The flashcards below were created by user
nthor
on FreezingBlue Flashcards.
-
2 reasons you should care about ethics
- respect for participants' rights
- protect yourself legally
-
2 sources of info on ethics
- apa statement
- institutional review boards
-
risk benefit rule
benefits should outweigh risks
-
4 rules of ethics
- informed consent
- freedom from coercion
- confidentiality
- debriefing
-
internal validity
- extent to which a set of research findings provides compelling information about causalitylow internal validity means not sure of cause and questionable findings
-
John Stuart Mill's 3 principles
- covariation
- temporal sequence
- eliminating confounds
-
covariation
- changes in one variable must correspond with changes in another variable
- cause and effect correlate
-
temporal sequence
- changes in thefirst variable must precede the changes in the second
- cause must come before effect
-
eliminating confounds
- must rule out other explanations
- also known as "third variable problem"
-
external validity
- extent to which a set of research findings providesan accurate description of what typically happens in the real world
- high generalizability means the study will apply to other people too
-
internal validity vs. external validity
- internal - accuracy
- external - generalizability
-
construct validity
- are you measuring or manipulating what you want
- good operational definitions lead to high construct validity
-
3 threats to construct validity
- confounding variable
- random error
- poor measurement
-
conceptual validity
how well a hypothesis maps on to the broader theory that it was designed to test
-
measurement
assigning numbers or names to objects and their attributes
-
4 types of measurement
- verbal
- behavioral
- behavioroid
- physiological
-
verbal measurement
- verbal response
- example: pain scale
-
behavioral measurement
- real behavior
- example: avoiding someone
-
behavioroid
- participants report on their own behavior
- example: condom use
-
physiological measurement
- biological processes
- example: brain imaging
-
indicators of theoretical constructs
what we can observe
-
operationalization of theoretical constructs
finding good indicators of theoretical constructs
-
4 types of measurement scales
- nominal
- ordinal
- interval
- ratio
-
reliability
consistency and repeatability of a study
-
3 types of reliability
- inter-observer
- internal consistency
- temporal consistency
-
inter-observer reliability
- different people judge and agree
- used to study behavior or products of behavior (writing sample)
-
internal consistency reliability
- multiple items are combined to create a score
- used with self-report scales or observations
-
temporal consistency reliability
- used to assess something that doesn't change on a regular basis
- test a group of people and then have them come back a second time to test them again
- test-retest reliability: over the span of 2 weeks, do you get the same score?
-
"more is better" rule of reliability
- more items (indicators, observers, observations, and occasions) lead to better reliability
- less vulnerable to chance
-
3 construct validity of scales
- predictive
- convergent
- discriminant
-
predictive construct validity
is the measure useful?
-
convergent construct validity
are we assessing what we want?
-
discriminant construct validity
are we assessing the wrong thing?
-
reliability vs. validity
- a study can be reliable but not necessarily valid
- example: a broken scale consistently reads 5 lbs over but your weight is not 5 lbs over
-
2 challenges of self-report questions
- judgement phase
- response translation phase
-
judgement phase
participants determine what kind of question is being asked and form some initial response
-
pilot testing
practice studies designed to help researchers refine measures in the real study
-
focus group
small and well represented group of people
-
open-ended questions
allow participants to respond any way they like without a structured scale
-
4 ways to word questions well
- KISS (informal language)
- avoid negations and double negatives
- avoid forced-choice items (double barreled)
- avoid questions that do not yield variance
-
anchors
- adjectives that lend meaning to the numbers on the scale
- 1 = not at all
- 5 = extremely
-
EGWA scale
empirically grounded and well anchored
-
3 steps to designing questionnaires
- step back and think
- write lots of questions
- analyze your scale and derive the best items
-
absolute scales
- participants report magnitude
- how many cigarettes do you smoke a day?
-
likert scale
rating scale
|
|