Forecasting the 2016 general election
Forecasting an event requires looking at the frequency of its occurrence and the current factors that are likely to lead to this outcome, and quantifying this uncertainty with a probability estimate ranging from zero to 100.
The forecast of November 15, 2015 favoured the People's National Party (PNP).
It started with the two parties having a 50-50 probability of winning based on the dead heat in the polls in previous elections.
Voter turnout is decreasing and is likely to be near 50 per cent so the diehards will determine the election.
The PNP president and prime minister, Portia Simpson Miller, has greater acceptance among the party's diehards than Jamaica Labour Party (JLP) leader Andrew Holness among his party's diehards.
The PNP is likely to get more private donations because of party unity and the economic 'successes', all of which gave the PNP a 55 per cent chance of winning.
Our forecasts follow 10 rules.
1. Comparison classes should inform your probability esti-mate: We calculated the frequency of wins and the margin of victories in all constituencies between the PNP and the JLP from 1962-2011.
2. Hunt for the right information: We collected data on voting, the economy, policies, scandals, the parties, and polls from 1976 to the present.
3. Adjust and update your forecast: The dead-baby scandal pushed the forecast back to a 50-50 probability of either party winning. The updated forecast gives the PNP a 60 per cent chance of winning, with the appointment of a new health minister, infrastructural develop-ment and repairs, the public-sector wage settlements, the minimum wage and National Housing Trust loan increases, the 'economic achievements', JUTC expansion, ganja reform, and JLP disunity, and so on. We try to disprove our forecast by searching for counter evidence.
4. Mathematical and statistical models can help: We are building two statistical models to look at how economic factors influence elections.
5. Post-mortem analysis helps you improve: We have been forecasting elections and other events over the globe in the last two years followed by a post-mortem of the few incorrect forecasts for feedback to enhance our skills.
6. Select the right questions to answer: The question of which party will win is broken into smaller questions that help us to focus on appropriate data.
7. Know the power players: We collected data on voters, the party leaders, deputy leaders and vice-presidents, the media, the private sector and external actors that are the major power players in the election.
8. Norms and protocols of institutions matter: The norms and rules of the parties, their organisation and leadership structure, the Westminster system, the electoral system and electioneering were studied.
9. Other perspectives than power politics can help: A very organised political underdog can upset the political apple cart so people and institutions that can stymie the party and leader are being studied.
10. Wild cards, accidents and black swans can throw off your forecasts: We made a list of the factors that we think may undermine the forecast.
Our seat-projection method uses frequency of wins and margin of victories in all constituencies from 1962-2011 to determine the garrison, traditional and the marginal seats and the party likely to win more seats.
This data is contextualised with news about candidates and constituencies, and monitoring the field. Garrison and traditional seats are sure wins for either party so the fiercest election battles will be in marginal seats.
The PNP is projected to get 40 seats (garrison nine, traditional 25, marginal six) and the JLP 23 seats (garrison four, traditional 11 and marginal eight).
Our updated and final forecast will be published before the election on February 25, 2016.
- Dr Christopher A.D. Charles is a senior lecturer in political psychology and Gleasia Reid is an MPhil/PhD student in political science at the University of the West Indies, Mona.