[o] GLIM 4, update 8 for Sun SPARCstation / Solaris on 4 Nov 2008 at 14:35:15 [o] (copyright) 1992 Royal Statistical Society, London [o] [i] ? unit 18 [f] ** directive expected but not found, at [? u] [f] [h] An invalid character was detected when the previous directive was completed and [h] the next one sought. Invalid character is u N.B. a directive begins with a [h] directive symbol $ [h] [i] ? $units 18$ [i] ? $data thrust$ [i] ? $read [i] $REA? 2340 2355 2362 2350 2348 [i] $REA? 2658 2650 2665 2640 [i] $REA? 2449 2458 2432 2437 2447 [i] $REA? 2410 2418 2397 2405 [i] ? $data block$ [i] ? $read 1 2 3 4 5 1 2 3 4 1 2 3 4 5 2 3 4 5 [i] ? $data treat$ [i] ? $read 1 1 1 1 1 2 2 2 2 3 3 3 3 3 4 4 4 4 [i] ? $look thrust block treat$ [o] THRUST BLOCK TREAT [o] 1 2340. 1.000 1.000 [o] 2 2355. 2.000 1.000 [o] 3 2362. 3.000 1.000 [o] 4 2350. 4.000 1.000 [o] 5 2348. 5.000 1.000 [o] 6 2658. 1.000 2.000 [o] 7 2650. 2.000 2.000 [o] 8 2665. 3.000 2.000 [o] 9 2640. 4.000 2.000 [o] 10 2449. 1.000 3.000 [o] 11 2458. 2.000 3.000 [o] 12 2432. 3.000 3.000 [o] 13 2437. 4.000 3.000 [o] 14 2447. 5.000 3.000 [o] 15 2410. 2.000 4.000 [o] 16 2418. 3.000 4.000 [o] 17 2397. 4.000 4.000 [o] 18 2405. 5.000 4.000 [i] ? $factor block 5 treat 4$ [i] ? $yvar thrust$ [i] ? $fit$ [o] deviance = 222074. [o] residual df = 17 [o] [i] ? $fit + block$ [o] deviance = 209292. (change = -12781.) [o] residual df = 13 (change = -4 ) [o] [i] ? $fit+treat$ [o] deviance = 826.88 (change = -208465.) [o] residual df = 10 (change = -3 ) [o] Anova Table Source SS df MS F Block 12781 4 3195.25 38.64225 Treat 208465 3 69488.33 840.3678 Error 826.88 10 82.688 the critical value : F0.05(4,10)=3.48, since 38.64>3.48, hence the block is sig. the critical value : F0.05(3,10)=3.71, since 840.37>3.71, hence the treat is sig. [i] ? $display e s r m$ [o] estimate s.e. parameter [o] 1 2350. 6.261 1 [o] 2 4.792 7.072 BLOCK(2) [o] 3 5.792 7.072 BLOCK(3) [o] 4 -7.458 7.072 BLOCK(4) [o] 5 -0.4545 7.755 BLOCK(5) [o] 6 302.0 6.215 TREAT(2) [o] 7 93.60 5.751 TREAT(3) [o] 8 56.37 6.215 TREAT(4) [o] scale parameter 82.69 95% CI for treatment t2-t1=302.0+/-t0.025(10)*6.215=302.0+/-2.228*6.215=(288.153, 315.847) t3-t1=93.6+/-t0.025(10)*5.751=93.6+/-2.228*5.751=(80.787, 106.4132) t4-t1=56.37+/-t0.025(10)*6.215=56.37+/-2.228*6.215=(42.523, 70.217) t3-t2=(93.60-302.0)+/-2.228*6.215=(-222.247, -194.553) t4-t2=(56.37-302.0)+/-2.228*6.716=(-260.563, -230.667) t4-t3=(56.37-93.60)+/-2.228*6.215=(-51.077,-23.383) zero is not in either of the intervals above, so all the treatments are sig. different from each other. [o] [o] standard errors of parameter estimate differences [o] 1 0.000 [o] 2 12.03 0.000 [o] 3 12.03 6.430 0.000 [o] 4 12.03 6.430 6.430 0.000 [o] 5 12.63 7.072 7.072 7.072 0.000 [o] 6 10.71 9.353 9.353 9.353 9.150 0.000 [o] 7 10.26 9.116 9.116 9.116 9.655 6.215 [o] 8 9.979 10.13 10.13 10.13 10.67 6.716 [o] 1 2 3 4 5 6 [o] [o] 7 0.000 [o] 8 6.215 0.000 [o] 7 8 [o] [o] unit observed fitted residual [o] 1 2340 2350. -10.466 [o] 2 2355 2355. -0.258 [o] 3 2362 2356. 5.742 [o] 4 2350 2343. 6.992 [o] 5 2348 2350. -2.011 [o] 6 2658 2652. 5.531 [o] 7 2650 2657. -7.260 [o] 8 2665 2658. 6.740 [o] 9 2640 2645. -5.010 [o] 10 2449 2444. 4.934 [o] 11 2458 2449. 9.142 [o] 12 2432 2450. -17.858 [o] 13 2437 2437. 0.392 [o] 14 2447 2444. 3.389 [o] 15 2410 2412. -1.624 [o] 16 2418 2413. 5.376 [o] 17 2397 2399. -2.374 [o] 18 2405 2406. -1.378 [o] [o] Current model: [o] [o] number of observations in model is 18 [o] [o] y-variate THRUST [o] weight * [o] offset * [o] [o] [o] probability distribution is NORMAL [o] link function is IDENTITY [o] scale parameter is to be estimated by the mean deviance [o] [o] linear model: [o] terms: 1+BLOCK+TREAT i] ? $cal res=y-%fv$ [i] ? $plot res treat$ [o] 10. + [o] | R [o] |R R [o] |R R R R [o] | R [o] | [o] 0. +R R [o] |R 3 [o] | [o] | R [o] | R [o] | [o] -10. +R [o] | [o] | [o] | [o] | [o] | R [o] -20. + [o] +---------+---------+---------+---------+---------+---------+ [o] 1.0 1.5 2.0 2.5 3.0 3.5 4.0 2*sqrt(MSerror)=2*sqrt(82.688)=18.1866 So no outliers [i] ? $sort res$ [i] ? $ca pos=%cu(1): pos=(pos-0.5)/%nu: pos=%nd(pos)$ [i] ? $plot res pos$ [o] 10. + [o] | R [o] | R R [o] | R R R R [o] | R [o] | [o] 0. + R R [o] | R R R R [o] | [o] | R [o] | R [o] | [o] -10. + R [o] | [o] | [o] | [o] | [o] | R [o] -20. + [o] +---------------+---------------+---------------+--------------- [o] -2. -1. 0. 1. do not seem linear, hence the normality assumption may not hold. [i] ? $fit$ [o] deviance = 222074. [o] residual df = 17 [o] [i] ? $fit+treat$ [o] deviance = 1268.9 (change = -220805.) [o] residual df = 14 (change = -3 ) [o] [i] ? $fit+block$ [o] deviance = 826.88 (change = -442.1) [o] residual df = 10 (change = -4 ) [o] SStreat, SSblock are not same as before.