TABLE II

Notation and Numerical Calculations for With Trend Over Time

NotationFormulaCalculated Value
Within-lot estimate of slopeFormulaβ̂ = −0.457
Variance of predicted lot meansFormulaSL2 = 4.197
Mean squared errorFormulaSE2 = 0.529
Predicted value of Y at t0Ŷ = Ȳ* + β̂(t0*)66.333 at t0 = 12
Estimated variance of Ŷ at t0Formula0.648 at t0 = 12
Estimated variance of YFormula4.363
Harmonic mean of the number of values in each lotaFormula1.454
Lot degrees of freedoms = I − 110
Error degrees of freedomFormula14
Effective sample sizeFormula6.729 at t0 = 12
Upper bound on variance of YFormula10.820
Constants used to compute U where χα:s2 and χα:r2 are chi-squared percentiles with area α to the left and s and r degrees of freedom, respectivelyC1 = sα:s2 − 11.538
C2 = rα:r2 − 11.131
Lower bound on one-sided tolerance intervalEquation 6L1 = 57.36
Upper bound on one-sided tolerance intervalEquation 7U1 = 75.31
Lower and upper on two-sided tolerance intervalEquation 8L2 = 57.25, U2 = 75.41
Lower and upper one-sided bounds using HK1 adjustment of two-sided tolerance intervalEquation 8 replacing Formula with ZPL1,HK1 = 58.13, U1,HK1 = 74.53
  • a This is an adjustment from the value suggested by Park and Burdick (18) but is computationally easier and provides very similar results.