Testing hypotheses (cont.)
test
implications—are
normally conditional (if-then) statements of the form,
“If conditions C occur then
event E will occur”
“conditions
C”:
·
Conditions
of some experiment
·
Observed
conditions
“If H (test
hypothesis) and A (auxiliary hypothesis) are true, then I (test
implication) is
true.”
Example—
If childbed fever is
caused by infectious matter (H) and chlorinated lime
destroys infectious matter (A), then if
persons attending the patients wash
their hands in a chlorinated lime solution (C)
deaths from childbed fever will
be reduced (E).
Forms
of reasoning involving auxiliary hypotheses—
Case 1—the test implication is true:
If H and A are
true, then I (if C then E) is true.
I is true.
(observation or outcome of experiment)
Therefore, H is true.
Case 2—the test implication is false:
If
H and A are true, then I (if C then E) is
true.
I is false.
(observation or outcome of experiment)
Therefore, either H or
A (and possibly both) are false.
According
to Hempel, in order for a statement to qualify as a scientific hypothesis,
it must be
empirically testable in
principle
[I.e., There must be some conceivable
observation or experiment the results of which
would
determine the truth or falsity of the hypothesis’s test implication.]
pseudo-hypothesis—a statement that appears to
be a scientific hypothesis but fails the
condition of empirical testability in principle
crucial
test—a test
(experimental or observational) intended to determine which of two
rival hypotheses is true
and which is false
reasoning—
If H1 is
true, then I1 (if C then E1) is
true.
If H2 is
true, then I2 (if C then E2) is
true.
[E1
and E2 are incompatible.]
C and E1.
[outcome of experiment]
Therefore, H1
is true and H2 is false.
Example—
the “tower experiment” to
decide between the geocentric (earth-centered) and
heliocentric (sun-centered)
theories of the solar system
1.
Because
auxiliary hypotheses are almost always needed to derive test
implications from test hypotheses, it is impossible to disprove either
of two competing hypotheses.
2.
Test
hypotheses cannot be “conclusively proved by any set of available data.” (pp.
27-28)
ad
hoc hypothesis—an
auxiliary hypothesis introduced for the sole purpose of saving a test
hypothesis
being threatened by adverse evidence
example—the phlogiston theory: phlogiston as having negative
weight
1.
There
is no precise criterion for ad hoc hypotheses.
2.
An
ad hoc hypothesis is motivated by the desire to protect someone’s
favored test hypothesis from refutation by the results of experiments.
3.
Generally
speaking, an ad hoc hypothesis leads to no additional test implications.