Correlational Research

Correlational Research
• The purpose of correlational research is
to discover relationships between two
or more variables.
• Relationship means that an individuals
status on one variable tends to reflect
his or her status on the other.
 Correlational Research
• Helps us understand related events, conditions,
and behaviors.
– Is there a relationship between educational levels of
farmers and crop yields?
• To make predictions of how one variable might
predict another
– Can high school grades be used to predict college
grades?
 Correlational Research
• To examine the possible existence of
causation
– Does physical exercise cause people to
lose weight?

CAUTION:
CAUTION:In Incorrelational
correlationalresearch
researchyou
youCAN
CAN
NOT
NOTabsolutely
absolutelysay
sayonce
oncevariable
variablecauses
causes
something
somethingtotohappen.
[Link]
Thiscan
canonly
onlybe
bedone
done
through
throughexperimental
experimentalresearch.
[Link]
Youcan
cansay
sayone
one
variable
variablemight
mightcause
causesomething
somethingelse
elseto
tohappen.
happen.
 Where does the data come
from for correlational
research?
• Surveys
• Scores on various tests or rating scales
• Demographic information
• Judges or expert ratings
 Correlational Research
Process
• Variables to be study are identified
• Questions and/or hypotheses are stated
• A sample is selected (a minimum of 30 is
needed)
• Data are collected
• Correlations are calculated
• Results are reported
 Terminology
• “Predictor” variable – the variable(s)
that are believed to predict the
outcome.
– Could be called an independent variable
 Terminology
• “Criterion” variable – the variable to be
predicted, the outcome
– Could be called the dependent variable
 Terminology
• Is level of education (predictor
variable) related to family income
(criterion variable)?
• Do people who eat more eggs
(predictor variable) have higher
cholesterol levels (criterion variable)?
 Which correlation to use?
Phi
correlation
Pearson
Spearman Product
rho Moment

Biserial
Correlation
Kendall
tau
 Pearson Product-Moment
Correlation
• Used when both the criterion and
predictor variable contain continuous
interval data such as test scores, years
of experience, money, etc.
 Examples of when to use the
Pearson Correlation!
Predictor Variable Criterion Variable
Years of Experience in Job Satisfaction score
Extension
Family Income EOC Test Scores
Distance from Krispy Weight
Kreme donut shop.
 Point Biserial Correlation
• When the predictor variable is a natural
(real) dichotomy (two categories) and
the criterion variable is interval or
continuous, the point biserial
correlation is used.
 Examples of when to use the
Point Biserial Correlation!
Predictor Variable Criterion Variable
4-H member or Not Leadership score
National Board EOC Test Scores
Certified or Not
Male or Female Salary
 Biserial Correlation
• When the predictor variable is an
artificial dichotomy (two categories)
and the criterion variable is interval or
continuous , the biserial correlation is
used.
 Examples of when to use the
Biserial Correlation!
Predictor Variable Criterion Variable
Tall or short Leadership score
Good looking or ugly Salary

More Popular or Less Self Concept Score

Popular
 Phi Correlation
• When the both the predictor and criterion
variables are natural dichotomies (two
categories), the phi correlation is used.
• If the dichotomies are artificial, the
tetrachoric correlation is used. This is rarely
the case in educational research
 Examples of when to use the
Phi Correlation!
Predictor Variable Criterion Variable
4-H member or not On School Honor Roll
or Not
Board Certified Member of Teachers
Teacher or Not Organization or Not
Male or Female Full Professor or Not
Spearman rho and Kendall tau
• When the both the predictor and
criterion variables are rankings, use
either the Spearman rho or Kendall tau
correlation.
– More than 20 cases – Spearman rho
– Less than 20 cases – Kendall tau
 Examples of when to use the
Spearman rho or Kendall tau
Correlation!
Predictor Variable Criterion Variable
Ranking of students in Ranking of students
high school graduating according to number
class of scholarship offers
Ranking of discipline Ranking of discipline
problems in 1990 problems in 2002
 Correlation Table
Predictor Variable Criterion Variable Correlation to Use
Interval (Continuous) Interval (Continuous) Pearson
Real Dichotomy Interval (Continuous) Point Biserial
Artificial Dichotomy Interval (Continuous) Biserial
Real Dichotomy Real Dichotomy Phi
Artificial Dichotomy Artificial Dichotomy Tetrachoric
Ranking Ranking Spearman rho for 20
or more rankings
Ranking Ranking Kendall’s tau for less
than 20 rankings
 Other Correlations
• You can perform multiple correlations using
such approaches as partial correlation,
multiple regression, discriminant analysis,
and factor analysis.
 Correlation Principles to
Remember
• For each individual in the research,
there must be at least two measures, or
it will be impossible to calculate a
correlation.
 Correlation Principles to
Remember
• A correlation may be statistically
significant (it didn’t happen by chance)
but be weak or low which means it is
nothing to get excited [Link] has no
practical significance.
More Principles to Remember
• A correlation is reported as r such as
r=.36.
More Principles to Remember
• The statistical probability is reported as p.
– Some researchers report the probability of the
correlation happening by chance was p>.05 (more
than 5 out of 100) or p<.05 (less than 5 out of 100) –
we hope for the later as researchers
– Other researchers report the actual probability; p=.03
• The first approach was used before the age of computers
• Either approach is acceptable.
More Principles to Remember
• In reporting correlations in research
reports you report both the r value and
the p.
 Correlations
• Correlations can range from –1.00 to 1.00
– A 1.00 is a perfect positive correlation
• As one variable increases, so does the other
– A -1.00 is a perfect negative correlation
• As one variable increases, the other variable decreases
– A .00 correlation indicates no correlation
• There is no relationship between one variable and
another
 Interpretation of the Strength
of Correlations
• .00 - .20 – Very Weak
• .21 - .40 – Weak Different
Differentstatisticians
statisticians
may
mayhave
havesimilar
similarbutbut
• .41 - .60 – Moderate slightly
slightlydifferent
different
• .61 - .80 – Strong scales.
scales.
• .81 – 1.00 - Very Strong
 Correlations
• Scatter plots are often used to depict
correlations
6000

5000
Calories per day

4000 This
Thischart
chart
shows
showsaa
3000
strong
strong
2000
positive
positive
1000 correlation
correlation
0
100 150 200 250 300 350 400
Weight
 Correlations
• Scatter plots are often used to depict
correlations
160 This
Thischart
chart
Minutes of Exercise per day

140
shows
showsaa
120
strong
strong
100
negative
negative
80
correlation
correlation
60
40
20
0
100 150 200 250 300 350 400
Weight
 Correlations
• Scatter plots are often used to depict
correlations
45
Miles from Krispy Creme

40
35
30 This
Thischart
chart
25 shows
shows
20 virtually
virtuallyno
no
15
correlation
correlation
10
5
0
100 150 200 250 300 350 400
Weight
 How can I calculate
correlations?
• Excel has a statistical function. It calculates
Pearson Product Moment correlations.
• SPSS (a statistical software program for
personal computers used by graduate
students) calculates correlations.