Structural equation modeling with latent variables to establish relationship between yield and its components for major crops of Haryana
Loading...
Date
2021-01
Authors
Journal Title
Journal ISSN
Volume Title
Publisher
CCSHAU, Hisar
Abstract
Structural equation modeling (SEM) is a powerful multivariate statistical analysis technique
used to analyze structural relationships with a wide range of applications in the plant sciences using
measured variables and latent constructs. This method is preferred over other methods because it
estimates the multiple and interrelated dependencies in a single analysis. SEM provides robust
estimates of path coefficients that characterizes complex phenomenon and biological processes. The
SEM of bread wheat has been hypothesized on the basis of the four latent variables viz. physiological
(ξ1), morphological (ξ2), fertility & quality (ξ3) parameters as exogenous latent constructs where as
grain parameter (η1) as endogenous latent construct as suggested by the preliminary exploratory factor
analysis. The final model has been assessed through fit indices viz. Chi square (16.25 at P-value
0.298), goodness of fit index (0.98), root mean square error approximation (0.031) and chi square
ratio (1.17). The SEM of basmati rice has been hypothesized on the basis of the four latent variables
viz. physiological (ξ1), morphological (ξ2) and fertility (ξ3) parameters as exogenous latent constructs
whereas grain parameter (η1) as endogenous latent construct as suggested by the preliminary
exploratory factor analysis. The final model has been assessed through fit indices like Chi square
(14.31 at P-value 0.426), goodness of fit index (0.99), root mean square error approximation (0.007)
and chi square ratio (1.02). The latent constructs in cotton are horizontal growth (ξ1) and
morphological (ξ2) parameters as exogenous where as biochemical (η1) and yield (η2) parameter as
endogenous latent construct. The final model of cotton has been assessed through fit indices like Chi
square (36.69 at P-value 0.484), goodness of fit index (0.99), root mean square error approximation
(0.000) and chi square ratio (0.99). The latent constructs in barley are phenological (ξ1) and grain (ξ2)
parameters as exogenous whereas yield (η1) parameter as endogenous latent construct. The final
model of barley has been assessed through fit indices like Chi square (29.18 at P-value 0.213),
goodness of fit index (0.98), root mean square error approximation (0.089) and Chi square ratio
(1.22).The structural equation model of pearl millet has been hypothesized on the basis of the four
latent variables viz. physiological (ξ1), morphological (ξ2) and fertility (ξ3) parameters as exogenous
latent constructs whereas yield parameter (η1) as endogenous latent construct as suggested by the
preliminary exploratory factor analysis. The final model has been assessed through fit indices chi
square (34.54 at P-value 0.302), goodness of fit index (0.96), root mean square error approximation
(0.047) and chi square ratio (1.15). The SEM model that fits well to the data indicated that there is a
positive influence of physiological and morphological parameters on the endogenous latent variable
yield whereas a positively highly significant influence of fertility parameter on the endogenous latent
variable yield was observed.