MLGI {GRRGI}R Documentation

Restricted Maximum Likelihood Estimates of Variance and Generalized Inference Confidence Limits for Gauge R and R Data

Description

Returns a matrix of the Restricted Maximum Likelihood Estimates of Variance for gauge R and R studies. Also provides the confidence intervals using Gerneralized Inference as developed by Hamada and Weerandi.

Usage

MLGI(data, prob, N)

Arguments

data Gauge R and R data, a matrix with columns named 'resp', 'part' and 'operator' in any order
prob desired confidence level such as 0.95 - a constant between 0 and 1
N desired number of simulations for Gerneralized Inference

Details

Returns a matrix with 17 rows and four columns with the estimates: SD:Part, SD:Operator, SD:PartOp, SD:Repeat, SD:Reproduce, SD:Gauge, SD:Total (standard deviations.) Then Var:Part, Var:Operator, Var:PartOp, Var:Repeat, Var:Reproduce, Var:Gauge, Var:Total (variances.) Then Gauge/Total, Gauge/Parts, Repeat/Gauge (ratios.)

The columns are 'REML est' (restricted maximum likelihood estimate,) 'Lower GI', 'Mid GI', 'Upper GI'. The Lower and Upper GI are the Generalized Inference confidence limits. 'MLGI' will eliminate the 'part operator' interaction factor if the P value of the likelihood ratio test is greater than 0.25.

Value

A matrix with 17 rows and four columns with the estimates.

Note

The 'data' argument needs to be a matrix with columns named 'resp', 'part' and 'operator'.

Author(s)

Walter Resch walt55128@msn.com

References

Package is based on Michael Hamada and Sam Weerandi (July 2000) Measurement System Assessment Via Generalized Inference, Journal of Quality Technology 32, 241-253

See Also

anovasat, RRgraph, ratiograph

Examples

data(HWstudy1)
MLGI(HWstudy1,0.95,1000)

[Package GRRGI version 1.1 Index]