EQUIVALENCE OF DIRECT , INDIRECT AND SLOPE ESTIMATORS OR AVERAGE DERIVATIVES

@inproceedings{StokerEQUIVALENCEOD,
  title={EQUIVALENCE OF DIRECT , INDIRECT AND SLOPE ESTIMATORS OR AVERAGE DERIVATIVES},
  author={Thomas M. Stoker}
}
If the regression of a response variable y on a k-vector of predictors x is denoted g(x)=E(yix), then the average derivative of y on x is defined as S=E(g'), where g'-ag/ax. This paper compares the statistical properties of four estimators of : a "direct" estimator , formed by averaging pointwise kernel estimators of the derivative g'(x); an "indirect" estimator f proposed by Hardle and Stoker(1987), based on averaging kernel density estimates: and two "slope" estimators d and df, which are… CONTINUE READING

Citations

Publications citing this paper.

Similar Papers