skip to main content skip to footer

Automated Hypothesis Tests and Standard Errors for Nonstandard Problems. (Revised)

Lord, Frederic M.
Publication Year:
Report Number:
ETS Research Bulletin
Document Type:
Page Count:
Subject/Key Words:
National Science Foundation (NSF), Computer Software, Data Processing, Error of Measurement, Hypothesis Testing, Mathematical Models, Statistical Analysis


Faced with a nonstandard, complicated practical problem in statistical inference, the applied statistician sometimes must use asymptotic approximations in order to compute standard errors and confidence intervals and to test hypotheses. This usually requires that he or she derive formulas for one or more statistics. He or she must then compute the numerical value of an estimate of some function of these variances and covariances. If a statistic is a nonlinear function of more than two or three sample statistics, the mathematical derivation of the necessary variance (and covariance) formulas may be burdensome, or even prohibitive. The purpose of the present paper is to call attention to computer program LASAHT that computes estimated asymptotic sampling variances and covariances numerically and carries out hypothesis tests without need for the statistician to derive formulas for them. (Author/RC) (18pp.)

Read More