Research Methods Summary Mila
Chapter 1:
Don’t rely on intuition and authority -> cognitive and motivational biases
Skepticism: evaluation based on careful logic and results
Empiricism: knowledge based on observation
Characteristics of scientific investigation:
o Data is central
o Scientists not alone: research is reported
o Adversarial: battle
o Peer review
Falsifiability: scientific ideas are testable
Pseudoscience
Goals behavioral science:
o Describe behavior: observation, relation
o Predict behavior: make better decisions
o Determine causes: change behavior
Temporal precedence: A -> B, not B -> A
Covariation cause and effect: when A -> B, when not A -> also not B
Elimination of alternative explanations: no confounding variable
o Explain behavior: why does behavior occurs
Basic research: answer questions of nature
Applied research: problems and solutions
Program evaluation: (MSD)
Induction: from data to theory
Deduction: using theory to make predictions about new data
Chapter 4:
Validity = accuracy
o Construct validity: are you measuring what you’re supposed to
o Internal validity: accurate conclusion about cause and effect
Temporal precedence
Covariation of cause and effect
Elimination of alternative explanations
o External validity: can you generalize findings
Operational definition: the set of procedures used to measure or manipulate the variable
Positive/negative/curvilinear/no relationship
Correlation coefficient: strength and direction of correlation (-1.0 till +1.0)
Nonexperimental method: correlational method
o Observation
o Used when it would otherwise be unpractical or unethical
o Difficult to determine direction of cause and effect
o Third-variable problem: alternative explanation
Confounding variable
o
Experimental method
o Direct manipulation
o Independent and dependent variable
, o Experimental control: all extraneous variables are kept constant
o Randomization
Field experiment: variable is manipulated in natural setting
o But no control
Participant variables: characteristics of individuals (age, gender)
Chapter 8
Posttest-only design:
o 2 equal groups -> no selection difference
Pretest-posttest design:
o Attrition/mortality
o Time consuming/ sensitize
Solomon four-group design
Between-subject design:
o Manipulation
o Randomization
o Elimination of confounding variable
o Measure dependent variable once
o Compare conditions in term of means
Within-subject design:
o Manipulation
o Counterbalancing or randomization of groups: because you don’t want order effects
o Order effects:
Practice effect
Fatigue effect
Carry-over effect: influence of one condition on the other condition
Spontaneous recovery
Autonomous maturation: in longer period people can change which could
become a confounding variable
o Elimination of confounding variables
o Measure dependent variable multiple times
Participant is in all conditions
o Compare conditions in term of means
Counterbalancing: determine order of conditions very specifically
o Intra-group: each participant receives one of orders on the list at random
o Intra-subject: each participant receives the conditions in multiple, sophisticated
orders.
o Balanced Latin square: A B L C L-1 D L-2 E etc.
Randomization of groups:
o Intra-group: order for each participant determined anew
o Intra-subject: each participant receives each condition very often in a random order
Mixed designs: improvement = posttest – pretest
Chapter 1:
Don’t rely on intuition and authority -> cognitive and motivational biases
Skepticism: evaluation based on careful logic and results
Empiricism: knowledge based on observation
Characteristics of scientific investigation:
o Data is central
o Scientists not alone: research is reported
o Adversarial: battle
o Peer review
Falsifiability: scientific ideas are testable
Pseudoscience
Goals behavioral science:
o Describe behavior: observation, relation
o Predict behavior: make better decisions
o Determine causes: change behavior
Temporal precedence: A -> B, not B -> A
Covariation cause and effect: when A -> B, when not A -> also not B
Elimination of alternative explanations: no confounding variable
o Explain behavior: why does behavior occurs
Basic research: answer questions of nature
Applied research: problems and solutions
Program evaluation: (MSD)
Induction: from data to theory
Deduction: using theory to make predictions about new data
Chapter 4:
Validity = accuracy
o Construct validity: are you measuring what you’re supposed to
o Internal validity: accurate conclusion about cause and effect
Temporal precedence
Covariation of cause and effect
Elimination of alternative explanations
o External validity: can you generalize findings
Operational definition: the set of procedures used to measure or manipulate the variable
Positive/negative/curvilinear/no relationship
Correlation coefficient: strength and direction of correlation (-1.0 till +1.0)
Nonexperimental method: correlational method
o Observation
o Used when it would otherwise be unpractical or unethical
o Difficult to determine direction of cause and effect
o Third-variable problem: alternative explanation
Confounding variable
o
Experimental method
o Direct manipulation
o Independent and dependent variable
, o Experimental control: all extraneous variables are kept constant
o Randomization
Field experiment: variable is manipulated in natural setting
o But no control
Participant variables: characteristics of individuals (age, gender)
Chapter 8
Posttest-only design:
o 2 equal groups -> no selection difference
Pretest-posttest design:
o Attrition/mortality
o Time consuming/ sensitize
Solomon four-group design
Between-subject design:
o Manipulation
o Randomization
o Elimination of confounding variable
o Measure dependent variable once
o Compare conditions in term of means
Within-subject design:
o Manipulation
o Counterbalancing or randomization of groups: because you don’t want order effects
o Order effects:
Practice effect
Fatigue effect
Carry-over effect: influence of one condition on the other condition
Spontaneous recovery
Autonomous maturation: in longer period people can change which could
become a confounding variable
o Elimination of confounding variables
o Measure dependent variable multiple times
Participant is in all conditions
o Compare conditions in term of means
Counterbalancing: determine order of conditions very specifically
o Intra-group: each participant receives one of orders on the list at random
o Intra-subject: each participant receives the conditions in multiple, sophisticated
orders.
o Balanced Latin square: A B L C L-1 D L-2 E etc.
Randomization of groups:
o Intra-group: order for each participant determined anew
o Intra-subject: each participant receives each condition very often in a random order
Mixed designs: improvement = posttest – pretest
