.- help for ^logprof^ (STB-56: sg148) .- Confidence Intervals based on the profile likelihood after logistic regression ------------------------------------------------------------------------------ ^logprof^ varname [^,^ ^lev^el^(^#^)^ ^nog^raph ^or^ { ^inc(^#^)^ | ^incfrac(^#^)^ } ^f^rom^(^#^)^ ^t^o^(^#^)^ ^sav^ing^(^filename^)^ ] ^logprof^ is only available after @logistic@ or @logit@. Description ----------- ^logprof^ calculates confidence intervals based on the profile likelihood for logistic regression models. The profile log-likelihood is calculated over a range of beta coefficients and confidence intervals obtained. Options ------- ^inc(^#^)^ specifies the interval between each value of the beta coefficient. If the confidence interval is unbounded, the increment may be too high. The default value of ^inc^ is 0.01 (unless ^incfrac^ is used instead). ^incfrac(^#^)^ is an alternative to inc, where the increment is based on the value of the estimated beta coefficient. For example, ^incfrac(100)^ would correspond to an increment of one-hundreth times the estimated beta coefficient. ^level(^#^)^ specifies the significance level, in percent, for confidence intervals; see help @level@. The default is 95%. ^or^ requests confidence intervals to be displayed in terms of an odds ratio. ^nograph^ suppresses the graphing of the profile likelihood. ^from(^#^)^ specifies the value of the beta coefficient to start iterating from. ^to(^#^)^ specifies the value of the beta coefficient to stop iterating at. ^saving(^filename^)^ requests the graph be saved. Both ^from^ and ^to^ have default values based on the logistic regression model previously fitted. Examples -------- . ^logistic dead year age, offset(offset)^ . ^logprof age, or^ . ^logistic foreign weight mpg^ . ^logprof weight, inc(0.001) f(-2) t(3)^ Author ------ Mark S. Pearce Department of Child Health University of Newcastle upon Tyne. m.s.pearce@@ncl.ac.uk Also see -------- On-line: help for @logistic@.