Table 2. Type I error probabilities for the linear regression and the GLM-based test statistics by module size and sample size.

module size	sample size (n)	linear regression ⋆	GLM-based LR test ⋆	GLM-based Larntz & Perlman ⋆
4	125	0.205 [0.179, 0.231]	0.109 [0.089, 0.129]	0.045 [0.032, 0.058]
4	450	0.056 [0.041, 0.071]	0.062 [0.047, 0.077]	0.043 [0.030, 0.056]
4	800	0.020 [0.011, 0.029]	0.053 [0.039, 0.067]	0.035 [0.023, 0.047]
5	125	0.197 [0.172, 0.222]	0.141 [0.119, 0.163]	0.048 [0.034, 0.062]
5	450	0.064 [0.048, 0.080]	0.067 [0.051, 0.083]	0.037 [0.025, 0.049]
5	800	0.022 [0.012, 0.032]	0.066 [0.050, 0.082]	0.048 [0.034, 0.062]
7	125	0.461 [0.430, 0.492]	0.314 [0.285, 0.343]	0.035 [0.023, 0.047]
7	450	0.314 [0.285, 0.343]	0.083* [0.065, 0.100]	0.036* [0.024, 0.048]
7	800	0.212 [0.186, 0.238]	0.070** [0.054, 0.087]	0.029** [0.018, 0.040]

^⋆ estimate [95% confidence interval]

* convergence rate of GLM: 0.991

** convergence rate of GLM: 0.993