Luis A. Afonso
Posts:
4,743
From:
LIsbon (Portugal)
Registered:
2/16/05


Current misleading claims to be discussed . . .
Posted:
May 26, 2014 5:50 PM


Current misleading claims to be discussed . . .
Towards H0 to Reject, as unlike, or not to do so, is all we are allowed to accede.
Slopping writing, saying and denotation, not excluding of course deficient University teaching, does lead to a campaign, leading mainly by Psychologists, against the Null Hypothesis Significance Test, NHST, (Frequentist way), that amazingly rages even today. ______When someone read, concerning the Null Hypothesis, H0: p=0, is automatically induced to think that the concern is focused in to find what conditions could be fulfilled in order to be sure that the parameter p is equal to zero. However, really, the aim is quite different: that is to obtain from the data sufficient evidence (conditioned by the significance level) the Null Hypothesis be so unlike that we should reject it. Note two important features: ____Is p=0 likeliness, not its value we are concerned to find out, ____This likeliness is gauging by a value we arbitrary set in advance: is alpha the probability not to reject a Null Hypothesis in case we should do. Note that I had the care to avoid: when the Null is true, so common, however easily misdirecting, found in textbooks. To those that are persuaded that the null can be stated true I used to ask a simple question: How many flips have you to perform in order to prove a coin is fair (probability heads on equal to ½)?
Luis A. Afonso

