-
empirical research
based on observations
-
objective research
verifiable, unbiased, repeatable, falsifiable, predictable, includes operational definitions
-
operational definitions
the way you define your variable
-
systematic research
organized, procedural, specific to one claim
-
best way to find causality
randomized experiments
-
Why must a hypothesis be grounded in theory?
A hypothesis should always explain what you expect to happen during the course of your experiment or research
-
theory
a well-established principle that has been developed to explain some aspect of the natural world
-
hypothesis
predictions grounded in theory (already known facts) about the world that might be true
-
predictions
an idea of the research’s outcome
ie. People will do ‘abc’ during ‘x’ amount of time in ‘y’ type of situation
-
variables
an event, situation or behavior that has a chance to be different
-
operational definition
the definition of the your dependent variables
-
Describing, predicting, determining causes for, and explaining behavior
are the goals of psychological research
-
Describing Behavior example...?
people have a rightward bias in head-turning when kissing
-
Predicting Behavior example...?
voting polls show that you are more likely to be a democrat if income is low
-
Predicting Behavior: beware...
- Direction of cause and effect
- Confounding variables - A variable whose effects on the response variable cannot be distinguished from one or more of the explanatory variables in the study
- Lurking variables - A variable whose effects on the response variable cannot be distinguished from one or more of the explanatory variables in the study, AND IS NOT INCORPORATED INTO THE DESIGN OF THE STUDY
-
Determining Causes for Behavior correlations:
A positive relationship has positive slope and r ~ 0.9
A negative relationship has negative slope and r ~ 0.9
No relationship will have zero slope and r = 0
A curvilinear relationship has different slopes - because as x increases, y decreases
-
Clarifying Explanations
Intermediate relationships between a dependent and independent variable
-
Explaining Behavior example...?
An increase in ice cream sales means an increase of the number of people getting cramps, and more people are drowning - CRAMPS are the clarifying explanation
-
An “expert” is not someone who has the
correct answers. It is
- someone who knows the evidence very well
- and thus can make a persuasive argument.
-
Any consumer of research still needs to
evaluate
the evidence being presented.
-
Therapeutic Touch – “Their failure to substantiate TT’s most fundamental claim is unrefuted evidence that the claims of TT are
groundless and that further professional use is unjustified.” Example of:
unobjective claims by TT's.
-
Randomized experiment more specific than broader term “experiment” in everyday language (e.g., ‘chemistry experiment’).
Define randomized experiment.
All participants in an experiment in psychology have the same equal chances of being chosen as anybody else.
-
Which plants received nitrogen in the "Effects of Manure on Potatoes" is an example of
randomization. The plots which receive treatments in determined randomly.
-
Scientific evidence is objective and systematic. What is not?
Intuition is not.
-
The "Hot Hand" in Basketball belief is an example of
Intuition being wrong, and not scientifically evidential.
-
Hypothesis
a statement about the world that may be true
-
Variable
- Any event, situation, or behavior
- that has at least 2 values
-
Operational Definition of Variables
- A definition of a variable in terms of how it is measured
- Examples: "aggression," "stress," "self-esteem"
- Operational definitions put concepts in the public realm where they can be criticized, tested, improved, or rejected
- All variables must have operational definitions to meet the objectivity criterion
-
Predictions
concern the outcomes of specific studies
-
Hypotheses are usually
stated at the level of theoretical constructs (“SES is associated with greater intelligence,” “Extraverts are high in sensation-seeking”)
-
Predictions are usually
stated at the level of operational definitions (“yearly income in US $ will be correlated with higher scores on the Stanford-Binet IQ test,” “people who score highly on the NEO extraversion scale will report a greater desire to try sky-diving”)
-
Theories generate and explain
sets of hypotheses.
-
Theory
is a conceptual framework that organizes and explains a larger body of facts.
-
Extravert / Introvert Study Theory:
Extroverts have low baseline levels of physiological arousal / Introverts have high baseline levels of physiological arousal
-
Extrovert / Introvert Hypothesis:
Extraverts will be high in sensation-seeking / Introverts will be low in sensation-seeking
-
Extrovert / Introvert Study Predictions:
- People who score highly on the NEO
- extraversion scale will report stronger desire to try sky-diving
-
Extrovert / Introvert Study Data / Observations:
- Data from a specific sample do or do not
- support the prediction
-
Extrovert / Introvert Study Conclusions:
- Theory is provisionally supported or
- hypotheses/theory may require modification
-
Problems in making causal statements:
- direction of cause & effect
- Confounding or “lurking” variables
-
Showing a relationship (correlation)
between two variables does not show
cause and effect.
-
Alternative vs. Clarifying Explanations:
- A lurking variable (like temperature)
- should correlate with both the
- independent and dependent variable in order to be a possible alternative
- explanation for a causal relationship between them.
-
In the "alcohol consumption leads to a longer life" study, what are possible causes of "wine effect"?
- Wine intake itself reduces risk of death
- Antioxidants in wine may be
- protective against heart disease/cancer
- Direction of causality
- Confounding or lurking variable of physical activity
-
Hypothesis: that electromagnetic waves produced by cell phones may have adverse effects on men’s sperm ... Study method: Self-report how often use cell phones ... Scientific evidence: Measure of sperm counts from semen samples ...
Study showed correlation between cell phone use and low sperm count.
Alternative explanations for the cell phone – sperm count correlation other than the idea that cell phone usage CAUSES lower sperm counts?
- Cell phones --> Electromagnetic radiation --> Kills sperm
- Stressful job --> Frequent cell phone use & Lower testosterone --> Lower sperm
- No causal relationship: correlation arises from third variable of stressful job correlated with both phone use and lower sperm.
-
Does the negative correlation between cell usage and sperm counts at least provide EVIDENCE that phone usage may cause lower sperm counts?
- Yes. Causal evidence from correlational data may depend on converging sources of evidence:
- –Is there a theory that explains/predicts a causal relationship?
- –In vitro studies of effects of electromagnetic waves on sperm?
- –Controlled experiments in animals that show same effect?
- –Although correlation does not *directly prove* causation, it can be useful EVIDENCE for causation
-
Limitations of Correlational Research
- The number of potential lurking variables is in principle infinite. –Cannot statistically control for an infinite number of variables using techniques like partial correlation. –Often cannot control for variables perfectly (e.g., when controlling for “education,” do all bachelor’s degrees indicate identical education?)
- Need to evaluate strength of evidence and likelihood of alternative explanations on a case by case basis.
-
In experimental research, the researcher manipulates
- at least one variable and then measures at least one outcome.
- EX: Subjects could be randomly assigned to either: (1) drink 5 cups of coffee a day or (2) drink 0 cups of coffee. Then compare cholesterol of 2 groups after a certain length of time (6 months, 1 year, 5 yrs…)
-
Manipulated variables are called
independent variables
-
Outcome variables are called
dependent variables
-
In non-experimental research, independent variables are
those hypothesized to be the cause.
-
In nonexperimental research, dependent variables are
those hypothesized to be the effect.
-
In experimental control, a key idea in experiments is that as much as possible,
ONLY the manipulated variables differ between conditions of testing.
-
In experimental control, the goal is to ISOLATE the... _____ ... so that it is the only possible cause of differences between conditions
independent variable
-
For experimental control, in a CORRELATIONAL study, individual differences between participants may act as
- lurking variables.
- EX: Already coffee-drinkers might exercise less, and this explains the "relationship" between coffee consumption and cholesterol.
-
In an experiment, ... ____ ...
should hold constant individual differences that could act as lurking variables, if sample size is large enough
random assignment
-
Say we have 100 subjects for our coffee experiment, 50 who exercise regularly and 50 who do not.
–If we randomly assign (e.g., flip a coin) to the coffee vs. no coffee conditions,
how many exercisers do we expect in each condition? How many no exercisers?
–Is exercise a lurking variable in this design?
- - 25 exercisers in each condition.
- - 25 no exercisers in each condition.
- - Yes:
-
Experimental control: In our coffee example, one group drinks 5 cups of coffee per day and the other group zero. Say the shop providing the coffee for the experiment is at the top of a steep hill on campus.
–Is that a problem?
Is coffee consumption the only thing that differs between the two experimental groups?
- - Yes: Non-exercisers might lose cholesterol BECAUSE they have to walk up a steep hill every day -- not because of drinking coffee. It would be impossible to tell the correlation -- walking up the hill is a confound.
- - No: If the shop was on a hill, then ONLY the coffee-drinkers would be getting daily exercise guaranteed, and non-drinkers would not.
-
Variables that vary along with the independent variable in an experiment (but that we did not intend to manipulate) are known as
confounding variables.
-
In an experiment, confounding variables co-vary (varies together with one variable) with one experimental group but
not the other.
-
In an experiment, confounding variables can act as alternative explanations, just like confounding variables in correlational research.
-
In an experiment, the key idea is that
the independent variable is no longer isolated if confounding variables exist
-
Sham surgeries in animal experiments
as much as possible make sure that only ONE variable is being manipulated between different conditions, since STRESS and DAMAGE can act as confounding variables during surgery.
-
What is a solution to the confounding variable problem of a shop on a steep hill in the coffee experiment example?
- - Take the shop off the hill
- - Make ALL subjects drink some liquid daily - real coffee, placebo "real" but fake coffee, and decaf as the control group
- - Make ALL subjects walk up the hill
-
In an experiment, if randomization is adequate, then
an infinite set of individual difference variables in principle are eliminated as possible lurking variables.
-
In an experiment, if a manipulation introduces a confounding variable, then
alternative explanations exist for differences between conditions.
-
In a perfectly run experiment, any difference between experimental groups on the dependent variable must be
caused by the manipulated variable (since ONLY the manipulated variable DIFFERS between the groups)
-
Experiments can in principle show causation. However, we still need
theory to interpret the effects of a manipulation.
-
Touch Therapy for Premature Infants Experiment: Human contact/touching improve health outcomes for premature infants? Say we performed a correlational study and greater touching associated with greater weight gain. Can we conclude that touching causes greater weight gain?
No: nurses or research assistants may have been drawn towards touching or differently interacting with healthier looking babies to begin with, therefore "showing" touching causes greater weight gain (when in fact, those healthy looking babies may have just grown more than the unhealthy looking babies due to physiological traits)
-
Touch Therapy for Premature Infants Experiment: 40 pre-term infants (< 36 weeks at birth) randomly assigned to a touch condition (experimental group) or no treatment (control group)
Experimental Condition: 15 minute session of stimulation each hour for 3 consecutive hours for 10 days (weekends off). Both “tactile” and “kinesthetic” stimulation. Tactile for 5 min., kinesthetic for 5 min., tactile again for 5 min. Control Condition: No treatment; normal nursing rounds. Weight
gain measured from hospital charts. Fed same amount of food.
Results: infants in Touch group gained more weight.
What if a correlational study had been run instead, simply measuring how often nurses touched babies and then measured outcomes of levels?
Different conclusions would arise from the study, because nurses may not know but they may be touching certain infants with certain "health" traits. That would have meant the possibility of only the healthy looking infants having a positive result from touch. It would have been an alternative explanation.
-
As a consumer of research, one of the first things you need to know is whether the research is correlational or experimental. Often in media reports, there is no explicit statement of this. Search for clues: words like randomized, placebo-controlled, investigators manipulated, etc.
When evaluating any research, need to reflexively think about and search for possible confounding variables that could act as alternative explanations
-
Determining that A causes B does not necessarily demonstrate why or fully “explain” the phenomenon.
-
Theories function to supply candidate explanations that can then be tested with other data collections.
-
"Basic" Psychology Research:
Attempts to answer fundamental questions regarding the operation of brain mechanisms and the generation of behavior. Often addresses “why” questions.
-
"Applied" Psychology Research:
Attempts to find solutions to practical problems.
-
What was the evidence that 1980's William Halsted's "radical" mastectomies worked better than simpler surgeries?
There was a theory: "cancer spreads in spatial gradients from site of tumor; greater chances of cure if remove more tissue expanding from site" ...and data suggesting that some of the patients were cured.
-
The randomized experiment of testing the effectiveness of radical mastectomies enrolled 1765 patients over 10 years. Patients with breast cancer were randomly assigned to (1) radical mastectomy, (2) simple mastectomy, (3) simple mastectomy + radiation.
Results?
The rates of relapse and death were not significantly different between treatment groups. Correlational data could not determine the effectiveness of this procedure, and one large experiment gave a fairly conclusive answer. Experiments are powerful: they can directly assess causality.
-
Non-experimental (observational, correlational) Methods
- Measure variables but do not manipulate them
- Advantages: Can observe people in real-life situations, can sample large numbers of people (representative surveys)
- Disadvantages: Difficult to demonstrate cause/effect relationships
-
Experimental Methods
- 2 things define experiment: (1) manipulate some variable, (2) randomly assign to conditions (or order of conditions)
- Advantages: Can demonstrate causal relationships
- Disadvantages: Laboratory settings are often unrealistic, may not generalize to real world
-
3 Validities
- Internal Validity - Can we say from our experiment that one variable ACTUALLY DID CAUSE another variable?
- Construct Validity - Does the operational definition of the variable reflect its intended theoretical meaning?
- External Validity - Can the results of a study be generalized to other subjects or contexts?
-
Regarding research ethics, science strives to be objective, so repeatable conclusions are based on concrete observations. Ethics is unavoidably subjective to some degree:
Being based on agreed upon standards, conventions and social norms, ethics is relative.
-
Who sits on the Institutional Review Board of Psychology?
IRB members should have the professional experience to provide appropriate scientific and ethical review. An IRB must have at least one scientist member and at least one member whose primary concerns are nonscientific. Additionally, there must be one member who is not otherwise affiliated with the institution (a community representative).
-
Dealing with subjects in a psychology experiment, the Cost / Benefits Analysis (beneficence)
weighs the positives and negatives of approving a psychology experiment. If regarding animals, the cost of maltreatment of the animals is weighed with the benefits of humans.
-
Regarding the use of deception in a psychology experiment, which out of the three had results that displayed in-group favoritism?
Blue Eyes / Brown Eyes Classroom Exp.
Milgrim's Obedience Exp.
Zimbardo's Stanford Prison Exp.
Blue Eyes / Brown Eyes Classroom Exp.
-
Regarding the use of deception in a psychology experiment, which out of the three had results that displayed influence over their actions against someone else?
Blue Eyes / Brown Eyes Classroom Exp.
Milgrim's Obedience Exp.
Zimbardo's Stanford Prison Exp.
Milgrim's Obedience Exp.
-
Regarding the use of deception in a psychology experiment, which out of the three had results that displayed role-playing techniques that escalated to threatening or harmful levels?
Blue Eyes / Brown Eyes Classroom Exp.
Milgrim's Obedience Exp.
Zimbardo's Stanford Prison Exp.
Zimbardo's Stanford Prison Exp.
-
Animal research: test the relationship between work-related stress and ulcers: one monkey of each of 4 pairs was the “Executive” – press lever at least once every 20 sec to avoid shocks; other monkey passive. Two 6-hr shifts a day, 7 days a week. Results: 3/4
Executives died of ulcers on days 9, 23, and 25. Passive monkeys did not develop ulcers, and ulcers develop during rest periods, not
during stress periods.
-
Data fraud can be noticed if
- data seem too perfect
- there is identical percent error (or, noise) are in different experiments
- an experiment is difficult to reproduce by other people
-
For the Intelligence Quotient, IQ, "intelligence" is related to the speed of mental processing, the ability to think creatively, a variety of mental and physical abilities, and a general factor. What were the Wechsler Adult Intelligence Scale topics subjects were tested on?
- Comprehension
- Information
- Arithmetic
- Block Design
-
Is the IQ test a reliable and valid measure of intelligence? Does it follow "reliability"? Does it follow the "Pearson product-moment correlation"? Define both.
- Reliability: test-retest reliability: if an individual is tested at two different times, will the scores be the same? (e.g. a scale measuring my weight)
- Pearson product-moment correlation coefficient (r): can range from -1.00 to +1.00
- r = +0.80 is a positive correlation, positive slope, when one variable increases, the other increases
- r = - 0.80 is a negative correlation, negative slope, when one variable increases, the other decreases.
-
Reliability:
Internal Consistency Reliability:
split-half reliability: If you divide the items of a test in half, would both halves have the same score?
|
|
