independent variables) (Li, 1975). The proportion of variance in the response variable explained by variance Knowledge of interrelationships between grain yield and its conin the predictor variable (partial coefficient of determitributing components will improve the efficiency of breeding programs through the use of appropriate selection indices. Previous path analy-nation) is the square of the path coefficient. ses studies in maize (Zea mays L.) treated yield components as first-In most studies involving path analysis, researchers order variables. The present study, based on evaluation of 90 expericonsidered the predictor characters as first-order varimental maize hybrids (comprising one diallel and one line ϫ tester ables to analyze their effects over a dependent or reset) at two locations in India, utilizes a sequential path model for sponse variable such as yield (Xu, 1986;Han et al., 1991; analysis of genetic associations among grain yield and its related traits Simon, 1993;Agrama, 1996;Board et al., 1997; Kumar by ordering the various variables in first-, second-, and third-order et al., 1999). This approach might result in multicollinpaths on the basis of their maximum direct effects and minimal collinearity for variables, particularly when correlations earity. The sequential path model showed distinct advantages over the among some of the characters are high (Hair et al., 1995; conventional path model in discerning the actual effects of different Samonte et al., 1998). There may also be difficulties predictor variables. Two first-order variables, namely 100-grain weight and total number of kernels per ear, revealed highest direct effects in (i) interpretation of the actual contribution of each on total grain weight (p ϭ 0.74 and p ϭ 0.78, respectively), while ear variable, since the effects are mixed or confounded belength, ear diameter, number of kernel rows, and number of kernels cause of collinearity (Hair et al., 1995), and (ii) suppleper row were found to fit as second-order variables. All direct effects mentation of unique explanatory predictions from were found to be significant, as indicated by bootstrap analysis. Test additional variables. Samonte et al. (1998) adopted a for the goodness-of-fit revealed that the sequential path model prosequential path analysis for determining the relationvided better fit to various datasets analyzed in the study. Correlations ships among yield and related characters in rice (Oryza between the predicted values of various response variables in the sativa L.) by organizing and analyzing various predictor second season dataset based on the path coefficients of the first season variables in first-, second-, and third-order paths. Howwere high, except for ear length and number of kernels per row. The ever, the collinearity of predictor variables was not tested applicability of the model has been confirmed through analysis of two additional datasets during 2000. The results indicated the utility for their agronomic performance during the monsoon season