Week 10 Flashcards
correlation studies
no manipulation
no independent or dependent variables
measuring to see if 2 or more variables are related
scatterplots
help us determine if relationships are linear
if related we can draw a line through the points
curve lines
a nonlinear relationship
such a a inverted U relationship
e..g the relationship between arousal and memory
positive relationship
both variables increase
negative relationships
one variable increases, other variable decreases
correlation coefficient
shows strength and direction of relationships
range from +1 to -1
+/- is the direction
1 is the strength
closer to 0
weaker correlation
what counts as a strong correlation is context dependent
Pearsons r and Spearman’s rho
both make the assumption that the relationship is linear
Pearsons r
data is interval/ratio
Spearman’s rho
data is ordinal
Pearsons r is
how far from the mean each participant falls
quantifies how far participants deviate from the mean on one variable predicts deviation from mean on other variables
Spearman’s rho is
ranks participants independently on each variable
quantifies the extent to which rank one predicts the rank on other variables
postive coefficient
postive relationship
negative coefficient
negative relationship
correlation coefficient is
a descriptive statistic
null hypothesis
no correlation
p value quantifies how much our observed sample correction differs from what we’d expect if the null was true
alternate hypothesis
can be one tailed or two tailed
what shouldn’t we do?
draw casual conclusions
what should we do?
consider how many correlations have been conducted
large datasets which lots of positive correlations produce spurious correlations
spurious correlations
not meaningful and don’t reflect a true relationship
random error
caused by unknown and unpredictable changes in the experiment
doesn’t effect the mean but changes the range of scores around the mean
systematic error
reproducible and consistent in the same direction
effects the mean
to reduce the error
repeat measurements
apply correction factors
pilot study
training the experimenter
training the participants
triangulation
triangulation
suing more than one measure of a particular behaviour
a group of measurement tools won’t have the same error
validity
how accurate is our research measuring the desired behaviour
internal validity
effects observed are due to independent variable not other factors
ecological validity
external
results can be generalised to other settings
population validity
external
results can be generalised to other people
predictive validity
external
whether our measure predicts other related behaviours
construct validity
external
are we measuring what we think we are measuring
face validity
on ‘face value’ it seems the measure is appropriate
reliability
consistency of our measure or results
internal reliability
applies only to questionnaires
do items thats purpose is to measure the same thing have consistent responses
external reliability
extent a measure varies from one use to another
test retest
same person fills out the same results it should be the same
random error in a questionnaire
natural small variations over time
systematic error in a questionnaire
social desirable responding