On Subjective Probability Forecasting

Abstract

{T}he subjective process of probability forecasting is analyzed.{I}t is found to contain a sorling aspect, in which the forecasterdistributes all instances into an ordered set of categories of likelihoodof occurrence, and a laboling aspect, in which the forecaster assignsan anticipated relative frequency, or probability, of occurrencefor each category. {T}hese two aspects are identified with the conceptsof sharpness and validity, which have been introduced by other writers.{T}he verification score proposed by {B}rier is shown to consistof the sum of measures of these two qualities. {A} satisfactory measureof synoptic skill is obtained by applying the {B}rier score to thesynoptic probability forecast and to a control forecast of the climatologicalprobability, and by expressing the difference as a percentage ofthe control score.{I}n an analysis of a large number of short-rangeprobability forecasts made by instructors and students in the synopticlaboratory of the {M}assachusetts {I}nstitute of {T}echnology itis found that even inexperienced forecasters are capable of displayingvalidity and skill except when dealing with events which occur veryrarely or nearly always. {S}kill for average or net conditions over24-hr periods is found to he roughly twice the skill in forecastsfor a particular instant and is found to vary with the directnesswith which the weather element can be inferred from prognostic charts.{T}he average of the judgment of two or more forecasters with comparableexperience is found to be a more skillful statement than the forecastof the most skilled individual.