Likelihood Ratio Test For Discrete Classes

LOD Score

Example

Varying Q

Maximum Likelihood: Testcross Data

Plotting Lod Scores

Homework Assignment # 9 Questions

  Click here for a
printer-friendly version

Varying Q

class AaBb Aabb aaBb aabb Total
observed 310 287 288 315 1200

z = Log10 [    L()    ]
L(0.5)

L(Q) = (Q)575 (1-Q)625

Log10[L(Q)] = 575Log10(Q) + 625Log10(1-Q)

Log10[L(0.5)] = 575Log10(0.5) + 625Log10(0.5)
  = 1200Log10(0.5)
  = 1200(-0.30103)
  = -361.24

Now we will vary Q and plot the lod score versus Q.

Say Q = 0.1, then Log10[L(Q)]

  = 575Log10(0.1) + 625Log10(0.9)
  = 575(-1) + 625(-0.0458)
  = -546.4

The Lod score when Q = 0.1 = -546.4 + 361.24 = -185.16
The Lod score when Q = 0.2

  = 575Log10(0.2) + 625Log10(0.8) - 1200Log10(0.5)
  = -101.24

The Lod score when Q = 0.3
  = 575Log10(0.3) + 625Log10(0.7) - 1200Log10(0.5)
  = -36.229

The Lod score when Q = 0.4
  = 575Log10(0.4) + 625Log10(0.6) - 1200Log=10(0.5)
  -6.231

The Lod score when Q = 0.45
  = 575Log10(0.45) + 625Log10(0.55)
  - 1200Log10(0.5)

  = -199.4028 - 162.27332 + 361.24
  = -0.43612

The Lod score when Q = 0.475
  = 575Log10(0.475) + 625Log10(0.525)
  - 1200Log10(0.5)

  = -185.901 - 174.90 + 361.24
  = 0.439

The Lod score when Q = 0.49
  = 575Log10(0.49) + 625Log10(0.525)
  - 1200Log10(0.5)

  = -178.1373 - 182.7686 + 361.24
  = 0.334

Now we can plot the Lod score as Q changes.

Copyright 2000©, Ted Helms

Back | Home | Top | Next
Home Forward Back