PLS for Path Diagrams Revisited, and Eztended
I extend the usual indices as linear compounds, by introducing proper indices and using concepts of fuzzy logic. A proposal is made to always study and report the full range of possible structural parameter values as determined by variations across all admissible indices. An appeal is made to use more structural information in the construction of indices. And finally, for a particular class of empirical applications, it is pointed out that proper indices for non-linear path models can also be obtained by combining Wolds mode B approach with a non-negative least squares routine.
linear indez fuzzy logic quadratic model with interaction simulation
Theo K. Dijkstra
University of Groningen, Economics & Econometrics & SNS Asset Management, The Netherlands
国际会议
The 6th International Conference on Partial Least Squares and Related Methods(第六届偏最小二乘及相关方法国际会议)
北京
英文
111-116
2009-09-04（万方平台首次上网日期，不代表论文的发表时间）