Multcompare function

Figure 9.21 is the output. The multcompare function provides an interactive plot that we can use to test for differences across groups. For example, if you click on L3, you can see that it is different from 12 sensors. Therefore, L3 must be the sensor that is wrong (i.e., odd in the sense that it is so different from all other sensors). On the other hand, R3 is only different from five other sensors. We do not know the cause of L3 s oddness. It could be bad data acquisition hardware, improper installation or placement of the L3 sensor, or some other reason. The point is that L3 s values are suspicious and will be removed from further analysis. The command ... 

Use multcompare(stats) where stats structure is the output of MATLAB anoval function, used in Step 3. multcompare generates an interactive plot where you can tell how many groups are compared with the group of interest. 

---