Correlational

Research
 Correlational Research
n  When we measure two or more variables and try to determine the
degree of relationship between them we are conducting a correlational
study.
n  A correlational study can be integrated with all descriptive studies, can
be added to the beginning or end of an experimental study.
n  Sir Francis Galton was the pioneer in intelligence studies and invented
correlation. He studied the relation between human ability and
inheritance. He organized the data in rows and columns, and first did
scatter plots.
n  He plotted the mean height of mother and father and the child. He
then drew a straight line through the scatterplot which he called the
regression line.

n  After reading
G a l t o n ’s w o r k , a
young statistician,
Pearson first
developed a formula
for the coefficient of
correlation=r
 Correlational Research

n  In
correlational studies, variables are
not manipulated, they are only
measured.
 What is a correlation?
n  Positive correlation,
n  Negative correlation,
n  scatter plots,
n  Significance of correlation, high-low correlation
n  Linearity assumption (e.g. Arousal and achievement)
 What is a correlation?
n  Restricting the range: R=.70
Decreasing the range in
one or both variables
decreases the correlation
(e.g. r=.70 and .26 in the
graphs)

n  Regression analysis gives

opportunity to make
predictions (e.g. ÖSS and
ability tests) R=.26
 Correlation between
IQ ve income is
significant; each point
increase in IQ test
scores is associated
with $202 to $616
Correlation between
IQ and wealth is not
significant.

Zagorsky, Jay. 2007. Do you have to be smart to

be rich? The impact of IQ on wealth, income
and financial distress. Intelligence 35: 489-501.
 Genomic Galton’s method: relies simply on the
average deviation of the two parents from the
explanation population average (corrected for age and sex). This
measure was calculated for a smaller set of 550
individuals for whom parental height data was
available.

The blue line shows the slope of The correlation between this
best fit for the data; In fact, predictor and measured height is
genotype score explained just much stronger than for the
3.8% of the variance in height, genotype score above: this
measure predicts around 40%
of the variance in height, 6 to
10 times more than is explained
by genotype score.
Yurii S Aulchenko, Maksim V Struchalin, Nadezhda M Belonogova, Tatiana I Axenovich, Michael N Weedon, Albert Hofman, Andre G
Uitterlinden, Manfred Kayser, Ben A Oostra, Cornelia M van Duijn, A Cecile J W Janssens, Pavel M Borodin (2009). Predicting human
height by Victorian and genomic methods European Journal of Human Genetics DOI:
 Correlation and causality:
n  In a study a positive correlation was found between
HAIR LOSS and HEART ATTACK. The advertisers
decided:
n  “To avoid heart attacks
Use hair loss treatment medicines!!!

n  In experimental studies «the independent variable is

manipulated when all else is held constant»; thus the
results can be attributed to the independent variable.
n  But in correlational research «all else is held
constant» is missing. This lack of control makes
correlational studies impossible to make a causal
conclusion. These are referred as directionality
problem and the third variable problem.
Problems in correlational research
n 1. Directionality:
Is there a relationship between TV watching
and aggression? How would you conduct such
a study?
–  A study found a positive correlation
between aggression and TV watching.
–  What is the direction?

–  Violent TV watching aggression

–  Aggression watching violent TV
 n  To resolve the problem they used cross
lagged panel correlation procedure (çapraz
atlamalı) 3. Violent TV
Violent TV preference in
preference in .05 13th grade
3rd grade

.01

.21 -.05

.31

Agression in .38 Agression in

3rd grade 13th grade
 Third variable problem
n  As we have mentioned before when the number
of churches increased, crime also increased.
When we find high correlation between two
variables how are we to interpret?
n  Is there a third variable?
n  In the example about TV and aggression, may
there be a third variable?
–  E.g. Parents can be aggressive and be a model for the
kids, at the same time children watch TV to avoid
contact with them.
–  Or lack of verbal fluency etc.
n  To evaluate the third variable partial correlation
procedures may be used.
Watching violent r=.43** Aggressive
TV behavior

R=.14
Watching violent Aggressive
TV behavior

Partial
r=.55** r=.46*
correlation

Parent’s Aggressive
behavior
MEDIATOR
(aracı)
DEĞİŞKEN
 If there are so many problems, why
are we conducting correlational
studies?
1. Practical reasons:
n  Sometimes we cannot manipulate the variables
(personality, age, gender etc.) In these circumstances
information about the relations between variables is
valuable.
n  In some studies prediction is targeted (not causal
explanation).
–  E.g. which personality characteristics are related with success?
–  Which tests predict safe driving?
–  What are the predictors of postpartum depression?
2. Ethical reasons:
§  Some studies can not be conducted experimentally
(e.g. Broca examined patients with speech disorder
and brain damage after death)
§  This is why animals are used as subjects in many
experimental studies investigating the relationship
between brain and behavior. In humans we use
correlational research for it.
Some examples of correlational
research
–  Psychological testing: Reliability: spit-half reliability; predictive
validity or criterion validity
–  Personality studies: e.g. We can find out a high correlation
between introversion and anxiety; negative correlation
between intraversion and sociability.
•  Seligman found out that depression and pessimism are highly
correlated. He proposed cognitive therapy to make a more positive
explanatory style and alleviate depression.
–  Cognitive neuroscience and imaging studies: fMRI studies
show that a region in the brain is active in the process, not
that it is necessary for the process.

–  Nature/nurture issue (Kalıtım/çevre): Comparing identical

twins reared together or separately: same genetics, different
environment
 Çift sayısı R for IQ
Birlikte büyütülmüş tek 4,672 .86
y. ikizleri
Ayrı büyütülmüş tek y. 65 .72
ikizleri
Birlikte büyütülmüş çift 5,546 .60
y. ikizleri
Birlikte büyütülmüş 26,473 .47
kardeşler
Ayrı büyütülmüş 203 .24
kardeşler
Kuzenler 1,176 .15
Bouchard ve McGue, 1981