[email protected].

ph
Spearman Rank-Order Correlation Coefficient
•This known as Spearman’s rho. The basic
logic underlying rho is that it tries to
predict the ranking of pairs of cases on the
dependent variable given their ranking on
the independent variable. However, it
makes use of the longer scale.
•This is a very simple and quick method
when there the paired data expressed in
ranks are less than 30.
Spearman Rank-Order Correlation Coefficient
•Spearman’s rho of the ranks of the two
variables is used to determine the degree
of relationship.
 Remember that Pearson’s r is
appropriate to use when data are
interval or ratio scale. For ordinal data,
Spearman Rho () is used.
But Spearman Rho () is
considered as a very special type of
Pearson’s r .
The formula is:

 6 d 2 
 ( rho ) = 1 −  
(
 n n − 1
2
) 

Where:
1 and 6 = constant
d = the difference in ranks and
n = the number of pairs
Example 1:
An instructor is interested in whether
the heavy use of formal exams as a form of
assessment is biased against students who
might perform better under different exam
conditions, such as verbal presentations. A
group of 15 students is selected and each
student is assessed in terms of their formal
examination skills. These 15 students is
selected and each rank-ordered on each of
these variables as indicated table below
with the calculations needed for rho.
 Rank on Rank Difference
Student Rank on Exam d squared
Presentation d
1 4 15 -11 121
2 6 3 3 9
3 9 14 -5 25
4 12 9 3 9
5 3 10 -7 49
6 13 11 2 4
7 5 6 -1 1
8 1 4 -3 9
9 14 8 6 36
10 2 1 1 1
11 10 2 8 64
12 7 5 2 4
13 15 7 8 64
14 8 12 -4 16
15 11 13 -2 4
Substituting these data into the equation for
rho:
 6 d 2 
 ( rho ) = 1 −  
 n ( n − 1) 
2

 6 ( 416 ) 
= 1−  
15 (15 − 1) 
2

= 0.26
 This indicates a weak, negative
association between the two types of
skills. The instructor might therefore
conclude that exams are not a good
indicator of other forms of learning skills:
• Students perform poorly in the exams
might perform well in verbal
presentations.
• Similarly, students who do well in the
exams might not relatively do all that
well when other skills are required.
 A mixed form of assessment methods
might give a better indication of students’
learning.
Example 2:
A group of BA students which called
themselves Unangosaurs Group made a
study about a degree of satisfaction of BSE
and BA students regarding their professors’
teaching expertise. One question asked of
the respondents was about the
characteristics they prefer most of their
professors in NEU. This question allowed
multiple responses which are presented in
the table below.
 BSE BA
Characteristics Preferred
f Rank f Rank
Approachable 229 1 227 1
Patient 161 6 166 5
Fair 179 5 150 6
Lively 187 4 237 3
Strict 19 9 98 9
With sense of humor 199 3 248 2
Has mastery of the subject matter 205 2 218 4
Reasonably strict/disciplinarian 93 8 101 8
Consistent 121 7 114 7
Total 1393 1609
 The researchers wanted to find out
if the characteristics of professors
preferred by BSE students do not
differ from the characteristics of
professors preferred BA students.
What they did was to assign ranks to
each attribute, and then they took the
correlation coefficient between the
rankings of the responses of the two
groups of students. Thus spearman’s
rho was used.
 BSE BA d
Characteristics Preferred d squared
Rank Rank
Approachable 1 1 0 0
Patient 6 5 1 1
Fair 5 6 -1 1
Lively 4 3 1 1
Strict 9 9 0 0
With sense of humor 3 2 1 1
Has mastery of the subject matter 2 4 -2 4
Reasonably strict/disciplinarian 8 8 0 0
Consistent 7 7 0 0
Total 8
Substituting these data into the equation for
rho:
 6d 2 
 ( rho ) = 1 −  
(
 n n − 1
2
) 
 6 (8) 
= 1−  
( )
 9 9 − 1 
2

48
= 1− = 0.93
720
 You get  = 0.93, which shows a high
correlation. But your work does not end
here. As what you have been doing with the
means, proportions, and Pearson’s r, you
have to test if “” is significant, and only
then you can conclude that the two groups
do not differ significantly on the
characteristics of professors they like most.
To test Spearman’s Rho (), follow the
same procedure as in testing the significance
of Pearson’s “r”. Below is the 5-step solution
using  = 0.05.
5 – step solution:
1. H0: = 0; There is no correlation between
the rankings of the characteristics of
professors preferred most by BSE and BA
students.
Ha:  ≠ 0; There is correlation between the
rankings of the characteristics of
professors preferred most by BSE and BA
students.
2. : 0.05; -computed = 0.93; -tab = 0.683
3. Direction Rule: Reject H0 if
 - comp (0.93)   - tab (0.683).
5 – step solution:

4. Decision: Reject H0 because 0.94 > 0.683