Measures of Subjective Probability

23
Measuring Subjective Probability in the Context of Medication Risks Niels Haase, Frank Renkewitz & Cornelia Betsch University of Erfurt (Germany)

description

Measuring Subjective P robability in the Context of Medication R isks Niels Haase , Frank Renkewitz & Cornelia Betsch University of Erfurt (Germany). Measures of Subjective Probability. 7-Point Rating Scale. 11-Point Rating Scale. Visual Analog Scale. - PowerPoint PPT Presentation

Transcript of Measures of Subjective Probability

Page 1: Measures of Subjective Probability

Measuring Subjective Probability in the Context of Medication RisksNiels Haase, Frank Renkewitz & Cornelia BetschUniversity of Erfurt (Germany)

Page 2: Measures of Subjective Probability

Haase, Renkewitz & Betsch: Measuring Subjective Probability 221st Annual Conference18-20 June 2012 ETH Zurich

Measures of Subjective Probability

7-Point Rating Scale

11-Point Rating Scale

Visual Analog Scale

Page 3: Measures of Subjective Probability

Haase, Renkewitz & Betsch: Measuring Subjective Probability 321st Annual Conference18-20 June 2012 ETH Zurich

Measures of Subjective Probability

Frequency Format

Percent Format

Page 4: Measures of Subjective Probability

Haase, Renkewitz & Betsch: Measuring Subjective Probability 421st Annual Conference18-20 June 2012 ETH Zurich

Measures of Subjective Probability

• Differences between scales– Resolution– Verbal vs. numeric

• Evaluative criteria– Usability– Test-retest reliability– Subjective confidence in judgment– Behavior prediction

Page 5: Measures of Subjective Probability

Haase, Renkewitz & Betsch: Measuring Subjective Probability 521st Annual Conference18-20 June 2012 ETH Zurich

Research Question & Design

• How do these scales perform when objective probabilities have been learned?

• SensitivityIndividual correlation between objective and subjective probabilities

• Context DependencyDifferentiation between ranges of probability

Objective Probabilities

Subjective Probability Judgment

Objective Probabilities

etc.(4 Levels of Probability)

Page 6: Measures of Subjective Probability

Haase, Renkewitz & Betsch: Measuring Subjective Probability 621st Annual Conference18-20 June 2012 ETH Zurich

Hypotheses

• Rating scales’ disadvantage due to low resolution and vague category quantifiers

• Visual analog scale’s disadvantage due to difficulties in its use

• H1: Numeric scales show higher sensitivity than the visual analog scale and the rating scales.

• H2: Rating scales and the visual analog scale will be more context-dependent.

Page 7: Measures of Subjective Probability

Haase, Renkewitz & Betsch: Measuring Subjective Probability 721st Annual Conference18-20 June 2012 ETH Zurich

Independent Variables

• 5 Scale Formats (between)

• 2 Ranges of probability (between):2%, 5%, 10%, 20%

vs. 42%, 45%, 50%, 60%

• Mode of presentation (within):SequentialGraphical (control)

Page 8: Measures of Subjective Probability

Haase, Renkewitz & Betsch: Measuring Subjective Probability 821st Annual Conference18-20 June 2012 ETH Zurich

Sequential Presentation

Page 9: Measures of Subjective Probability

Haase, Renkewitz & Betsch: Measuring Subjective Probability 921st Annual Conference18-20 June 2012 ETH Zurich

Sequential Encoding: Sensitivity

Page 10: Measures of Subjective Probability

Haase, Renkewitz & Betsch: Measuring Subjective Probability 1021st Annual Conference18-20 June 2012 ETH Zurich

Sequential Encoding: Sensitivity

Low Range of Probabilities: 2%-20% High Range of Probabilities: 42%-60%

Page 11: Measures of Subjective Probability

Haase, Renkewitz & Betsch: Measuring Subjective Probability 1121st Annual Conference18-20 June 2012 ETH Zurich

Sequential Encoding: Context Dependency7-point rating scale Percent format

42%20% 42%20%

Page 12: Measures of Subjective Probability

Haase, Renkewitz & Betsch: Measuring Subjective Probability 1221st Annual Conference18-20 June 2012 ETH Zurich

• All scales were highly sensitive

• Numeric scales were generally superior

• Rating and visual analog scales were context dependent

• Differences between scales are reduced when encoding is more error-prone

Summary: Sequential Encoding

Page 13: Measures of Subjective Probability

Haase, Renkewitz & Betsch: Measuring Subjective Probability 1321st Annual Conference18-20 June 2012 ETH Zurich

Probability of 42%:

Graphical Presentation

Page 14: Measures of Subjective Probability

Haase, Renkewitz & Betsch: Measuring Subjective Probability 1421st Annual Conference18-20 June 2012 ETH Zurich

Sequential Encoding vs. Graphical Encoding

Page 15: Measures of Subjective Probability

Haase, Renkewitz & Betsch: Measuring Subjective Probability 1521st Annual Conference18-20 June 2012 ETH Zurich

Graphical Encoding: Sensitivity

Page 16: Measures of Subjective Probability

Haase, Renkewitz & Betsch: Measuring Subjective Probability 1621st Annual Conference18-20 June 2012 ETH Zurich

Graphical Encoding: Sensitivity

Low Range of Probabilities: 2%-20% High Range of Probabilities: 42%-60%

Page 17: Measures of Subjective Probability

Haase, Renkewitz & Betsch: Measuring Subjective Probability 1721st Annual Conference18-20 June 2012 ETH Zurich

Graphical Encoding: Context Dependency7-point rating scale Percent format

42%20% 42%20%

Page 18: Measures of Subjective Probability

Haase, Renkewitz & Betsch: Measuring Subjective Probability 1821st Annual Conference18-20 June 2012 ETH Zurich

• Altogether higher sensitivity

• Differences between measures remain the same, numeric formats superior

• Reduced context dependency

• No effect of range, i.e. encoding difficulty

Summary: Graphical Encoding

Page 19: Measures of Subjective Probability

Haase, Renkewitz & Betsch: Measuring Subjective Probability 1921st Annual Conference18-20 June 2012 ETH Zurich

Experiment 2: Effect of encoding error

• Encoding more error-prone– 5 outcomes within 1 sequence– Shorter inter-stimulus time– 3 repeated measurements

• 2 Scale formats (between)7-point rating scale vs. Percent format

• 2 Ranges of probability (within):10%, 20%, 30%, 40%, 50% vs. 50%, 60%, 70%, 80%, 90%

Page 20: Measures of Subjective Probability

Haase, Renkewitz & Betsch: Measuring Subjective Probability 2021st Annual Conference18-20 June 2012 ETH Zurich

Experiment 2: Sensitivity

Small Probabilities: 10%-50%

Large Probabilities: 50%-90%

Page 21: Measures of Subjective Probability

Haase, Renkewitz & Betsch: Measuring Subjective Probability 2121st Annual Conference18-20 June 2012 ETH Zurich

Experiment 2: Context Dependency7-point rating scale Percent format

10% 20% 30% 40% 50% 50% 60% 70% 80% 90%Objective Probabilities

10% 20% 30% 40% 50% 50% 60% 70% 80% 90%Objective Probabilities

Small Probabilities: 10%-50%

Large Probabilities: 50%-90%

50% 50% 50% 50%

Page 22: Measures of Subjective Probability

Haase, Renkewitz & Betsch: Measuring Subjective Probability 2221st Annual Conference18-20 June 2012 ETH Zurich

• Different scale formats vary in sensitivity with numeric measures generally faring the best

• Sensitivity also varies as a function of error in infomation at the time of judgment

• Scale characteristics (e.g. resolution) become less important with increasing error

• Context dependency increases with increasing error

• In the realm of probabilities rating scales are not suitable for between-subjects research

Conclusions & Implications

Page 23: Measures of Subjective Probability

Haase, Renkewitz & Betsch: Measuring Subjective Probability 2321st Annual Conference18-20 June 2012 ETH Zurich

Thank you.