Analysis of Dichotomous Outcomes
A Case Study of Constituency Characteristics' Data in the 2006 General Elections in Uganda
What influence do demographic, socio-economic and political characteristics of Constituencies have in influencing the results of an election? To answer this question, the Hosmer-Lemeshow logistic regression technique for step-wise analysis of dichotomous outcomes was used on election statistics of 2006 and data from surveys carried out in Uganda.
Participation of the Constituency voters, their average monthly per Capita incomes and political party dominance characterising a Constituency were found to be reliable, statistically for the restricted model , in correctly predicting 70% of the data.
The results would imply, for strategists in multiparty systems, that political and coalition actors could optimise their decisions or vote chances by articulating alternative policy matters in their campaign agenda.
Unobservable income and other forms of inducements or coercion on which data are seldom collected or lacking may influence outcomes of polls.
These could provide room for future studies.