Hypothetico-Deductive method

Popper's theory of corroboration
E is a severe test of T with respect to background theory B:
S(E,T,B) = Defn prob(E,T) is much greater than prob (E,B)
NB: prob(E,T) is probability of evidence E given theory T; prob (T,E) is probability of theory T given evidence E

If it's true for the severe test, then the prediction P is said to corroborate theory T.

Degree of corrooratuib of T by E with respect to background theory B:
Corr(E,T,B) = Defn [prob(E,T)-prob(E,B)]/[prob(E,T)+prob(E,B)]
(worst case -1 and best case +1)

Given a number of theories of which T has the highest degree of corroboration, should one use T for the purpose of prediction, or of action, rather than one of the other less well corroborated theories? The dispute here concerns the problem of induction. Popper rejected induction, but cannot avoid it entirely.