Probing For Hypotheses
Marvin Levine · 2022
From current perspectives, the most important detail of Reading No. 12 was the introduction of the blank-trial series as a probe for the H. This technique was demonstrated in four-dimensional problems with a set of four blank trials. The most fundamental validation derives from a feature insufficiently stressed within the articles. During four blank trials S may produce any one of 16 response patterns. Eight of these are patterns corresponding to the simple Hs; the remaining eight are patterns inconsistent with these Hs and with the theory as initially stated. This feature provides a built-in null hypothesis: If during blank trials Ss were responding randomly, then the proportion of blank-trial sets showing simple H patterns should approximate 0.5. The obtained result was characteristically above 0.9. Other results also attested to the validity of the probe.