Models as Instruments, With Applications to Moment Structure Analysis
The paper discusses some very general model to evaluate the quality of estimates, and of models that these estimates are based on. Methods are based on a simple geometrical argument, and on expansions of the loss functions around the estimate, the target, and the replication. We give both delta-method and Jackknife computational procedures to estimate the relevant quantities.
KeywordsLoss Function Specification Error Sample Moment Absolute Moment Multivariate Behavioral Research
Unable to display preview. Download preview PDF.
- 2.P. J. Bickel, C. A. J. Klaassen, Y. Ritov, and J. A. WellnerEfficient and Adaptive Estimation for Semipammetric ModelsJohn Hopkins University Press, Baltimore, MD, 1993.Google Scholar
- 7.J. de LeeuwModel Selection in Multinomial ExperimentsOn Model Uncertainty and its Statistical Im-plications (T. J. Dijkstra, ed.), Springer Verlag, New York, NY, 1988.Google Scholar
- 8.A.E. Gelfand, D.K. Dey, and H. ChangModel Distributions using Predictive Distributions with Implementation via Sampling-Based MethodsBayesian Statistics 4, Oxford University Press, New York, NY, 1992.Google Scholar
- 10.J. HurtAsymptotic Expansions for Moments of Functions of StatisticsSecond Prague Symposium on Asymptotic Statistics, Akademie, Prague, Czechoslovakia, 1978.Google Scholar
- 12.H. Linhart and W. ZucchiniModel SelectionWiley, New York, NY, 1986.Google Scholar