5. Process Improvement
5.6. Case Studies
5.6.2. Sonoluminescent Light Intensity Case Study

5.6.2.6.

Important Factors: |Effects| Plot

Purpose

The |effects| plot displays the results of a Yates analysis in both a tabular and a graphical format. It is used to distinguish between important and unimportant effects.

Sample |Effects| Plot

Conclusions from the |effects| plot

We can make the following conclusions from the |effects| plot.

1. A ranked list of main effects and interaction terms is:
   X2
   X7
   X1*X3 (confounded with X2*X7 and X4*X6)
   X1
   X3
   X2*X3 (confounded with X4*X5 and X1*X7)
   X1*X2 (confounded with X3*X7 and X5*X6)
   X3*X4 (confounded with X1*X6 and X2*X5)
   X1*X4 (confounded with X3*X6 and X5*X7)
   X6
   X5
   X1*X2*X4 (confounded with other 3-factor interactions)
   X4
   X2*X4 (confounded with X3*X5 and X6*X7)
   X1*X5 (confounded with X2*X6 and X4*X7)
   

2. From the graph, there is a clear dividing line between the first seven effects (all |effect| > 50) and the last eight effects (all |effect| < 20). This suggests we retain the first seven terms as "important" and discard the remaining as "unimportant".

3. Again, the confounding structure on the right reminds us that, for example, the nominal effect size of 70.0125 for X1*X3 (molarity*pH) can come from an X1*X3 interaction, an X2*X7 (solute*clamping) interaction, an X4*X6 (gas*horn depth) interaction, or any mixture of the three interactions.