Descriptive Statistics: Correlation Types of Correlations
uDescribes the relationship between two u Variable X increases
or more variables. u Variable Y increases
uDescribes the strength of the relationship in Positive Correlation
terms of a number from -1.0 to +1.0.
Value ranging from .00 to 1.00
uDescribes the direction of the relationship
as positive or negative. Example: the more you eat, the more
weight you will gain
Types of Correlations Types of Correlations
u Variable X decreases u Variable X increases
u Variable Y decreases u Variable Y decreases
Positive Correlation Negative Correlation
Value ranging from .00 to 1.00 Value ranging from -1.00 to .00
Example: the less you study, the lower Example: the older you are, the less
your test score will be flexible your body is
Types of Correlations Correlation Strength
u Variable X decreases
u Variable Y increases u .00 - .20 Weak or none
Negative Correlation u .20 - .40 Weak
Value ranging from -1.00 to .00 u .40 - .60 Moderate
Example: the less time you study, the u .60 - .80 Strong
more errors you will make u .80 - 1.00 Very strong
1
Positive or Negative? Caution!
u IQ and reading achievement u Correlation does not indicate causation.
u Anxiety and test scores u Correlation only establishes that a relationship
u Amount of calories consumed and weight gain. exists; it reflects the amount of variability that
u Amount of exercise and weight gain is shared between two variables and what they
have in common.
u Reading achievement and math achievement
u Foot size and math ability Examples:
u Amount of ice sold and number of bee stings.
u SAT scores and GPA in college.
Correlation: IQ and GPA
A Picture of Correlation
u IQ GPA
u A scattergram or scatter plot visually represents u 110 2.5
140
a correlation u 140 4.0 130
u The X axis is on the horizontal u 80 1.0 120
u The Y axis is on the vertical. u 100 2.0 110
IQ
100
u 130 3.5 90
u 90 1.5 80
70
u 120 3.0 0 1 2 3 4
u 70 .5 GPA
Correlation: IQ and Errors Correlation: IQ and Weight
u IQ u IQ Weight
Errors u 120 170
u 80 14 140 u 100 160 140
130
u 120 6 130
u 70 120 120
120
u 100 10 110
u 140 130
IQ
110 100
IQ
u 90 12 100 u 90 200 90
u 130 4 80
u 130 110
90
70
80
u 110 8 u 80 150 110 120 130 140 150 160 170 180 190 200
70 Weight
u 140 2 0 5 10 15 u 110 140
u
Errors
70 16
2
Caution Coefficient of Determination
u Do not interpret the coefficient of u Square the coefficient of correlation.
correlation as a percent!
u r = .50
u r 2 = .25 or 25 %
u If you want to know the percentage of
u Twenty five percent of the variance in one
variance in one variable that is accounted
for by the variance in the other variable, variable can be accounted for by the
compute the coefficient of determination variance in the other variable.
Example: Coefficient of Factors Influencing
Determination Correlation
u The correlation between IQ and reading at u When interpreting the correlation
its highest level: r = .60 coefficient, always consider the nature of
the population in which the two variables
u r2 = .36 or 36 % were observed.
u The correlation coefficient will vary from
u Thirty six percent of reading achievement is one population to another.
related to IQ. Reading achievement and IQ
share 36% of the variance.
Factors Influencing Factors Influencing
Correlation Correlation
u The relationship of variables may differ u Higher correlations are expected in a
from population to population. heterogeneous population than in a
u Example: Physical prowess and age are homogeneous one.
correlated between the ages of 10 and 16. u Example: In elementary and high school, there
u Example: Physical prowess and age are not is a positive correlation between height and
correlated between the ages of 20 and 26. success in basketball.
u Example: In the pros, there is no such
correlation.
3
Factors Influencing Choosing Correlation
Correlation Formulas
u There may be a correlation between two u X is nominal data
variables not because there is a relationship u Y is nominal data
between them but because both are related Correlation Formula: Phi coefficient
to a third variable. Example: Correlation of sex (male/female)
u Example: Average teacher salary for 20 years and choice of car color (red, black, blue,
and the cost of hard liquor. white, silver)
Choosing Correlation Choosing Correlation
Formulas Formulas
u X is nominal data u X is nominal data
u Y is ordinal data u Y is interval data
Correlation Formula: Rank biserial coefficient Correlation Formula: Point biserial
Example: Correlation of race and rank in school
Example: Correlation of sex and GPA
Choosing Correlation Choosing Correlation
Formulas Formulas
u X is ordinal data u X is interval
u Y is ordinal or interval data (interval data u Y is interval
must be converted to ordinal) u Correlation Formula: Pearson correlation coefficient
u Correlation Formula: Spearman rank coefficient u Example: Age and the number of minutes it takes to
u Example: Correlation between rank and GPA
solve a problem