Policy – What are the issues that drive voter choice? What policy mix will maximise votes? How to predict an election?
SCENARIO:
An opposition political party is facing an election in six months. They ask Future and Simple to find out what issues are important to certain segments of voters, and what mix of policies is most likely to win them the most votes.
F&S Approach
We propose a longitudinal study, comprising:
a MaxDiff experiment on issues of importance and their ranking, repeated at 3 or 4 intervals;
a DCM experiment testing a range of Government and opposition policy settings;
and an Information Conditioning experiment to test campaign advertising.
The repeated MaxDiff experiments will hone in on issues and assist policy formulation. Once draft policies are formulated, we will use the DCM to predict the optimum levels (e.g. tax rates) and mix (e.g. removal or instatement of a benefit) of policies. Just before the campaign begins, we will use the Information Conditioning to identify the most effective advertisements.
The DCM will predict the outcome of the election for a range of policy settings and advertising campaigns.
Plan and Deliverables
Literature Review (including the party’s current data and intelligence)
Voter focus groups to build a list of issues of importance
Best-Worst (max-diff) fieldwork. Over 4 studies, 2 weeks apart, we will derive scales of the rank and strength of importance of each issue. We will investigate:
Whether certain groups of voters form their scales of importance based on their party allegiance, and other groups vote based on how a party’s policies align with their scale of importance, and how to identify and describe each group.
How sophisticated a group of voters’ understanding of a policy or issue is, and whether the issue/policy comprises strong countervailing elements. For example, in a recent project, we have found a strong importance ranking for ‘Environmental Sustainability’ without any clear importance structure for a range of sub-categories, leading us to believe that there is not a deep knowledge of the sub-categories (underlying issues). In the same study we found that ‘Worker and Employment Rights’ were highly important, but a sub-category involving union activity was significantly lower in scale than all other sub-categories. One can assume this is a strong countervailing element of an otherwise important issue.
How we can identify latent classes in the population, comparing their importance scales with demographic and attitudinal responses.
DCM fieldwork. About three months out from the election, the results of the MaxDiff exercise have assisted draft policy development, and communication strategies. We now test a range of possible government and opposition policy and personnel, and predict the election outcome for any permutation of government and opposition policies. We produce a DSS which will allow strategists and policy makers to ‘war game’ the policy duel.
Information Conditioning. There are a few important factors we have not simulated in our predictive models – chiefly campaign advertising. Just before the campaign begins, we use another DCM, this time with Information Conditioning, to assess the impact on preference for certain policies, and on voting intention, of a range of advertisements. We identify the most effective advertisements and messages – both in general, and for specific segments.
Deliver an omnibus Choice Model, in the form of an easy to interrogate, interactive DSS (Decision Support System). This model will predict the election outcome for any permutation of government and opposition policies, personnel and advertising.
