Home

Projections vs Surveys – which is more accurate in Elections 2019

Reading Time: 2 minutes

The main way of forecasting election results is by conducting a survey and asking respondents who they intend to vote if the election was to be held on the next day. The results from surveys are often not as reliable as we hope due to a number of factors – primarily the high costs of conducting a survey.

The reliability of surveys depend on a number of factors including how the sample was drawn, the number of people interviewed, were respondents interviewed telephonically or face-to-face, how honestly respondents replied  etc. These factors have contributed to conflicting estimates of what to expect after May 8.

Recent surveys suggest the ANC will get between 51 or 61 percent of the votes. Another way of forecasting results is to project the past performance of political parties forward. By, for example, looking at the percentage of votes each party received in the last eight elections (i.e. the last twenty years) it is possible to project that pattern forward for relatively short periods.

The reliability of the projections hinge on the past pattern of votes being maintained. If, for example, a party has been gaining support with over the past eight elections it is assumed that it will continue to gain support over the short term.

When the national vote share of the ANC is projected it seems that it is likely to obtain 55% of the total vote on May 8.

The ANC’s support in Gauteng has declined more rapidly and the projections indicate that it will obtain only 44% of votes cast in that province.

As far as the ANC is concerned the projections tend to be in line with the more pessimistic surveys. They do however indicate what can be expected. Several surveys, for example,  indicate that the ANC will get a majority of votes in Gauteng. However for the ANC to win that province, not only will the decline over the past election have to be halted, it has to be reversed sharply.

Author

MOST READ