WOULD I LIE TO YOU? What people tell the pollsters
In June 2017 there was a General Election in the UK. The organization Ipsos MORI, along with other Exit Pollsters proved to be fairly accurate in predicting the final outcome.
Table 1: Final number of Parliamentary seats for each of the GB political parties after the 2017 General Election
| PARTY | EXIT POLL | FINAL RESULT |
| Conservative | 314 | 318 |
| Labour | 266 | 261 |
| SNP | 34 | 35 |
| Liberal Democrats | 14 | 12 |
| Other | 22 | 23 |
Such Exit Polls have a good track record of accuracy in predicting the final results of a number of previous elections. Indeed, Ipsos MORI, the BBC and Sky News, all trumpeted their successful Exit Poll predictions.
However, just how informative are Exit Polls? Given that their correlation with actual election results is almost 1:1, there is a high degree of redundancy in the prediction. It is as if the pollster is standing on a beach, at the water’s edge, asking people paddling from the sea “Did you make a splash?
In essence Exit Polls are purely a measure of the ‘truthfulness’ of voters’ responses to the question “How did you just vote?” For the most part voters are replying honestly about their voting decision because the behaviour has just occurred only moments earlier. Individuals are inclined to answer (psychological compliance to the interviewer’s request), and to answer truthfully in order to avoid ‘cognitive dissonance’. In other words, respondents would feel ‘uncomfortable’ lying about a recent piece of behaviour.
The Exit Poll figures and the associated election-result ‘predictions’ made by statisticians, politicians and specialist correspondents plays out as a piece of real theatre. The tension builds as the actual counts from the constituencies trickle-in, beginning only a few hours after the polling stations have closed, and the graphs, coloured maps and ‘swing-o-meters’ support or embarrass studio guests in equal measure.
However, any predictive quality contained in the Exit Poll results exists for a few hours only, as an ‘emotion of expectation’ in a media-generated drama.
The Exit Polls and actual voting behaviour are happening close together both in lapsed-time and psychological-space. Consequently, the quality of ‘accuracy’ being measured does not belong in the realm of prediction, but is an indication of how well the sample of voters in the Exit Polls matches the sample of actual voters. The ‘prediction’ relies on the Exit Poll sample being representative (in terms of demographics) of the actual voting public.
Tellingly, this predictive-accuracy achieved by the Exit Polls, is not replicated by the pre-election surveys.
Table 2: Final voting share achieved by the political parties in the 2017 General Election
| PARTY | Ipsos MORI prediction | FINAL RESULT |
| Conservative | 44% | 43.5% |
| Labour | 36% | 41% |
| SNP | 5% | 3.1% |
| Liberal Democrats | 7% | 7.6% |
| Green | 2% | 1.7% |
| UKIP | 4% | 1.9% |
| Others | 1% | 1.2% |
Here it can be seen (Table 2 above) that the share of votes for the Labour Party was underestimated by 5%.
The 5% error, although seemingly small, meant that the prediction of a Conservative majority was wrong and though remaining the largest party, it no longer had an overall majority in the new Parliament.
Ipsos MORI suggested that the error was due overestimating the ‘age-effect’. It is ‘known’ amongst pollsters that younger people are more likely to support ‘anti-establishment’ parties but are less likely to bother voting. It is supposed that more youngsters turned-out to vote than expected, and hence the 5% difference.
What people really think of our politicians?
In the run-up to the 2017 Election, the author conducted a small study involving a small random sample of eligible UK voters.
Unlike the traditional Pollster-surveys, this study used a unique technique developed formerly for use in marketing and brand research, (registered as Emotix™)[1].
Participants asked to respond intuitively to a pairing of a party logo and its association with a likely or unlikely characteristic.
The resulting data showed an estimate the strength of feeling that respondents associated with each of the main political parties.
Each Party had a cluster of word associations, which represented its brand–image in the mind of the General Public.
The Party brand-images that emerged were very interesting and contained both positive and negative characteristics.
Conservative
Positives: Decisive; Determined; Dynamic; Fair; Reliable and Tough.
Negatives: Arrogant; Boring; Materialistic; Pompous and Ruthless.
Labour:
Positives: Compassionate; Determined and Fair.
Negatives: Self-centred and Untrustworthy.
Liberal-Democrats:
Positives: Determined
Negatives: Boring; Pompous and Self-centred.
In numerical terms, the ‘spikes’ or ‘strengths’ in the Party-profiles are shown below in Table 3 below:
Table 3: Strength of the characteristic associations across the three main parties
| Conservative | Labour | Liberal Democrats | |
| Arrogant | 80 | 50 | 45 |
| Boring | 85 | 55 | 65 |
| Compassionate | 65 | 70 | 55 |
| Decisive | 75 | 60 | 45 |
| Determined | 75 | 80 | 65 |
| Dynamic | 75 | 45 | 30 |
| Fair | 65 | 70 | 45 |
| Materialistic | 75 | 50 | 50 |
| Pompous | 80 | 55 | 60 |
| Reliable | 75 | 55 | 55 |
| Ruthless | 90 | 55 | 55 |
| Self-centred | 70 | 65 | 55 |
| Tough | 75 | 60 | 45 |
| Untrustworthy | 50 | 70 | 40 |
The Conservative party is seen as tougher, more dynamic and more decisive than either Labour or the Lib-Dems.
The Conservatives are also seen as more arrogant, more pompous and more ruthless.
However, the results also show that Conservatives are perceived as being capable of being fair and compassionate and are considered more reliable than the other two main opposition parties.
Labour, as one would expect are seen as being more compassionate than the other two parties. They are also perceived as being fairer. However, for whatever reason they are perceived as being much more untrustworthy than either the Conservatives or the Lib-Dems.
The pattern revealed in the results is worrying for the Lib-Dems. They have a ‘flat’ profile, with no outstanding strengths when compared to the Conservatives and Labour Parties. The results are showing a Party with a weak personality. Within their profile the stand-out perceptions are determined, boring and pompous.
The results show how potential voters were thinking and feeling about the political parties[2]. If asked outright, their responses might be completely different. These results show the ‘hidden’ or subconscious associations that would-be voters hold, and their inference for future voting behaviour is clear.
KEY TAKEAWAY: This polling study illustrates the usefulness of measuring emotional responses. This information has a richness and quality that reveals the nature of each Political Party as perceived by the public. The resultant ‘brand-images’ are in fact clusters of emotional associations held in the sub-conscious part of the voters’ minds and will have a powerful influence on their future decision making.
Dr N Marlow
November 2020
[1] The design of the online engine is underpinned by Implicit Attitude Theory (IAT). This is a widely accepted theory in psychology, and has been tested in various forms since its inception in the late 1990’s.
(For a seminal paper, see: Greenwald, A. G., McGhee, D. E., & Schwartz, J. K. L.(1998). Measuring individual differences in implicit cognition: The Implicit Association Test, Journal of Personality and Social Psychology, 74, 1464-1480)
[2] The results show prospective voters can simultaneously hold both positive and negative thoughts about a particular political party. This seemingly illogical position would be difficult to detect using traditional questionnaires.