Home > RippleStat > Decsion Theory > Video of Hypothesis Testing Model of Inferential Statistics


Instructional Software for
Statistics & Experimental Psychology

RippleSoft Software

Video of Hypothesis Testing Model of Decision Theory Application

Figure 1. The user chooses a new random condition--the user will not know whether the new treatment produces results different from the established treatment.

Then the user runs the experiment and "rejects Ho"--there is evidence the new treatment is different.

Now the practitioner uses the new treatment and discovers she doesn't get a different result from when she was using the established method. Perhaps the practitioner decides, "I can't trust what those scientists tell me."

What happened? Well, since this is a simulation we can gather further evidence. If the researcher conducts the experiment many times (224 times) she would discover that about half the time the Ho is rejected--that is indicating the power of the statistical test is ony about 50%: "Power is the probability of rejecting a false Ho."

If many practitioners try the new method, about half of them (55%) discover the new technique is different. Half don't find any difference.

The new technique is "different." It just isn't a whole lot different. It is statistically significant ("reliable") and perhaps not usefully so. Other factors are important, such as, the new treatment might be cheaper, or less dangerous, etc.