Intro to Experimental PSYC

Intro to Research Methods in
Please read Chapters 4 and 5 in
Major Topics
• Research -- what it is.
• Planning a research project.
• Selecting the research units (“subjects”) from
which data will be collected.
• Constructing or obtaining the instruments used
to collect data.
• Relative merits of commonly employed research
• Analysis of the data.
• Preparing a research report.
Social Sciences Research
Research in the social sciences (including
psychology) is
– theoretical (sometimes),
– empirical
– nomothetic, and
– probabilistic
A Scientific Theory
• A model of how basic constructs and
measured variables are related to one
• Constructed by induction – inferring the
general from the specific.
• Generates predictions (hypotheses) by
deduction – inferring the specific from the
Tested by attempting to falsify.
From theory, derive hypotheses.
Gather relevant data.
If data match hypotheses, theory
• If not, modify theory or abandon it.
• Is not just a hunch.
• Explains all known facts in domain of
• Must be refutable.
Exploratory Research
• Not all psyc research is theory-driven.
• I can wonder about the relationship
between X and Y without any guiding
• Exploratory research provides the raw
materials from which theories are refined.
• Epistemology
– origins and nature of human knowledge
• Rationalism
– knowledge through thought
– Pythagoreans, Socrates
• Empiricism
– knowledge through sensory experience
• Applies to the general case.
– We study individuals
– to explain, predict, and control behavior
– not just in one individual, but in most.
• As opposed to idiographic.
– where the focus is on a single individual.
• Our interest is in the entire population
– for example, correlates of perceived
attractiveness in all humans.
• But our data (sample) represents only a
small proportion of that population.
• So our conclusions cannot be free of
possible error
– must be stated in probabilistic terms.
Three Basic Types of Research
• Descriptive (Univariate)
• Relational
• Causal
Descriptive Research
• Univariate – work with only one variable at
a time.
• Example: How many people dream in
Relational Research
• Determine how variables are related to
one another.
• Is age related to dreaming in color?
• Is sex/gender related to dreaming in color?
• Is arousal of certain brain areas related to
dreaming in color?
Causal Research
• Is X a cause of Y?
– If I manipulate X, will Y change?
• Establish that X and Y are related.
• Rule out (noncausal) alternative
• Employ experimental methodology.
• Eliminate confounds.
Third Variable Problems
Research and Time
• Cross Sectional Research
– Compare political attitudes of 20 vs 50 years
– Differences due to maturation or ??
• Longitudinal Research
– Follow multiple cohorts, measure at 20, 30,
40, 50 years of age.
• Repeated Measures and Time Series
Patterns of Relationships
Between Continuous Variables
Positive Linear
Negative Linear
See Bivariate Linear Correlation.
Null hypotheses
Alternative hypotheses
Nondirectional hypotheses
Directional hypotheses
Sharp/Point null hypotheses
Loose/Range null hypotheses
A Research Project
From Start to Finish
Formulating the Initial Broad Question
• Experiencing a practical problem.
• Familiarity with past research and theory.
• RFP – requests for research proposals.
• Curiosity about everyday experiences.
– my curiosity regarding attitudes about animal
Narrow Down the Question
• Want a question that can be well
addressed in a single research study.
• Focus in on one or a few parameters
– Theory may suggest which parameters are
most important
– Or you just might have a hunch while sitting
on the couch watching the news.
– How is misanthropy related to attitude about
The Research Hypothesis
• Theory, insight, or hunch may suggest an
answer to the question posed.
• This becomes the research hypothesis.
• Ethical cost/benefit analysis and a
relationship between misanthropy and
attitudes about animals.
• How to manipulate or measure the
concepts in the research hypothesis.
• Manipulate misanthropy by exposing some
subjects to depictions of evil humans.
• Develop questionnaire to measure
Reviewing the Literature
• Has somebody else already answered this
• If so, what additional research is
suggested by those results?
• Have others addressed similar questions?
• How did others operationalize constructs,
recruit subjects, get grants, analyze the
data, and so on.
Procedural Details
and Feasibility
Prepare a step-by-step plan.
Can you afford it?
If not, are there less expensive methods?
How many subjects will you need?
How can you motivate your subjects?
From whom will you need permission,
cooperation, or assistance?
• Are there statistical procedures available
to analyze the data you will collect in a
way that will answer your question?
• Can you conduct such analysis yourself or
will you have to hire a research
Gathering & Analyzing the Data
• Consider conducting a pilot study.
• You can count on some things going
wrong. Be prepared to deal with them.
• Screen your data to determine if you can
analyze them the way you intended to. If
not, adopt alternative analysis.
• Be on the lookout for unanticipated
findings and be prepared to shift your
attention to them.
Using Your Statistical Results to
Answer Your Question
• If your question was simple, your design
experimental, and your results what you
expected, this is a breeze.
• Otherwise, be ready to sweat it out.
Writing a Research Report
• Share your results with the world.
• Use the style appropriate for your
• For most psychologists, that is the style of
the American Psychological Association.
• I have (or will have) taught you the basics
of APA style.
Research Validity
Will your research lead to conclusions that
fit the available data and stand up to
Statistical Conclusion Validity
Internal Validity
Construct Validity
External Validity
Statistical Conclusion Validity
Determine the extent to which variables
are related.
• Power
• Efficiency of Estimation
• Robustness
Internal Validity
Is there a causal relationship between
these variables?
• Experimental research
• Various threats to internal validity
Construct Validity
Have we operationalized our constructs
• We have demonstrated a relationship
between our measured variables.
• Does this make us confident that there is a
relationship between the underlying
External Validity
Can we generalize the results?
• to other types of subjects
• to other situations
• to other measurements of the constructs
• etc.