วันพุธที่ 9 ตุลาคม พ.ศ. 2556

PLS-SEM

Partial Least Squares-SEM

ลักษณะของ PLS

1. Partial Least Squares is just like Principal Component (PC) Regression except in how the  component scores are comput
 2. PC regression = weights are calculated from the covariance matrix of the predictors- PLS = weights reflect the covariance structure between predictors and response
3. While conceptually not too much of a stretch, it requires a more complicated iterative algorithm
           4. Like in regression, the goal is to maximize the correlation between the response(s) and component scores
ภายใต้กฎที่ว่า
 
กล่าวโดยสรุปการประเมินผล PLS-SEM
Reflective Measurement Models
Formative Measurement Models
  1.     Internal consistency (composite reliability) ร่วมกับ Cronbachs Alpha) 
2.     Indicator reliability
3.     Convergent validity (AVE > 0.5)
4.     Discriminant validity (Fornell-Lacker Analysis)
        1.  Convergent validity (R2 แต่ละตัว)
          2. Collinearity among indicators (VIF)
        3.   Significance and Relavance od outer weights
1.     Coefficients of determination (R2 = จำนวนตัวแปรตาม)
2.     Predictive relevance (Q2)
3.     Size and significance of path coefficients
4.     f2 effect sizes
5.     q2 effect sizes (if necessary)
Source: Hair, J. F. Jr., Hult, G. T. M., Ringle, C.M.,Sarstedt, M. (2013) A Premier Partial Least Squares Structural Equation Modeling (PLS-SEM). London: Sagepublications.com

ไม่มีความคิดเห็น:

แสดงความคิดเห็น