimental manipulation
41
 Hypothesis: happiness enhances pain resistance.
Experimental
Manipulation
Independent
Variable
Experimental
group
Imagine you
won the lottery
– what will you
buy first?
Happy
state
Control
group
List what you
will need to buy
this month.
Normal /
baseline
state
Directly
controlled by
experimenter
Not directly
controlled by
experimenter
Our induction of the Independent
Variable (happiness) is indirect.
Dependent
Variable
Cold Presser
Task (ice bucket)
Cold Presser
Task
(How long can you
keep your hand in
the bucket…)
Foundations of
Research
Indirect experimental manipulation
42
 Hypothesis: happiness enhances pain resistance.
Experimental
Manipulation
Independent
Variable
Experimental
group
Imagine you
won the lottery
– what will you
buy first?
Happy
state
Control
group
…what will you
need to buy this
month?
Normal /
baseline
state
Dependent
Variable
Cold Presser
Task (ice bucket)
Cold Presser
Task
A Manipulation Check tests whether the
experimental manipulation actually induced the
Independent Variable
(…were experimental group participants actually made happy…)
Foundations of Quasi-independent variables
Research
43
3. Create a quasi-Independent variable
using a measured variable.

Most common approach to experimental-like analyses in
observational research areas


Anthropology, paleontology, earth sciences, sociology….
Categorize participants by measuring (not manipulating)
something:

Physical characteristics, e.g. species / sub-species boundaries

Locations, e.g., housing patterns, geologic strata

Scores based on a measured behavioral or social category
44
Foundations of Quasi-independent variables
Research
3. Create a quasi-Independent variable
using a measured variable.
Examples from behavioral science:

Scores over / under an established “cut point”,



Scores based on a frequency a distribution:

Median split: top v. bottom half.

Extreme scores: top v. bottom 10% of scores.
Simple self-identification:


e.g., over 4 depression symptoms on a standard scale.
e.g., “Republican” v. “Democrat”.
Behavioral index:

Used any drug in previous year v. not.

Voted in 2012 v. not.
Not a “True” IV:
Participants not
randomly
assigned to
groups.
Using a measured variable to create groups
Foundations of
Research



45
Administer depression scale, count the # symptoms rated 2 or 3.
Form groups based on a cut point; e.g., > 4 symptoms = quasi-clinical
depression.
Participants are assigned to groups based on their ratings, not
random assignment.
Below is a list of different feelings. Circle the number that shows how many days you felt each
of these over the PAST WEEK.
I was bothered by things that usually do
not bother me.
I felt I could not shake off the blues even
with help from my friends or family.
I had trouble keeping my mind on what
I was doing.
I felt depressed.
I felt that everything I did was an effort.
My sleep was restless.
I was happy.
I enjoyed life.
I felt sad.
Rarely or
none of
the time
A Little
of the Time
A moderate
amount of
the time
Most or all of
the time
(less than 1 day)
(1 or 2 days)
(3 - 4 days)
(5 - 7 days)
0
1
2
3
0
1
2
3
0
1
2
3
2
2
2
2
2
2
3
3
3
3
3
3
# of symptoms
rated0 2 or 3 1
0
0
0
0
0
1
1
1
1
1
Foundations of
Research
Overview: Types of Variables

Independent variable(s)

Dependent variable(s)

Control variables

Random variables

Confound
46
Foundations of
Research

47
Overview: Types of Variables
Independent variable(s)
 We hypothesize that the IV
explains the phenomenon we
are interested in.
 We assume it to be
Independent of any other
variable in the experiment.
 We manipulate or create it.
 Defines our ‘contrast space’;
what is being compared to what
(e.g., high v. low arousal
conditions…).

Dependent variable(s)
 The phenomenon we are
explaining:
•
Performance…
•
Learning…
•
Motivation…
 What we assess as the
outcome measure.
Foundations of
Research


Overview: Types of Variables
Independent variable(s)
Control variables

Dependent variable(s)
 Attributes of the experiment we keep constant between the
experimental & control groups;
•
Physical & social environment of the study;
•
Room,
•
Instructions & apparatus,
•
Experimenter….
48
Foundations of
Research


Overview: Types of Variables
Independent variable(s)
Control variables

Dependent variable(s)
 Attributes of the experiment we keep constant between the
experimental & control groups;
•
Physical & social environment of the study;
•
Basic procedures;
•
Recruitment and enrollment,
•
informed consent,
•
instructions,
•
baseline assessments…
49
Foundations of
Research


Overview: Types of Variables
Independent variable(s)
Control variables

Dependent variable(s)
 Attributes of the experiment we keep constant between the
experimental & control groups;
•
Physical & social environment of the study;
•
Basic procedures;
•
 everything except the Independent Variable. 
 So, we control for variables that may affect the results
(…recruitment methods…) by ensuring they are the same for all
groups.
50
Foundations of
Research



Overview: Types of Variables
Independent variable(s)
Control variables

51
Dependent variable(s)
Random variables
 Variables (other than the IV) that we do not control; we allow them to
vary randomly within and between groups
•
Demographics; age, ethnicity, education…
•
Attitudes & beliefs, psychological states…
 We consider these irrelevant – we have no hypotheses or interest in
their effect on the outcome variable (DV).
 By random assigning participants these variables should “even
out” across groups…
•
All groups should end up with about the same distribution of age,
education levels, etc.
•
If the groups differ in one of these variables it may represent a
confound.
Foundations of
Research




Overview: Types of Variables
Independent variable(s)
Control variables
Random variables
Confounds

52
Dependent variable(s)
 An unanticipated or unmeasured variable that differed between
groups, and may have led to the results rather than (or in addition to)
the IV.
• A control variable that was not, in fact, adequately controlled…
• A random variable that differed between groups and affected the results:
• A failure of the random assignment procedure; for some reason the
groups different in, e.g., education level, ethnicity…;
• A variable that was important enough the it should have been
carefully controlled.
Foundations of
Research
Overview: Types of Variables

Independent variable(s)
Control variables
Random variables

Confounds



Dependent variable(s)
 A variable other than the IV that affected the results.
 Confounds make it difficult (impossible?) to interpret the results .
• An confound we do not know of may create the illusion that our results
supported the hypothesis,
…when in fact the results were due to something else entirely,
…e.g., a mistake we made in the experimental design, biased (nonrandom) group assignment, etc..
• A known confound may be quasi-controlled through statistical analyses.
53
Foundations of
Research

Overview: Types of Variables
Independent variable(s)
 Our explanatory variable; manipulated.

Dependent variable(s)
 The phenomenon we are explaining; assessed.

Control variables
 Attributes of the experiment we keep constant between
experimental & control groups.

Random variables
 Variables we allow to vary randomly within and between groups.

Confounds
 A variable other than the IV that affected the results.
 Confounds can make results difficult (impossible?) to interpret.
54