Path Coefficient Analysis Autumn
PATH COEFFICIENT ANALYSIS
Professor Dr. A. K. M. Aminul Islam, GPB, BSMRAU, Gazipur
Path Analysis
A path coefficient is simply a standardized partial regression coefficient and as such it
measures the direct influence of one variable upon another and permits the separation of
correlation coefficients into components of direct and indirect effects.
The concept of path analysis was originally developed by Sewell Wright in 1921, but the
technique was first used for plant selection by Dewey and Lu in 1959.
Main features of path coefficient analysis
Main features of path coefficient analysis are given below:
i. Path analysis measures the causes of association between two variables.
ii. Path coefficient measures the proportion of standard deviation of dependent
variable which is due to variation in independent variables.
iii. Analysis of path coefficient is based on all possible simple correlations among
various characters.
iv. It provides information about direct and indirect effects of independent variables on
dependent variables.
v. When independent variables are correlated, path coefficient analysis separates the
linear correlation between the dependent and independent variables into direct and
indirect effects of the independent variable.
vi. When independent variables are uncorrelated, path coefficients are the simple
correlation coefficient between the dependent and independent variables.
vii. Path coefficient has a direction, it may be positive or negative.
viii. Path coefficients may be greater or less than unity.
ix. It also estimates residual effects.
x. Path analysis helps in determining yield contributing characters and thus is useful in
indirect selection.
Path Diagram
In path analysis, a line diagram which is constructed with the help of simple correlation
coefficients among various characters included under study is referred to as path diagram.
It is constructed before estimation of direct and indirect effects. The dependent variable,
say yield is kept at one side and all independent variables on the other side. The nature of
the causal system is represented diagrammatically as follows:
Let us consider that the yield ‘Y’ in barley is the function (effect) of various components
(causal factors) like ears / plant (X1), grains / ears (X2) and 100-grain weight (X3) etc.
1
Path Coefficient Analysis Autumn
PX1Y or P14
X1
Y PX2Y or P24 rX1X2
(r12)
rX1X3
PX3Y or P34 X2 (r13)
rX2X3
PRY (r23)
X3
R
Fig. Path diagram of X1, X2 and X3 on Y
- P14, P24 and P34 are the path coefficients of X1, X2 and X3 on Y
- r12, r23 and r13 are the correlation coefficients between the respective variables.
- R is the residual effect and PRY is the path coefficients of R on Y.
Analytical Procedure for path coefficient analysis
Now the routes of correlations between Y and other variables are:
Table 1. Correlation coefficients between yield and its primary components in
barley
Characters Ears / plant Grains / ears 100-grain Grain yield
(X1) (X2) weight (X3) (Y) or X4
Ears / plant (X1) 1.00 (r11) r12 r13 r14
Grains / ears (X2) r12 1.00 (r22) r23 r24
100-grain weight (X3) r13 r23 1.00 (r33) r34
Grain yield (Y) or X4 r14 r24 r34 1.00 (r44)
Step 1: Construct the path coefficient equations or simulations of equations
a) Correlation coefficient between X1 and Y (r14)
Direct effect of X1 on Y = P14
Indirect effect of X1 on Y via X2 = r12 P24
Indirect effect of X1 on Y via X2 = r13 P34
r14 = P14 + r12 P24 + r13 P34
2
Path Coefficient Analysis Autumn
b) Correlation coefficient between X2 and Y (r24)
Direct effect of X2 on Y = P24
Indirect effect of X2 on Y via X1 = r12 P14
Indirect effect of X2 on Y via X3 = r23 P34
r24 = P24 + r12 P14 + r23 P34
c) Correlation coefficient between X3 and Y (r34)
Direct effect of X3 on Y = P34
Indirect effect of X3 on Y via X1 = r13 P14
Indirect effect of X3 on Y via X2 = r23 P24
r34 = P34 + r13 P14 + r23 P24
The whole system of relationship between the variables can be explained by the
following equations –
a) r14 = P14 + r12 P24 + r13 P34 ---------------- 1
b) r24 = P24 + r12 P14 + r23 P34 ---------------- 2
c) r34 = P34 + r13 P14 + r23 P24 ---------------- 3
Step 2: Solution of simultaneous equations
From simultaneous equations we can get the values of P14, P24 and P34 which are the
direct effect of X1, X2 and X3 on Y by putting or fitting the values of correlation
coefficients (rij) from the table 1. This can be done by the following three methods:
a) Elimination procedure
b) Matrix method
c) Modified approach
a) Elimination procedure:
i. Divide each equation by the coefficients of its own direct effect
r14 P14 + r12 P24 + r13 P34 ---------------- 4
1.0 = 1.0
r24 P24 + r12 P14 + r23 P34 ---------------- 5
r12 = r12
r34 P34 + r13 P14 + r23 P24
= ---------------- 6
r13 r13
3
Path Coefficient Analysis Autumn
ii. Subtract the equation 4 from 5 and 6
Equation 5 – Equation 4 ---------------- 7
Equation 6 – Equation 4 ---------------- 8
iii. Divide the equations by the coefficients of P24
Equation 7 ÷ Coefficients of P24 of equation 7 ---------------- 9
Equation 8 ÷ Coefficients of P24 of equation 8 ---------------- 10
iv. Subtract the equation 9 from equation 10
P34 = ?
v. Put the value of P34 into the equation 10
P24 = ?
vi. Put the value of P24 and P34 into the equation 6
P14 = ?
b) Matrix method:
The simultaneous equations 1, 2 and 3 expressed above may be presented in the
following matrix form:
r14 r11 r12 r13 P14
r24 = r21 r22 r23 . P24
r34 r31 r32 r33 P34
Or, A = B . C
Here the values of A and B are known. We have to calculate the value of C vector.
C= B– 1. A
Thus we require B– 1 before the estimating C vector
C= B– 1. A
4
Path Coefficient Analysis Autumn
P14 r11b r12b r13b r14
P24 = r21b r22b r23b . r24
P34 r31b r32b r33b r34
Where,
a) P14 = r11b r14 + r12b r24 + r13b r34
b) P24 = r21b r14 + r22b r24 + r23b r34
c) P34 = r31b r14 + r32b r24 + r33b r34
b) Modified approach:
There is a short cut and also easier procedure to calculate path values (Chaudhury et al.,
1980). This modified approach is same as pivotal condensation method of matrix
inversion.
[Ref. Chaudhury, B. D.; Yadava, T. P. and Paroda, R. S. 1980. A modified method for
path analysis. Genet. Iber., 32: 93 - 104.]
Step 3: Calculation of Residual effect
The individual effect can be obtained by the following formula:
1 = P2R4 + P214 + P224 + P234 + 2P14 r12 P24 + 2P14 r13 P34 + 2P24 r23 P34
P2R4 = {1 – (P214 + P224 + P234 + 2P14 r12 P24 + 2P14 r13 P34 + 2P24 r23 P34)}
PR4 = √{1 – (P214 + P224 + P234 + 2P14 r12 P24 + 2P14 r13 P34 + 2P24 r23 P34)}
The residual effect can also be obtained by the following formula:
1 = P2R4 + PIY rIY
In this example,
1 = P2R4 + P14 r14 + P24 r23 + P34 r34
P2R4 = {1 – (P14 r14 + P24 r23 + P34 r34)}
P2R4 = √{1 – (P14 r14 + P24 r23 + P34 r34)}
5
Path Coefficient Analysis Autumn
Step 4: Calculation of direct and indirect effects
a) Ears per plant (X1) and grain yield (Y):
Direct effect of X1 on Y = P14
Indirect effect of X1 on Y via X2 = r12 P24
Indirect effect of X1 on Y via X2 = r13 P34
Total effect (direct + indirect) = r14
b) Grains per ear (X2) and grain yield (Y):
Direct effect of X2 on Y = P24
Indirect effect of X2 on Y via X1 = r12 P14
Indirect effect of X2 on Y via X3 = r23 P34
Total effect (direct + indirect) = r24
c) 100 grain weight (X3) and grain yield (Y)
Direct effect of X3 on Y = P34
Indirect effect of X3 on Y via X1 = r13 P14
Indirect effect of X3 on Y via X2 = r23 P24
Total effect (direct + indirect) = r34
Step 5: Interpretation of results
a) If the correlation coefficient between a causal factor and the effect is almost equal
to its direct effect, then the correlation explains the true relationship and a direct
selection through this trait will be effective.
b) If the correlation coefficient is positive, but the direct effect is negative or
negligible, the indirect effect seems to be cause of correlation. In such situations,
the indirect causal factors are to be considered simultaneously for selection.
c) Correlation coefficient may be negative but the direct effect is positive and high.
Under these circumstances, a restricted simultaneous selection model is to be
followed, i.e. restrictions are imposed to nullify the undesirable indirect effects in
order to make use of the direct effect. Direct selection for such trait should be
practiced to reduce the undesirable indirect effects.
d) If the value of residual effect is moderate or high, it indicates that besides the
character studied, there are some other attributes which contribute for yield.
