Obsidian | Level 7

Additive Two-Way ANOVA Model - nonsignificant model but significant parameter estimates?

My knowledge on two-way ANOVAs is a bit shaky so I'm hoping for some clarification. So it appears my overall model is significant (p < 0.0001). However, I am a bit confused why pressure is nonsignificant and temperature is significant in the Type III ANOVA table. Yet individual levels of pressure are significant and temperature 270 has no values. Are these the difference between main effects and simple effects?

Furthermore, I was going to look at pairwise comparisons, but it that wrong when there is no interaction? That is, I can't really control for temperature when comparing different pressure levels and vise versa.

Data

``````DATA DATA.Temp;
input Pressure Temp Strength @@;
datalines;
120 250 9.60 120 260 11.28 120 270 9.00
130 250 9.69 130 260 10.10 130 270 9.57
140 250 8.43 140 260 11.01 140 270 9.03
150 250 9.98 150 260 10.44 150 270 9.80
;
RUN;``````

Model Procedure

``````PROC GLM data = DATA.Temp plots = ALL;
class pressure temp;
model strength = pressure temp / clparm noint solution;
means pressure temp / hovtest = Levene;
lsmeans pressure temp / pdiff = all adjust = Tukey;
RUN;``````

Output

 Source DF Sum of Squares Mean Square F Value Pr > F Model 6 1164.195417 194.032569 540.51 <.0001 Error 6 2.153883 0.358981 Uncorrected Total 12 1166.349300
 Source DF Type III SS Mean Square F Value Pr > F Pressure 3 0.58069167 0.19356389 0.54 0.6727 Temp 2 4.65765000 2.32882500 6.49 0.0316
 Parameter Estimate StandardError t Value Pr > |t| 95% Confidence Limits Pressure  120 9.482500000 B 0.42366293 22.38 <.0001 8.445834157 10.519165843 Pressure  130 9.309166667 B 0.42366293 21.97 <.0001 8.272500824 10.345832510 Pressure  140 9.012500000 B 0.42366293 21.27 <.0001 7.975834157 10.049165843 Pressure  150 9.595833333 B 0.42366293 22.65 <.0001 8.559167490 10.632499176 Temp      250 0.075000000 B 0.42366293 0.18 0.8653 -0.961665843 1.111665843 Temp      260 1.357500000 B 0.42366293 3.20 0.0185 0.320834157 2.394165843 Temp      270 0.000000000 B . . . . .

1 ACCEPTED SOLUTION

Accepted Solutions
Diamond | Level 26

Re: Additive Two-Way ANOVA Model - nonsignificant model but significant parameter estimates?

The Type III test is testing to see if the levels of each variable are different from one another. So the levels of TEMP are different from one another, the levels of PRESSURE are not.

The parameter estimate tests are testing to see if the estimate is different from zero. So for PRESSURE, the estimates are different from zero.

--
Paige Miller
Diamond | Level 26

Re: Additive Two-Way ANOVA Model - nonsignificant model but significant parameter estimates?

The Type III test is testing to see if the levels of each variable are different from one another. So the levels of TEMP are different from one another, the levels of PRESSURE are not.

The parameter estimate tests are testing to see if the estimate is different from zero. So for PRESSURE, the estimates are different from zero.

--
Paige Miller
Discussion stats