external validity
can outcome of observation be applied and expected in other settings
- does it apply to other settings, people and time?
internal validity
does the study answer design and analysis answer the og research question
case studies
- focuses on a specific case, mostly just 1 person
- pro; give rich info
- con: low external validity (doesnt always apply to everyone else)
naturalistic observation
- examines what people do in natural context, in their own environemtns
-con: reactivity: when people know they are being watched, they act differently - con: possible low internal validity
- pro: can have high external validity (if no reactivity)
archival study
- doing research using existing data
- pro: less invasive
- con: lack of quality control (dont exactly know if this information i true or good)
surveys
- pro: ease of administration, convenient
- con: response/error bias
- malingering and social desirability - faking answers to make people feel bad for you
- anonymity helps
- con: ambiguity (do people fully understand what your trying to measure)
- will answer differently based on how they interpret the questions
reliability
- to be reliable a test must produce similar answers over an over again
- internal validity
- test-retest reliability
- inter-rater reliability
internal consistency
- do results agree
- all answers should lead to similar and consistent results
- shouldn’t have conflicting results
test-retest reliability
- same test different timeline
- will the same test provide the same results two days apart?
inter-rater reliability
do two people agree on the resuts
validity
valid if it measures what its supposed to measure
- face validity
- convergent validity
- divergent validity
face validity
- does the test measures what its supposed to
- does a survey ask about cars if they want to know how often someone bikes
convergent validity
test that measure the same things should have the same results
divergent validity
test that measure different things should have different results
correlation studies
- related things but not causation
two variables have a relationship
advantage: can make predictions
disadvantage: cannot infer causality
correlation graphs
- positive if going up (both variables increasing)(+1)
- negative if going down (one inc, one dec)(-1)
- no correlation: straight line
- closer dots are on scatter plots, more correlated
illusory correlation
looks like theres a correlation but in reality there isnt
- jinx, or spiritual beliefs
random assignment
randomly assigns people to groups in experimental design
confounds
variable that can alternatively affect your variables
- can think that y cause x but really z is affecting them both
- could be placebo effect or
- participant demands (behaving the way researcher wants you to)
- or experimental effect (researcher has bias)
experimental design
- has random assignment
- for cause and effect
- has manipulation
quasi-experimental design
no random assignment but still manipulation
- some things cant be random (marital status, ethnicity, experiences)
W.E.I.R.D
mostly study weird samaples because we are in the western world
W: western
E: educated
I: industrialized
R: rich
D: democratic
descriptive statistics
- characterizes or summarizes the data
- mean, median, mode
- density function: formula that estimates the line of best fit
positive skewed distribution (stats)
- long tail on the right
- mean gets pulled to the right
