When it launched its election manifesto in February, the Democratic Alliance (DA) took a swipe at the quality of expert commentary on the election and their analysis of party support trends. The party accused the analysts of systematically underestimating how well the DA would perform in the upcoming national elections and, by implication, undermining confidence in them.

The comments by the DA follow a series of surveys which show that support for the DA has declined in recent years. The latest report by the Institute of Race Relations (IRR) has been the most optimistic with respect to DA support.

The IRR survey indicates that support for the DA had declined by about 1% since the 2014 national election. They indicate that the DA would get about 22% of the vote if the election was to be held now. Two weeks ago, a survey by the Association for Free Research and International Cooperation (Afric) indicated that the DA would get less than 10% of the vote compared to 17% for the EFF and 58% for the ANC.

This survey indicates that, in sharp contrast to the DA’s expectation of growing support, should an election be held now the DA would no longer be the official opposition.

All of these polls are sorely lacking in one key regard – the intention to vote for a party as declared in a survey has a relatively poor correlation to election results. There is a significant gap between a respondent expressing support for a party and them actually voting for it. Despite supporting a political party many voters may not actually cast a vote for it for a range of reasons. Perhaps the individual is registered or, most importantly, simply do not feel motivated on voting day.

In the 2014 national election, almost half of the voting age population (46%) did not vote. Part of this was due to many citizens of voting age not being registered to vote. However, almost one quarter of those who were registered did not cast a ballot on the day. There are strong indications that the voting rate is sure to drop further in the upcoming election possibly resulting in more than half the voting age population not cast ballots. It is this factor that the surveys do not adequately account for in their estimates.

The likelihood of an individual voting varies dramatically between social, demographic and economic groups. Usually older, more educated and wealthier individuals are significantly more likely to exercise their vote than people who are younger, less educated or economically marginalised.

As a result surveys routinely understate the vote share of parties (like the DA) whose supporters tend to be more educated, wealthier and older.  Conversely the surveys routinely overstate the vote share of parties (like the EFF) who have a younger support base or whose supporters are more likely to be unemployed or otherwise marginalised.

Moreover, if the vote share of any one party is distorted by a survey the share of all other parties are invariably misrepresented – bringing into question the reliability of the survey.  As a result of the above,  the DA is almost certainly correct when it claims that surveys understate its share of votes. Similarly the surveys almost certainly overstate the extent to which the EFF will win votes in May. This said, even the most rigorous analyst has to question whether of not there has been a real decline (or simply a stagnation) in DA support. To answer this they have to look to other data like media coverage and by-elections.

There is one advantage the DA has over all other political parties. In 2016 it, with alliance partners, won control over a significant number of municipalities including the metropoles of Johannesburg, Tshwane and Nelson Mandela Bay. Had the DA been seen as running those areas well their new role should have impacted positively on support for the party. Rather their new influence has proved to be a double edge sword. After 2016 expectations of the party in terms of the way in which their municipalities were run were suddenly raised. This raising of expectations was largely of the DA’s own making as the promised service improvements and the rooting out of corruption etc.

However even if their new role resulted in improved municipal administration the benefits to individual voters was, at the very best, diffuse. While expectations can rise instantaneously the turning around of a municipality’s’ administration takes time and when it does happen the results are (initially) barely perceptible. The lag between reality and expectation is certain to lead some disillusionment with the new administrations – irrespective of how well it is performing.

However the new administrations can make an immediate impact on how well residents are represented. As far as democracy is concerned appearing to hear voters is as least as important as improving services. Here, in many instances, the DA has failed spectacularly.

In several instances, the DA has rewarded party loyalists by offering them ‘safe seats’ in by-elections. Examples of this include the party putting up a PR councilor in Johannesburg’s Ward 88 – a seat considered a ‘safe’ DA ward. The candidate was duly elected but then proved unable to make the transition from representing the party (as a PR councilor) to representing the residents as a ward councilor. While the move may have bolstered the DA’s capacity in Johannesburg Metropolitan Council it resulted in increasing disillusionment among residents of Ward 88.  In the process of giving the seat to an acolyte the worst stereotypes of the DA (including that of being “arrogant”) were affirmed and the party’s’ performance in the ward is sure to suffer.

It is important not to underestimate the impact poor representation can have on voters. In at least one instance the consequences have been spectacular. After winning 62% of the vote in the 2016 local government election the DA lost eThekwini Ward 77 to the ANC in a by-election. In the space of a year their support dropped from 62% of votes cast to 38%. While it is difficult to gauge how widespread poor representation by DA party representatives is there is strong anecdotal evidence that it may cost the DA a significant level of support.

While the DA may be correct in their criticism of the analysis of surveys the party cannot rely on voters drawing a distinction between how well it represents them at local level and how well it represents them in parliament. It is the quality of representation in key wards rather than the content of manifestos that will hinder progress. At the very least the DA should approach the May election with some trepidation and not bet on the pundits having misread survey results.