home *** CD-ROM | disk | FTP | other *** search
/ Apple Reference & Present…nuary (Partner) - Disc 2 / The Apple Reference and Presentations Library (Disc 2)(January 1994).iso / SAS Institute / US English / JMP Statistical / JMP-PPC / JMP Demo Help / JMP Demo Help.rsrc / PICT_4503_Lack of Fit (-Loglikelihood).png < prev    next >
Portable Network Graphic  |  1994-04-29  |  3.5 KB  |  366x232  |  1-bit (2 colors)
   ocr: LogLikelihood for Total Error is the loglikelihood for the fitted model, the as the Error line in the Analysis of Loglikelihood report. T - for Pure Error is the loglikelihood that would result from fi separate response probabilities for each sample as distingu from separate X values. for Lack of Fit is the difference petween Total Error and PI loglikelihoods. ChiSquare is twice the loglikelihood for Lack of Fit, forming, a likelihoc chi-square test for the hypothesis that the model fits as we saturated model that fitted response probabilites for each S combination. Sometimes this IS caleddmee ...