You are on page 1of 64

Special Correlation

Methods

Biserial Correlation

Point Biserial Correlation

Tetrachoric Correlation

Phi-Coefficient Correlation
• The value of 0 indicates that the absence of any
relationship between the variables.

• The + and - signs indicate the direction of relationship


between the two variables.

•  Positive sign shows that the variables change in same


direction while the negative sign shows that the variables
change in opposite directions.
Biserial Correlation

• Used when one variable is continuous and the other


is artificially dichotomized.
• Normality of distribution.
• A large N.
Formula for calculating Biserial r

where,

• rbis = Biserial correlation coefficient 


• Mp = Mean of the 1st group
• Mq = Mean of the 2nd group
• p = proportion of the 1st group
• q = proportion of the 2nd group
• u or y = height of the normal curve ordinate dividing the two parts p and
q
• σ = Standard deviation of the entire group
Calculation
16
16
16
16
16
16
16
16
5
Practice
Scores Group 1 Group 2

80-84 3 5
75-79 5 7

70-74 7 10

65-69 10 15

60-64 6 17

55-59 5 0

50-54 4 6
Point Biserial Correlation

• used to measure the strength and direction of the association


that exists between one continuous variable and one
truly/naturally dichotomous variable.

• When items are scored simply as 1 if correct & 0 if incorrect,


i.e., as right or wrong.

• Useful in the analysis of the items of a test, i.e., in item-test


correlations.
• df = N – 2
= 100 – 2 = 98

• Check Table 25 to test the significance of point


biserial r at 0.5 & 0.1 levels of significance.
• As the value of rpbis (0.52) exceeds .197 at .05 & .256 at .01 levels

of significance, it can be taken to be significant.


Scores P Q
Practice
95-99 1 0
90-94 1 1
• Calculate point
biserial correlation 85-89 6 0
for the given data. 80-84 11 2
75-79 6 4
70-74 9 6
65-69 3 8
60-64 2 3
55-59 1 2
50-54 - 6
45-49 - 2
40-44 - 3
35-39 - 1
30-34 - 1
Total 40 39
Tetrachoric Correlation

• Used when both variables are continuous and


artificially dichotomized.

• Example – Social maturity & Intelligence

Socially mature Socially immature Above avg. Below avg.


2x2 Contingency table

X Total

B A A+B
Y
D C C+D

Total B+D A+C A+B+C+D


Formula
Calculation
• Divide AD by BC.

• If BC > AC, then divide BC by AD & attach a –ve sign


to rt.

• Enter Table K with this ratio and read rt directly.


Practice

Q. Find Tetrachoric r in test evaluation of 125 college juniors.

X Variable
Non-science
Majors Science Majors

Above test mean 30 44


Y Variable

Below test mean 36 15


Phi-coefficient Correlation

• Used when both variables are continuous and


truly/naturally dichotomized.

• It is perhaps most useful in item analysis when we


want to know the item-item correlations.
2x2 Contingency table

X Total

B A A+B
Y
D C C+D

Total B+D A+C A+B+C+D


Calculation

• Two items X and Y are part of a test of 100 items. Item X is passed by 100 and failed
by 125 students in a group of 225. Item Y is passed by 135 and failed by 90 in the
same sample. Find the correlation between X and Y.

Item X

Failed Passed Total

55 80 135
Passed B A A+B
Item Y

70 20 90
Failed D C C+D

125 100 225


Total B+D A+C A+B+C+D
Formula
• χ² = Nɸ²

• χ² = 225 x (.36)²

• χ² = 29.2

• Since the value of χ² exceeds .01 level of


significance, it can be taken to be highly
significant. Therefore, the null hypothesis is
rejected.
Practice

Q. Find Phi-coefficient r between variables X and Y.

X Variable
Sub-groups No Yes

Yes 30 44
Y Variable

No 36 15

You might also like