BHS 20401 Methods in Behavioral Sciences I

1
BHS 204-01Methods in Behavioral Sciences I

• April 9, 2003
• Chapter 2 (Stanovich) Falsifiability How to
  Foil Little Green Men in the Head

2
Logic of Experimentation

• Two forms of logic
• Deduction moving from general principles to
  specific conclusions.
• Induction moving from specific observations to
  general principles.
• Induction is used during naturalistic and
  exploratory research.
• Deduction is used during experiments.

3
Propositional Logic

• Modus Ponens (confirmatory)
• If p then q
• Observe p
• Conclude q
• Modus Tollens (disconfirmatory)
• If p then q
• Observe not-q
• Conclude not-p

4
Logical Fallacies

• Affirming the consequent
• If p then q
• Observe q
• Conclude p
• Denying the antecedent
• If p then q
• Observe not-p
• Conclude not-q

5
Falsifiability

• Seeking support for hypotheses commits the
  logical fallacy of affirming the consequent.
• Instead, we must test hypotheses by seeking
  disconfirmatory evidence.
• A testable theory is one that can be proven wrong
  if it is wrong.
• It must have the chance to fail.
• A theory cannot explain every outcome.

6
Poppers Approach

• Predictions from theory (hypotheses) must be
  specific.
• They must state what will happen and what will
  not happen.
• General predictions or all-encompassing
  predictions cannot be tested (are unfalsifiable).
• When data accumulates that contradicts theory,
  then theory must be changed.
• Data is not thrown out explanations are.

7
Two Hypotheses

• Null hypothesis (H0)
• There will be no difference between treatment and
  control groups (no treatment effect).
• Alternative hypothesis (H1)
• There will be a difference of a particular kind.
• Directionality states how the treatment group
  will differ from the control group.
• We test the null hypothesis and by rejecting it
  (disconfirming it) can accept the alternative.

8
The Neyman-Pearson Approach

• Two theories can be compared by predicting
  incompatible outcomes
• If theory A is correct, hypothesis A will be
  confirmed and B will be disconfirmed.
• If theory B is correct, hypothesis B will be
  confirmed and A will be disconfirmed.
• The comparison is not with the control group ( is
  there a treatment effect or not) but with the
  predictions made by the two theories.

9
Errors are Important

• We learn something, even when an experiment does
  not work does not produce the expected
  result.
• Knowledge advances when we find that our ideas
  are wrong and can abandon incorrect beliefs.
• We must be willing to let evidence guide belief
  not the other way around.
• Scientists criticize each others ideas in an
  ongoing dialectic that produces change.

10
Working on the Fringes

• Interesting questions are those that
• Exist at the fringes of knowledge.
• Can be tested using existing methods.
• Many questions are important but untestable.
• Some questions are interesting to the public but
  not to scientists because they have already been
  answered
• ESP and other paranormal claims, astrology.