Path coefficient
Jump to navigation
Jump to search
Path coefficients are standardized versions of linear regression weights which can be used in examining the possible causal linkage between statistical variables in the structural equation modeling approach. The standardization involves multiplying the ordinary regression coefficient by the standard deviations of the corresponding explanatory variable: these can then be compared to assess the relative effects of the variables within the fitted regression model. The idea of standardization can be extended to apply to partial regression coefficients.
The term "path coefficient" derives from Wright (1921), where a particular diagram-based approach was used to consider the relations between variables in a multivariate system.[1]
See also
[edit | edit source]Notes
[edit | edit source]- ^ Dodge, Y. (2003) The Oxford Dictionary of Statistical Terms, OUP. Lua error in Module:Citation/CS1/Configuration at line 2172: attempt to index field '?' (a nil value).
References
[edit | edit source]- Shipley, B. (2000) Cause and Correlation in Biology: A User's Guide to Path Analysis, Structural Equations and Causal Inference Cambridge University Press. Lua error in Module:Citation/CS1/Configuration at line 2172: attempt to index field '?' (a nil value).
- Wright, S. (1921) "Correlation and causation", Journal of Agricultural Research, 20, 557–585.