L4 - scientific/experimental method Flashcards
Braitenberg vehicle
tempted to describe machines with very simple internal structures using psychological language
dynamic, complex, unpredictable behaviours can emerge from simple systems
Ockham’s razor
given several explanations, simplest should be accepted (easiest to test)
reductionism
only need mechanistic explanation
but do we need a psychological explantion/would it add something about the true nature?
cannot empirically exclude philosophical possibilities, only know what we can observe
problem: emergent behaviour - emergent behaviour may be unpredictable
rationalism
using reason & logical arguments
observation unecessary and potentially misleading
empiricism
hypotheses need to be (dis)confirmed by observation using empirical, descriptive, and experimental methods
observation reproduced more -> more certain
scientific method
observation->hypothesis->prediction->experiment->analysis-»
fact = data + theory
inductivism
observations->theory->hypothesis->more observations->laws
“all swans are white”
falsificationism
refutable statement deduced from theory, experiment set up to falsify not confirm theory
strict testing
theory can be changed & tested again
Kuhn’s cycle of revolutions
“science progresses funeral by funeral”
preparadigm period->normal science->anomaly->crisis->revolution->normal science
observer effects
our “reality” manifests in a specific way based on our interaction with it
experimentation assumes that only factor affecting behaviour is objectively manipulated variables but this is incorrect
experimenter bias
exposure of rodents to male experimentors produces pain inhibition (pheromones)
what is true behaviour?
demand characteristics
try to do what they think researchers want to see to please them
demand characteristics are cues in experiment that convey hypothesis
Milgram experiment
participants willing to kill when following commands from perceived authority figure but many problems (demand characteristics, artificiality, representativeness)
representativeness
research predominantly on male while North-American psychology undergraduates (WEIRD) - not representative of humanity
doubtful whether results can be reproduced in other populations
artificiality
bizarre tasks in research labs not representative of real life situations
result of reducitonist approach
also applies to animals!
can try field study without being noticed
descriptive methods
naturalistic observation
Hawthorne effect & experiment
several workplace conditions manipulated but productivity increased as long as observed, regardless of light, pay, etc
reactivity
change of behaviour when know observed
experimenter expectancy effect & experiment
students told to train rats that are either genetically good or bad at running mazes, bias of experimenter changes outcome (give off cues based on expectation that affect behaviour)
response bias
tendency for participants to respond innaccurately to self-report questions
not passive, actively integrate multiple sources of information to generate response in given situation
correlation & correlation studies
describe & predict relationship between variables, not causal
directionality problem
direction of causation in correlational study ambiguous
positive correlation
variables move in same direction
negative correlation
inverse relationship
zero correlation
not predctably related
third variable
correlation may be due to third, unmeasured, variable
experimental method
test prediction based on hypothesis
interested in explanations (C&E)
manipulate variable(s) while maintaining ceteris paribus
experimental gorup
treatment level of IV
control group
receive no/unrelated treatment
comparison level of IV
confound(ing variable)
anything affecting DC that varies unintentionally caries between IVs eg age, experience
internal validity
quality of experiment: if observed effects are due to IV and not confounds
external validity
degree to which results can be generalised (tricky balance with internal validity)
descriptive statistics
measurements -> series of data points - distribution/spread of data
frequency distribution
number of data points in each bin
central tendency
number which best represents distribution
but which one depends on shape
mode
most often
median
50% above, 50% below
mean
sum & divide
normal distribution
symmetrical bell curve, most variables in nature approximate
mode=median=mean
large sample needed for good Gaussian distribution
variability
how much data points vary from each other & mean, how wide the spread
how well central tendency describes distribution
standard deviation
average distance of data points from mean
68% in 1 SD
95.5% in 2 SDs
99.7% in 3 SDs
logical empiricism
science rests on objective empirical observations that are independent of observer and theory
statistical significance
is difference meaningful enough to attribute to treatment? (dif between groups bigger than can be expected by chance)
null hypothesis H0 - no difference (try to reject using T-test)
experimental hypothesis H1
empirical structuralism
external validity does not exist
observations only valid within its specific context and theoretical framework you approach it with
argument against positivism (observations are objectively true)
realm of applicability of theory expands as you continue testing
