Lecture Preview
n What is good research design?
n Explore the scientific method
n Discuss the ethics of experimentation
n Review statistics
n Peer review
Facilitated Communication: A Cautionary Tale
n Facilitated communication was a “revolutionary” treatment for autism (extraordinary claims)
n Biklen (1990) thought that autism was primarily a motor disorder
n Experimenter sat next to nonverbal child with autism and guided the child’s hand over a keyboard
n Students seemed to make stunning progress in communication
Facilitated Communication: A Cautionary Tale
n Students began making allegations of brutal sexual abuse
n Dozens of controlled studies examined the phenomenon and found that the words came solely from the minds of the facilitators (much like a Ouija board)
n It is still difficult for proper research findings to get into the mainstream psyche; some people still practice facilitated communication
The Beauty and Necessity of Good Research Design
n But I know it works!
n Often our impressions are wrong
n Prefrontal lobotomy - example of what happens when we rely on our subjective impressions
n Egaz Moniz won the Nobel prize for this procedure
n Controlled studies showed it didn’t work
Prefrontal Lobotomy: Psychosurgery and Reliance on Subjective Impressions
Heuristics and Biases: How We Can Be Fooled
n Heuristics - mental shortcuts or rules of thumb
n Reduce the cognitive energy required to solve problems
n We tend to oversimplify reality
n Imagine yourself driving from Reno, Nevada, to San Diego, California - what compass direction would you take?
Common Heuristics Studied by Kahneman and Tversky
n Representativeness - “like goes with like”
n Base rate - how common a characteristic or behavior is in the general population
n Base rate fallacy
n Availability - “off the top of my head”
n Estimating the likelihood of an occurrence based on the ease with which it comes to our minds
Cognitive Biases: Systematic Errors in Thinking
n Hindsight bias (“I knew it all along”) - tendency to overestimate how well we could have successfully forecasted known outcomes
(e.g., “I knew they were the perfect couple”)
(e.g., “I knew they were the perfect couple”)
n Overconfidence - tendency to overestimate our ability to make correct predictions
n These errors can lead to confidence in false conclusions
Scientific Method: A Toolbox of Skills
n Allows us to test specific hypothesesderived from broader theories of how things work
n Theories are never “proven,” but hypotheses can be disconfirmed
n Naturalistic Observation - watching behavior in real-world settings with
n High degree of external validity - extent to which we can generalize our findings to the real world
n Low degree of internal validity - extent to which we can draw cause-and-effect inferences
n Case study designs
n Depth is traded for breadth
n Common with rare types of brain damage
n Helpful in providing existence proofs, but can be misleading and anecdotal
n Correlational designs
n Correlation can vary from –1 to +1
n 0 means no relationship
n Depicted in a scatterplot - each dot represents a single person’s data
n Illusory Correlation - perception of a statistical association where none exists (e.g., crime and the full moon)
n Correlation cannot determine causation - merely shows things are related or associated
Experimental Design: What Makes a Study an Experiment?
n Random assignment of participants to conditions
n Experimental Group - receives the manipulation
n Control Group - does not receive the manipulation
n Independent Variable - experimenter manipulates
n Dependent Variable - experimenter measures to see whether manipulation had an effect
n Confounds - any difference between the experimental and control groups, other than the independent variable; makes independent variable effects uninterpretable
n Cause and effect - possible to infer, with random assignment and manipulation of independent variable
Pitfalls of Experimental Design
n Placebo effect - improvement resulting from the mere expectation of improvement
• Subjects must be blind - unaware of whether they are in the experimental or control group
• Placebos show many of the same characteristics as real drugs
n Nocebo effect - harm resulting from the mere expectation of harm (e.g., voodoo doll phenomenon)
n Experimenter expectancy effect - phenomenon in which researchers’ hypotheses lead them to unintentionally bias a study outcome
n Clever Hans, the mathematical horse
n Rosenthal’s undergrads and maze-bright, maze-dull rats
n Double-blind design - neither researchers nor subjects know who is in the experimental or control group
n Hawthorne effect - phenomenon in which participants’ knowledge that they’re being studied can affect their behavior
n Demand characteristics - cues that participants pick up from a study that allow them to generate guesses regarding the researcher’s hypotheses
n To minimize Hawthorne effects:
• Covert observation
• Participant observation
Asking People About Themselves and Others
n Random selection - key to generalizability; ensures every person in a population has an equal chance of being chosen to participate
n Evaluating Measures:
n Reliability - consistency of measurement
n Validity - extent to which a measure assesses what it claims to measure
n A test must be reliable to be valid, but a reliable test can still be completely invalid
Self-Report Measures and Surveys
n Self-report measures - questionnaires assessing a variety of characteristics (e.g., interests, traits)
n Surveys - measure opinions, attitudes
n Question phraseology is crucial
n Pros
n Easy to administer
n Direct (self) assessment of person’s state
n Cons
n Accuracy is skewed for certain groups (narcissists)
n Potential for dishonesty
nResponse sets - tendencies of research subjects to distort their responses
nPositive impression management
nMalingering
Ratings Data: How Do They Rate?
n Halo effect - tendency of ratings of one positive characteristic to spill over to influence the ratings of other positive characteristics
n Leniency effect - tendency of raters to provide ratings that are overly generous
n Error of central tendency - an unwillingness to provide extreme ratings (low or high)
Ethical Issues in Research Design
n Tuskegee Study (1932 to 1972)
n African American men living in rural Alabama diagnosed with syphilis
n U.S. Public Health Service never informed, or treated, the men
n Merely studied the course of the disease: 28 men died of syphilis, 100 of related complications, 40 wives were infected,
19 children were born with it
19 children were born with it
n In 1997, President Clinton offered a formal apology
Modern Ethical Guidelines
n Institutional Review Board (IRB)
n Informed Consent
n Justification of deception
• Milgram’s obedience study
n Debriefing of subjects afterward
n Animal Research
n Institutional Animal Care and Use Committee (IACUC)
n About 8% of psychological research uses animals
Statistics: The Language of Psychological Research
n Descriptive statistics - numerical characteristics of the nature of the data set
n Central tendency - where the group tends to cluster
• Mean - average of all scores
• Median - middle score in the data set
• Mode - most frequent score in the data set
n Dispersion - sense of how loosely or tightly bunched scores are
• Range - difference between the highest and lowest scores
• Standard deviation - measure of dispersion that takes into account how far each data point is from the mean
Statistics: The Language of Psychological Research
n Inferential statistics - mathematical methods that allow us to determine whether we can generalize findings from our sample to the population
n Statistical significance - finding would have occurred by chance less than 1 in 20 times
n Practical significance - real-world importance
n Statistical deceptions
n Example: Truncated line graphs
Evaluating Psychology in the Media
n Most reporters are not scientists, so
n Consider the source - tabloid vs. Discover
n Beware of
• Sharpening - exaggerating the central message of the study
• Leveling - minimizing the less-central details
• Pseudosymmetry - appearance of scientific controversy where none exists while purporting to provide “balanced coverage”
• Example: Four paragraphs supporting ESP, four paragraphs against it
0 comments:
Post a Comment