While the term likely voter model may give the impression that its purpose is to model the results of a real election, this is not what it does. The aim of a likely voter model is to model the results of a hypothetical election at the time of data collection. It is thus a tool that measures the situation in the present but does not predict the future. We can liken this to measuring a patient’s temperature, where the result of the measurement indicates the patient’s temperature at that given time, but says very little about what the patient’s temperature will be next week.
As we move closer in time to a real election the results of a likely voter model should increasingly resemble what the actual results will be, as long as there are no extraordinary intervening events. Nevertheless, it is important to bear in mind when interpreting the results of the model that many voters make their final decision about who to vote for in the last weeks running up to an election. The degree of uncertainty in the model can thus remain quite high up until the last moment.
Unlike likely voter models, there are also election forecasts, which genuinely try to determine how the elections are going to turn out. As well as data from public opinion research, these models may also make use of economic data, such as the unemployment rate or the level of GDP, and take into account the history of past elections, or may also consider expectations in society about the results, as reflected for example in the odds set by bookmakers. This type of prediction is however not very common in the Czech Republic.
