Range Effects in Experiments on People

1
Range Effects in Experiments on People

• E.C. Poulton (1975)

2
Outline

• Types of range effects
• (Asymmetrical) Transfer of training
• Time-order effects
• Stimulus and response probability
• How to avoid range effects

3
Types of Range Effects

• Transfer of training
• Asymmetrical transfer
• Time-order errors
• Central tendency of judgment
• Stimulus generalization
• Stimulus and response probability
• Probability learning

4
Transfer of TrainingAsymmetrical Transfer

• Compensatory vs. pursuit tracking tasks
• A subjects expert task performance depends on
  which task they are trained on first.
• Training on a pursuit tracking task before a
  compensatory tracking task leads to lower error
  rates for the compensatory tracking task.

5
(No Transcript)
6
Time Order ErrorsCentral-Tendency Range Effects

• Ratings of Sensory Magnitude
• Objective, or direct measures of sensory
  magnitude are impossible to obtain.
• Subjects vary their responses based on the range
  of sensory stimuli used.
• Examples Acceptable dB level of noise, aesthetic
  preference for isosceles triangles, no-lose bets,
  shape constancy, size constancy, step-tracking

7
Time Order Errors Stimulus Generalization

• When comparing colors of variable wavelength to a
  memorized standard, the addition of stimuli on
  each end of the spectrum increases the number of
  confusions in the middle.

8
Time Order Errors Stimulus Generalization

• Adjacent categories
• When making absolute judgments of tones, the
  middle categories also get confused with one
  another.
• The larger the range of tones to be remembered,
  the more confusion there will be in the middle
  range.

9
Time Order Errors Stimulus Generalization

• Serial positions
• When memorizing a list, the first and last few
  positions are easily remembered, but the items in
  the middle of the list get confused with one
  another.

10
Stimulus Probability

• RT depends on the probability of the stimulus and
  the probability of the response.
• Subjects learn probabilities and prepare
  themselves accordingly
• Leads to quicker reaction time in two-dimension
  conditions

11
Stimulus ProbabilityProbability Learning

• Maximizing Subjects are rewarded for correctly
  guessing the stimuli source, so they always pick
  the most frequent source
• Probability matching Subjects arent rewarded,
  but they use the information gathered and predict
  each source about in proportion to the frequency
  of signals coming from it
• Averaging Subjects predict each source equally
  often

12
Reaction Time and Signal Intensity

• Reaction times for responding to tones were
  longer after hearing a very loud tone.
• Likewise, reaction times were shorter after
  hearing a soft tone.
• Asymmetrical range effect because it is not
  possible to respond much faster than .2 seconds,
  but it can take much longer.

13
Summary

• Any experimental design in which each person
  performs more than one condition (repeated
  measures) is going to incorporate some sort of
  range effect.
• Within subjects designs should only be used in
  combination with a between subjects design, or
  for the purpose of studying range effects.