Methods for the Analysis of Choices and Evaluations

Notes on choice intentions measurement and Options Analysis application

  1. Two varieties of choice intention

  2. Using choice intentions data to estimate share of choice

  3. Using choice intentions data for Options Analysis validity assessment

Two varieties of choice intention

Two types of intention data can be usefully distinguished:

  1. Immediate choice intention: An immediate intention refers to a choice that would be made right now.

  2. Anticipated choice intention: This refers to the most likely next choice. It is implied by the use of "usual", "most common", "expected" or other analagous term for choice intention.

The distinction between these two is suggestive of the need for care in the framing of intentions questions and interviewer guidance in the collection of survey data.

Where the nature of the choice intention is left unstated, respondents tend to assume that such questions refer to the "most probable" i.e. anticipated choice intention. But ambiguity is unavoidable if the nature of intention is unqualified.

The data for questions on choice intention have varying implications for:

  1. the estimation of share of choices in a population, and

  2. Options Analysis validation procedures and analysis.

It is not necessary to pre-judge the issues involved as both questions can be asked in the same survey. The results should be analysed for substantive differences. If differences are not substantive then the issues outlined below may not be relevant. Such a finding suggests that appeal uncertainty amongst the preferred options is not affecting these choices.

Where there is strong separation amongst individuals' assessments of option appeal or an absence of uncertainty, or an appropriate combination of both, it is to be expected that questions on both intention perspectives will yield the same or very similar results.

Within the Options Analysis conceptual framework, all else being equal, the only reason for obtaining different choice results from the two questions, is an individual's uncertainty of appeal amongst the most preferred options. Overlap between an individual's option appeal distributions for the most preferred options, especially if these uncertainty distributions are negatively correlated, will suggest variation in choice amongst these options. Here, anticipated choice will usually be the option with the highest mean appeal. An immediate choice intention, where different from the anticipated choice, will likely refer to an option of lesser probability.


Using choice intentions data to estimate share of choice

Choice intentions data based on the immediate decisions of respondents may be aggregated to estimate option shares of choice.

On the other hand, aggregated data for anticipated choices will tend to exaggerate the share of the more common selections with a corresponding under-representation of less common choices.

However, at the level of the individual, data on an immediate choice will be less predictive of future choices than will data for anticipated choice.


Using choice intentions data for Options Analysis validity assessment

Appeal data is concerned with option evaluation. Relations between individual option appeal are used to infer choice intention. Each individual's statement of choice intention is used as an independent criterion for testing such inference.

With Options Analysis based on single-point measures of appeal, each measure reflects the sum of past and current experience and knowledge concerning an option. Since inference of option choice is based on highest appeal, which will invariably carry the highest choice probability, the results will reflect anticipated rather than immediate choice intentions. Using anticipated choice intentions is the most appropriate criterion measure for assessing the validity of these Options Analysis results.

Where there are substantial differences between estimates of shares for immediate and anticipated choice intention data it will be necessary to qualify Options Analysis using appeal-point data for the unanalysed effects of appeal uncertainty. The general results for such Options Analysis will be:

  1. For shares of the more popular options to be overstated, and less popular options understated.

  2. For individual changes in choice to be sharp rather than smooth - although overall change results will probably be similar. (Uncertainty suggests marginal changes in choice probabilities rather than whole-choice changes.)

Otherwise, the standard forms of Options Analysis may be confidently applied with the measures of appeal, advantage and disadvantage substantially unaffected by these considerations.


Notes index


Copyright 2012 Corporate R & D Ltd
All rights reserved. Email: