in Uncategorized

Political Psychology, the theory and the practice

Political psychology, looking at the way people make decisions about voting, and how politicians can manipulate them is a fascinating subject. This is an essay I wrote on the subject which gives some idea of the theory and practice of the discipline. It’s a little long (5000 words) but I like to think it’s well worth the trouble.

Political psychology aims to provide a methodical approach to an area which is usually considered more art than science. The political arena is more composed of hunches, personalities and ideas than it is of hard facts and well-tested statistics. In the last half-century, the scientific theory of persuasion has begun to find its place in the political spectrum. It has not been an easy development.

A lot of the early research confirmed intuitive beliefs about persuasion, but it failed to provide insights that would be of real value to a jobbing political consultant. More recently, as the machinery of political psychology has become more sophisticated and flexible, it has begun to prove itself a useful tool both for understanding political behaviour and for the business of winning elections.

As we shall see, the systematic approach to politics and persuasion has been most successful when it works within the broad stream of political science, rather than confining itself to laboratory-based studies, and when it engages with the reality of the modern political marketplace by providing valuable methodological guidance.

Since Plato?s time, and until the last century, persuasion had been seen as the business of the rhetorician, not the scientist. It was seen as a craft of words, similar to poetry or novel-writing, drawing inspiration from the work of politicians and philosophers of the classical era onwards.

The first steps towards the development of a science of persuasion and propaganda were taken in the era of the Great War. The change happened for a number of reasons ? increasing news coverage, increased literacy, and perhaps most of all, the sheer human toll which the war was taking on the people of England and Germany ?. It proved necessary to come up with comprehensive strategies to keep the entire nation motivated, and to keep allied nations loyal to the cause. The UK had its Ministry of Information and other countries rapidly developed equivalent functions.

Lasswell was one of the early theorists and practitioners of modern propaganda. As well as playing his role in the Second World War, he laid down a simple theoretical basis for the field of persuasion. Although his framework did not always provide many answers to the questions that arise, it certainly did provide a vocabulary with which to ask them.

The framework which Lasswell provided served as a core on which to build a science of persuasion. His model divided the process into parts which can be summarised as follows (Stone, 1974):
? Who?
? Says what?
? To whom?
? In what channel?
? With what effect?

To be fair, Lasswell conceived of this list as a checklist for the practitioner rather than as a theoretical framework for researchers. It is a reductionist model, and it clearly has faults. It does not capture the fact that all of these elements are snugly intertwined and any change in one of them (for example communicating with a different audience) is likely to result in implications for all of the other areas. The divisions themselves are somewhat simplistic. In his component ?says what? Lasswell does not distinguish between the intent of the communicator and the actual words which are used to communicate that intent. Thus a great deal of the richness of communication is lost. One other major failing, particularly from the point of view of the modern political practitioner, is that the approach focuses on conveying a message imploring a particular course of action, rather than focusing on the ?selling? of a particular candidate. In the modern election, the candidate?s image is as much concern to his political consultant as are the issues which he stands for.

On the other hand, Lasswell?s model provided a way to break down something, which, taken as a whole really allowed very little scope for analysis. It allowed for an independent variable to be manipulated (usually ?who? or ?to whom? or ?in what channel?) while the dependent variable (usually ?with what effect?) was measured to allow causal links to be determined.

Experimental Work
Much of the experimental work in political psychology in the 1960?s and 1970?s was driven by Lasswell?s research. The most prominent group carrying out this research was Carl Hovland who led the highly influential ?Yale Group?. This group attempted to use this model to dissect the nature of persuasion, with an emphasis on political persuasion. Hovland?s work, together with the work of his collaborators and others who followed up on his work, was oriented around a laboratory situation, where a very specific persuasive stimulus was used and the effect was measured in a comprehensive, managed way. (Stone, 1974)
They came up with interesting findings around each of these areas which have certain practical implications for political parties.

In terms of ?who? delivers the message, they found that the perceived credibility, expertise and attractiveness of the communicator were critical to persuasiveness. A strong speaker, who spoke without hesitation was more effective than a communicator with punctuated speech (Lind and O?Barr, 1979, reported in Petty and Cacioppo, 1996). A fast talker seemed to be more convincing than a slower one (Miller et al, 1976, in Petty and Cacioppo). Eye contact with the audience was also shown to be important (Hemsley and Doob, 1978).

This finding has been reflected most prominently in the use of autocue systems (the ?idiot box? or ?truth machine?, first used extensively by Ronald Reagan) which allow politicians to read from the screen while appearing to be looking directly at the audience.

Hovland and his colleagues found that certain groups would be more receptive to a message than others (Stone, 1976). They found that children were at their most persuadable around the age of 9. They found that women were more easily persuaded than men (although researchers now accept that this was the result of the subject matter rather than the subject). Personality factors such as self-esteem and intelligence could also play a part. They found that subjects could be ?inoculated? against persuasion, through the presentation and rebuttal of a weaker version of the argument.

This idea of ?inoculation? is partly reflected in the public relations practice of preferring that a candidate release bad news about himself or herself, rather than having it revealed in the media or by an opponent. It gives the opportunity to neutralise and control the message, and perhaps even turn it to his or her advantage.

Sherif, Sherif and Nebergall (1965), reported in Stone (1976) found that subjects with extreme views were less likely to shift their views than more moderate subjects.

This has clear implications for political campaigning. Rather than appealing to the entire audience, the canny campaigner now seeks out the most changeable group in the electorate and targets messages at them.

They found that different channels lent themselves to different presentations of messages. For example, a radio or television message needed to be quite simple. A written argument, on the other hand, could be expected to be quite complex, since readers could take their time in comprehending its argument.
Hovland graphed the effect of the persuasion effort over time. He expected the effect to diminish as memory of the arguments presented diminished, but in fact it did not ? it diminished much more slowly. Interestingly, as time went on the credibility of the persuader became less of an issue ? the persuader with credibility was no more effective than the persuader without.

The work of Hovland and others drew certain conclusions about the nature of the message as well. For example, they found that in a message containing a mix of good and bad news, it was better to present the good news first. However, other studies have contradicted this, and it seems likely that there are many interactions between different aspects of the communication which could have had an impact. This would certainly not have surprised a practitioner like Lasswell or Bernays. Crafting a political message does not lend itself to a simple set of rules-of-thumb.

Similarly, the use of fear in the message has been studied, and the findings of Janis and Feshbach (as reported in Cacioppo and Petty, 1996) are that these messages are effective, if there is a clear action that the recipient can take to avoid the negative consequences. As will be discussed later, the overall effect of negative messages in a political campaign was oversimplified and lost in experimental research

Politicians can also take guidance about how to present their arguments ? when it is appropriate to deal with the counter-arguments, and how many arguments should be given -. According to the McGuire (1964, reported in Stone, 1976) it is more important to present both sides where the audience initially disagrees with the position being taken.

Outside of the Yale group, Zajonc was investigating persuasion at a much more simple level. Rather than considering arguments and personalities, Zajonc concentrated on the importance of ?mere exposure? in increasing the attractiveness of an object or person. Zajonc took very simple stimuli, such as the names of vegetables or pictures randomly chosen from college yearbooks, and exposed them a varying number of times to the stimulus. Zajonc showed that continued exposure to a stimulus will increase the attractiveness of that stimulus regardless of what it actually is.

Perlman and Oskamp (1971) (according to Stone, 1974) investigated how different settings or different clothes effected the perception of study participants. Zajonc prepared photographs of the same models in different situations, i.e., in one condition, a model would be dressed as a clergyman, but in another he was dressed as a prisoner. They found evidence that associations with positive or negative objects or situations had a profound effect on participants? liking of models.

The work of Zajonc shows the importance of repeating an image and repeating a message in order to improve acceptance. This is a characteristic not only of western democracies, but also religious groups and despotic governments. Similarly, Perlman and Oskamp?s research provides obvious guidance for the associations that politicians should form about themselves. In a recent party political film, Fianna Fail?s footage of Bertie Ahern shows him in conversation with Bill Clinton, a figure held in almost universal high regard in Ireland. No footage was shown of Ahern with other players in the peace process, such as Gerry Adams, Tony Blair or David Trimble, since some members of the potential audience might take offence at these figures.

But there are some big practical issues with this research. In general, because of the difficulty of controlling the relevant variables in an area as complex as this, these studies had the following failings:

– They looked at one aspect of persuasion in isolation from all others. This meant that other related areas (such as image building) were neglected.
– As Hovland pointed out himself, it is premature to generalise from the typical samples (students and occasionally, army personnel) to ?real people?. The group had an age, wealth and education profile which was untypical. Cost and inconvenience make it difficult to get any other audience, although pioneers like Iyengar carried out studies in malls to get around these problems. Such experiments relied heavily on sponsorship, however, leading to further problems.
– Further, the experiments did not investigate the variety of personalities and attitudes of the general public, and how they typically interact with political messages. As an example, these experiments miss out on the different attitudes of the very different effects of campaigns on the most and least attentive voters.
– The environment where these experiments were carried out was generally somewhat artificial, resembling a classroom rather than a situation where voters would typically be likely to absorb political information. A participant might be put sitting beside a perfect stranger to watch a political advertisement ? a situation that would never occur in the real world.
– The experimenters were in a position to cut off other stimuli and demand the complete attention of study participants. Politicians do not have this luxury. The experimenters had a captive audience, who had some motivation to understand the argument and complete the tasks. In practice, voters do not have this motivation and do not take time out to listen to the arguments of politicians they do not support. As Zimbardo and Ebbeson say, ?The people you may want most in your audience are often least likely to be there.?
– The experiments miss out on the biggest practical problem most politicians face – how to get voters to listen to their message at all. The participants in the studies would have put some effort into evaluating the arguments presented. In practice, voters do not necessarily do this. The only issues they give extensive attention to in the real world are issues which effect them directly, and they may evaluate this type of issue differently from an issue which will have a neutral effect for them.
– The reality is that elections are not decided on the basis of arguments and their persuasiveness alone. The Yale experiments do not address the important role of issues such as the politician?s image, partisanship and voter turnout. However, these aspects of decisionmaking do not lend themselves to extensive study in the laboratory.
– The experiments miss out on the social context. For every claim in a political campaign, there is a rapid rebuttal. Repartee with political correspondents and hecklers is part of every election
Petty and Cacioppo (96) describe the elaboration-likelihood model which synthesises the most salient results of this research with an understanding of its clear failings. The model takes into account that there is more than one kind of persuasion at work, and that persuasion is a gradual, ongoing task, not something which occurs after one or two sittings. According to the elaboration-likelihood model, there are two main routes to change opinion and attitudes

1. The Central Route. This is the type of persuasion that the experiments of Hovland and others examined. Arguments are presented and evaluated, and if the subject?s attitude changes, it is likely to remain changed. The problem with this route is, that the subject must be in a position to process the information in the arguments. Thus it is often ineffective, because the subject may not have time to take in the arguments, may not know the relevant facts to be able to fully evaluate them, and is probably not motivated, because the issue is not particularly important to him or her.

2. The Peripheral Route. This is the type of persuasion that Zajonc and later, Perlman and Oskamp studied. It consists of repeated exposure to the stimulus, in combination with positive stimuli. The effect of it is likely to be short-lived, and could be supplanted by an alternative candidate or issue. Obviously it is not possible to convey anything but the most simple message in this way. However, this is the only way to reach a voter about an issue that they are not personally involved in or don?t know anything about.

Ironically, the very fact that a subject is convinced initially by the peripheral route and votes for a candidate or buys a product, may lead him or her to spend more time learning about the candidate?s policies or the product?s benefits, resulting in Central-Route persuasion taking place.

This model helps us understand some of the outlines of modern political messaging. We can see the different role that a TV advertisement plays, which is an attempt to provide exposure and association and something like a newspaper article, which is oriented at providing more full arguments which will lead down the ?central route?.

The key contribution of the elaboration-likelihood model to modern politics is probably the sound-bite. The model made clear that the ability of the public to absorb a message is limited, and that messages were best kept very simple. In television, the simple message is often accompanied by images which associate the politician with positive stimuli.

Survey Research
Other types of research addressed the problems that the Yale Group failed to address. By analysing the results of elections and surveys such as the National Election Survey, scientists have been able to get an insight into the actual operation of political persuasion.

The Yale Group?s work was not the only work that was done to study political behaviour. At the same time, considerable further analysis was being done of the results of elections and the apparent factors, using statistical techniques such as multivariate analysis. Although these studies did not focus on persuasion, they do give an important insight into the factors which drive voting.

?Panel studies? were used in US presidential elections in 1940, 1948 and thereafter and gave insights into changes in voting in the course of an election. Panel studies operated by keeping in touch with the same voter over the course of the campaign. As a result of the studies, Sears claimed (1969, in Stone, 1974) that up to 90 percent of the American electorate voted on a partisan basis in Congressional elections. Rossi also drew the conclusion that it was ?the least politically motivated and least knowledgeable? voters who were most likely to shift their votes.

More optimistically, V.O. Key (1966, in Stone) found that between 1932 and 1966, the changes in issue concerns have been reflected in choice of president. Of course, if such a finding were to be made today, it would be open to a much more pessimistic explanation ? that rather than the presidents changing to reflect concerns, that politicians and the media have manipulated the voters? agenda to suit the strengths of the candidates that were running.

That this is possible is reflected in the work of Iyengar, who found that exposure to news about particular topics on the television can shift an items importance within a voters perception (Iyengar, 2001).

Research into results of previous elections also showed that economic performance was a key factor in electoral success. As early as the 1940?s Bean recognised that the state of the economy was a critical part of the voting equation. Incumbents were more likely to lose during an economic downturn, and more likely to hold on during a positive period. (Lewin, 1991)

As a result, governing parties are likely to treat an election primarily as a referendum on the economic performance of the current government. In the election in Ireland in 2001, where the government was riding on the back of major economic growth, issues around the economy were the mainstay of the government?s campaign, providing the ?message of the day? for every second day of the campaign (conversation with Fianna Fail party researcher, 2002)
The analysis by Norpoth and Iyenegar (2002) of the Bush-Gore election, using results from exit surveys of voters attitudes was able to make a useful analysis of the factors which led to Gore?s defeat. He was able to determine that a failure to grab the middle ground, rather than a weak personality or the shadow of scandal were the underlying factors.

As with experimental work, there are major weaknesses in survey-based and electoral analysis-based political research. It is impossible to show conclusive causality. It is difficult to show the direction of causality; for example, it is hard to tell whether a particular person supported the Nice Treaty because he is a member of the Irish Green Party, or if he favours the Green Party because of their stance on Europe. There are always confounding factors which make it difficult or impossible to gauge the importance of any one factor, such as the effect of a particular newspaper item. Surveys are often dependent on accurate, honest recall by participants. Participants may forget what they voted for and why, or may even be ashamed to admit it to the interviewer. They are unlikely to be able to accurately recall what political broadcasts they actually heard. Results of attitude surveys are often not reflected in how people actually act. For example, they may be in favour of the idea of a waste management plan, but opposed to the building of an incinerator. No matter what the economic situation, a large proportion of voters will complain that they have difficulty making ends meet.

The best political researchers have come to rely on a mix of the two techniques. For example, Ansolabehere and Iyenegar (1997) delved further into the question of negative campaigning and the reasons for its success. By using the results of media monitoring and election analysis together with an experimental methodology, they were able to come to far more sophisticated conclusions about the effects of negative campaigning than Hovland could hope to. Their analysis, based on regression analysis of 34 Senate elections showed that negative campaigning worked by discouraging people in the middle ground from voting and so polarizing the electorate. At the same time, they conducted studies, in a naturalised setting in a shopping centre, where they showed a variety of political ads to partisan voters.

Their findings were interesting and directly relevant to the operations of political parties. Ansolabehere and Iyenegar found that political advertising works by reinforcing existing partisan lines and increasing the likelihood that partisan voters will come out to vote. They found that this was particularly the case with negative advertising, which was likely to discourage swing voters from voting. Certain candidates were able to benefit from this artificial depression in turnout.
Ansolabehere and Iyenegar also found that negative advertising also works because of its high news impact. The researchers referred to the well-known ?Peace, little girl? advertisement shown during the Lyndon B. Johnson?s presidential campaign against Barry Goldwater in 1964 as an example. It consisted of a picture of a young girl innocently counting the leaves of a daisy. When she has counted to nine, a male voiceover begins to count down, and the camera zooms in on the little girl?s face. When the count reaches zero, the little girl is replaced with a nuclear explosion, and the voice of Johnson, saying ?These are the stakes?to make a world in which all of God?s children can live, or to go into the dark. We must either love each other, or we must die?.

The advertisement was supposed to be played only once. Such was the newsworthiness, however, that it was played twice more over that weekend by other networks and without payment. The fuss created and the replaying of the ad created a great deal of coverage for the campaign. In fact nowadays, some political campaigners in congressional elections simply make a controversial advertisement and send it to the local television stations, without paying for a spot at all. They depend on the editorial staff deciding to play it on the strength of its newsworthiness.

The most recent studies have used similar methodological improvements to better understand the role of the media in political persuasion. Iyenegar (2001) describes the weaknesses of earlier studies, which failed to take account of the fact that most campaigns feature ?offsetting messages? which make it difficult to directly measure the effect of a single media spot. Furthermore, the researcher concludes that rather than having a direct persuasive effect, the media?s main role is to set the agenda for its consumers. Voters give greater weight to issues that have been raised and discussed in the media. Incumbent candidates and political elites can use this to their advantage, since they are usually in a position to make the news and so direct the media agenda to their areas of strength.

These findings from the latest research confirm theoretically what seasoned political pundits have probably instinctively known for some decades. Influence over the media is one of the key levers for electoral success, and that the influence is most effectively used to control agendas and to repeat messages, rather than as a channel for direct persuasion. This is something that political parties have been taking advantage of for some decades before it became clear in the research.

The Practitioner?s View
A number of important consequences and lessons for political parties are described above. But the best practitioners are looking not only at the results of research in political psychology, but also at the methodology which underlies it. They understand the strengths and weaknesses of different approaches and know that they have to take these lessons on board if they are to make the best of their advertising and polling budgets. In many ways, the methodologies of the academic scientist have been shaped to serve as the technological armaments of the modern electioneer.

Clinton?s election team was well aware, on the basis of surveys of voting partisanship and previous voting patterns that a relatively small proportion of ?swing voters? would decide the result of the election. Only by identifying these voters and appealing to them could he hope to win a second term.
Clinton?s polling firm, Penn and Schoen devised the ?neuropersonality poll? and administered it by telephone to tens of thousands of voters. They hoped to find a way to understand the mindset of the swing voter, and find out what differentiated Republican-oriented swing voters from their Democratic counterparts. This would allow the broad outlines of policy, and indeed the broad thrust of campaign appeals to be designed so that it would be appealing to these groups. (Stengel and Pooley, 1996)

But surveys and broad general direction were not enough. Clinton and his chief adviser, Dick Morris needed to build a bridge between the broad general thrust of their campaign and the specific messages that they would use to draw the attention of the swing groups. This required a mix of talents. On the one hand, the scriptwriters and TV producers provided the creative development that was necessary to interpret the results of the Neuropersonality Poll. On the other hand, the polling company had to experimentally verify that the images and scripts produced would actually have the desired effects.

This gave birth to the method of ?mall-testing? of political advertisements, which was used for the first time in the 1996 US presidential election. Subjects were solicited in shopping centres in swing states and were first asked questions about their background and political affiliation. They were shown a TV advertisement, and then asked another series of questions. The whole process took less than 10 minutes, and allowed a satisfactory sample of around 200 participants to be collected in an evening. The results allowed the campaign team to make decisions about tactics on the basis of objective numbers, rather than on the basis of ideology and subjective opinion alone. For example, the results allowed the pollsters to determine that the sentence ?The Republicans want to cut Medicare? was much more convincing than ?The Republicans want to cut Medicare so they can pay for a $245 billion tax cut for the wealthy.

Penn and Schoen did not limit themselves to testing with actual ads. They were anxious to determine how the opposition?s attacks were effecting their vote, and how they could be counteracted. They did this by fabricating opposition advertisements using animatronic technology and monitoring voters? reactions to them. As a result, Clinton?s team was able to quickly and confidently rebut his opponent?s economic plans. The team was also able to mock up its own ads before making them, thus saving the full expense of producing an advertisement that was unlikely to succeed.

Naturally enough, the techniques that are feasible and economic in a national campaign are not necessarily applicable to a local electoral race. However, there are alternative strategies, based on similar principles that have equal potential.

Politicians in the US have access to unparalleled resources to help them target the exact voters who are most likely to be receptive to their messages. A commercial database company, Aristotle, brings together information from credit, marketing and voter databases to provide politicians with the information they need. Using Aristotle?s website, aspiring politicians can create and download a list of voters or contributors who live in a particular area, have a particular affiliation and fit within certain economic or social criteria, such as number and age of children, number and type of cars owned, household income and so on. In many states, the records also contain information about political affiliations, making it easier to identify partisan and independent voters.

It is easy to see how these databases can be used to direct specific information at particular groups, in a way that would be far more sophisticated and fine-grained than what was done in recent presidential polls at national level. Iyenegar?s recent research into ?slate-mail? (the US term for circulars delivered during an election campaign) shows that this medium has large, unexploited potential.

In the UK and Ireland, the same techniques have not been applied, because the law resists the collection of such personal information on computer. Even the electronic distribution of voting registers has been heavily circumscribed preventing the extensive use of computer technology which has revolutionized political research and US electioneering.

However targeting goes on in a much less formal manner, where canvassers focus on particular areas where they were most likely to benefit from a partisan vote or gain from a swing of non-aligned voters. The practice is supported by information from informal vote tallies, which give a clear indication of shifts in voting patterns on a ward-by-ward and even street-by-street level. As new technology is introduced, however, it appears that the amount of fine-grained information available will diminish greatly.


Although the theory has often fallen behind the practice, academic psychology now provides some valuable insights for the jobbing political consultant. Whilst the conclusions will probably not be startling for most political commentators, the academic treatment means the facts are being presented now with far greater rigour than before. The methodologies employed are beginning to be sophisticated enough to give useful information about the undertones of political success. However, academic psychologists have to continue to pay careful attention to the real-life nous of professional experts and political parties.
The conclusions of the research lead us far past persuasion and the tricks and mnemonics of early propagandists. Agenda-setting, identification and targetting of swing groups and manipulation of turnout are among the key techniques of the modern pundit. Persuasion, in its traditional sense, now seems less important. Changing the attitudes of the electorate is no longer the centre of attention. It is far more important to identify potentially sympathetic voters and to build the agenda around issues which are important to them.

Professional experts and political parties also have a great deal to learn from the academics as regards the methodologies which should provide their technological armaments in future campaigns. There have been genuine leaps forward in the way political research is carried out in the last few years. The potential of the new techniques is only beginning to be exploited by political parties.

As the academic and professional streams of political psychology begin to form bridges, the area of persuasion and political communication is also changing form. Rather than being an uncomfortable combination of art and science, it is beginning to synthesise into something more akin to a craft. The political craftsman draws inspiration and knowledge from the best of both worlds.


Stone, W., (1974). The Psychology of Politics. Free Press: New York

Iyengar, S. (2001). Engineering Consent: The Renaissance of Mass Communications Research in Politics. Retrieved from website on 10 January 2003

Cacioppo, R. and Petty, J., (1996) Attitudes and Persuasion: Classic and Contemporary Approaches. Boulder, Colo. ; Oxford : Westview.

Norpoth, H., Iyengar, S, (2002) ?It Was About Ideology, Stupid: The Presidential Vote 2000?. Retrieved from]

Iyengar, S. (2001) Engineering Consent: The Renaissance of Mass Communication Research in Politics. Retrieved 10 January 2003 from website

Ansolabehere, S., Iyengar, S. (1997) Going Negative: How Political Advertisements Shrink and Polarize the Electorate. New York: Free Press

Stengel, R. and Pooley, E. (1996) ?Masters Of The Message?, Time, November 6, 1996. Retrieved 20 December 2002 from website

Write a Comment


  1. Nice work. It ads a bit of food for thought. I often think that the deadlock of the 2000 election is really the triumph of polling. Gore and Bush were both looking and polling for the issues that would attract 51% of the voters in key states. They both did tons of polling and hit the messages as coached. It would not surprise me to see more 50/50 splits in the elections ahead.


  • » VoteTube » Daisy Girl October 13, 2003

    […] The ‘Peace, Little Girl’ ad, as it is also known marked the beginning of the modern political ad. At the time, no one had ever seen anything like it and the news programs showed it a number of times. It is interesting to think about what gives it its emotional strength  Allow me to quote from my own magnum opus: […]

  • Blood Pressure October 13, 2003

    blood pressure diet