This experiment illustrates how all the choices are bundled up in the experiment, handled by the way in which the Alternative hypothesis is expressed and in the corresponding test statistic, rather than as multiple disconnected experiments in which the Bonferroni penalty, the upper p-level bound, has to be paid at the end in a much more costly way. The experiment was conducted in accordance with one of our standard experimental protocols to test the Null hypothesis against the Alternative hypothesis that in the Chumash nearly all the Prime Ministers have one of their appellations with ELSs in a more compact meeting with one of the ELSs of a key word for Prime Minister than expected by chance.

Experimental Protocol

Text	The Five Books of Moses; the Chumash
Skip Specification	Expected Number of ELS = 10
Resonance Specification	Maximum ELS row skip on a cylinder = 10 Maximum ELS column skip on a cylinder = 10
Monkey Text Control Population	ELS random placement
Compactness Measure	Maximum distance between the letters of an ELS pair
Number of Trials	10,000
Significance Level of test	1%

The p-level of the experiment was 6/10,000 so we reject the Null hypothesis.

Of course we want to see the tables associated with each prime minister. But we must remember that the protocol of the experiment did not look for a best table and the test statistic was based on what it found in all tables. So looking at one best area table is not necessarily representative.

< Next >