Abstract
Some of the most important propositions in the political marketing literature hinge on assumptions about the electorate. In particular, voters are presumed to react in different ways to different orientations or postures. Yet there are theoretical reasons for questioning some of these assumptions, and certainly they have seldom been empirically tested. Here, we focus on one prominent example of political marketing research: Lees-Marshment’s orientations’ model. We investigate how the public reacts to product and market orientation, whether they see a trade-off between the two (a point in dispute among political marketing scholars), and whether partisans differ from non-partisan voters by being more inclined to value product over market orientation. Evidence from two mass sample surveys of the British public (both conducted online by YouGov) demonstrates important heterogeneity within the electorate, casts doubt on the core assumptions underlying some political marketing arguments and raises broader questions about what voters are looking for in a party.
Parties, voters and political marketing
The major political parties in Britain and elsewhere face a dilemma. They have had to respond to a decline in partisanship and hence a thinning out of their traditional constituencies since the 1970s (Dalton, 2000; Schmitt and Holmberg, 1995) by broadening and diversifying their appeal. Such competitive pressures and their consequences were not an entirely new phenomenon (Downs, 1957; Kirchheimer, 1966). Yet it is fair to say that parties have substantially transformed in recent decades not just by utilizing more professional methods of electioneering but ultimately by internalizing an organizational philosophy that has been described as a ‘political market orientation’ (Ormrod, 2005).
This concept of political market orientation and the way in which it is contrasted with the electoral strategies of more ideological and cleavage-based parties has a lot in common with Wellhofer’s (1990) earlier distinction between market models of ‘vote maximization’ (essentially Downsian competition) and ‘vote production’ (the mobilization of traditional, cleavage-based support by ultimately ‘lazy monopolists’). The difference is that, whereas Wellhofer described these as two different (but equally problematic) market models of politics, political marketing scholars outline an evolutionary process which fairly closely reflects recent patterns of partisan change in Britain and elsewhere: Parties in secular, welfare states were decreasingly able to rely on appeals to class, religion and ideology and increasingly forced to broaden their bases of support among diverse interest groups. In marketing terms, this general trend of weakening party allegiances represents a shift in market power from producers to consumers, precisely the change that transformed business philosophy from production to marketing. (Scammell, 1999: 726)
Scholars in political marketing are divided about the electoral implications of these differing orientations or postures. Lees-Marshment states baldly that ‘if major parties want to win general elections in Britain today they need to adopt a market orientation’ (2001b: 1081). However, there is dispute about the cross-national applicability of such a claim (Lilleker and Lees-Marshment, 2005; Ormrod, 2006), and more recent contributions to the literature have cast doubt on the validity of such a clear prescription (Coleman, 2006; Henneberg, 2006; Henneberg et al., 2009; Henneberg and O’Shaughnessy, 2009). In particular, Henneberg (2006) argues that a party adopting the posture of following, which he also describes as being ‘customer-led’ or ‘market-driven’, can fall into the trap of being perceived as a ‘tactical populist’ (pp. 32 ff.), a party that simply strives to win at all costs by following public opinion, seemingly lacking clear principles. A wider tendency towards market-oriented, customer-led or market-driven parties across the spectrum may even be responsible for the decline in political trust (Dalton, 2004) and falling turnout (Lutz and Marsh, 2007) across Western democracies.
We should also bear in mind that party identification and cleavage voting have diminished but not disappeared over recent decades. Herein lies the dilemma mentioned at the outset. Parties must balance their appeals between, on the one hand, attracting new followers through ascertaining and then satisfying their political demands, and, on the other hand, reassuring the remnants of their traditional following through continuing emphasis on core political principles. A minimum requirement to that effect had already been spelt out by Downs (1957), who argued that, while parties will need to move towards the median voter to maximize their vote, they cannot leapfrog other parties if they want to retain some basic credibility. 2 A further challenge for market-driven parties arises from their increased sensitivity to diversity among electorates and hence from marketing’s emphasis on segmentation and targeting (Smith and Hirst, 2001). Customizing their offerings to the particular demands of specific target audiences may render parties vulnerable to a challenge of inconsistency or arbitrariness in their political stances.
The responses of political-marketing scholars to this dilemma have varied. Lees-Marshment (2001a, b) is the most obviously comfortable with the notion that there is a definite opposition between product and market orientation, or at least a continuum running from pure product to pure market orientation, and that parties have to choose a position along this continuum. Others reject such a trade-off, arguing that postures of leading and following are not mutually exclusive and that parties can simultaneously embrace both (Henneberg, 2006: 33, 38 ff.). In particular, the concept of ‘relationship marketing’ – ‘building long-term trust and commitment based relationships’ (Henneberg and O’Shaughnessy, 2009: 10) – has been posited as a means of reconciling marketing’s conceptualization of politics as a process of exchange with the social bases of partisan politics. Henneberg et al. (2009) argue that this approach is compatible with – even conducive to – a deliberative model of democracy. As such, it has the potential to be part of the solution to public disenchantment with democratic politics rather than part of the problem.
One major concern, then, for political marketing research is to clarify the nature of the relationship between leading and following. How – and how easily – can parties be flexible and adaptive to the electoral market while still emphasizing their core values and beliefs (or ‘brand heritage’ as Henneberg (2006: 38) describes these in marketing terms) in order to reassure partisans and to appear generally principled. Answering these questions involves both theoretical and empirical work, a good deal of which is under way. However, one important facet of the task is to consider how voters themselves react to parties’ postures or orientations. And that facet has been neglected so far, by both political marketing scholars and students of electoral behaviour. Addressing that neglect is the purpose of this article.
One striking feature of political marketing research is that its empirical focus is almost always on parties rather than voters; that is, on sellers rather than buyers in the market for votes (see Lloyd (2006) for an exception). And this may partly be the case because ‘[marketing] models are not designed to offer post-hoc scientific explanations of voting behaviour but to produce a broad-brush framework within which parties and candidates can think about the development of strategy’ (Bartle and Griffiths, 2002: 30). Marketing theories tend to be prescriptive, particularly those proposing that political parties ought to become ‘relationship builders’ (and thereby both lead and follow), while acknowledging that no such parties exist to date (Henneberg, 2006: 38 ff.). The respective merits and shortcomings of leading and following, and the extent to which there is a trade-off between the two, are points that are being argued rather than tested. And these arguments frequently involve assumptions – whether explicit or implicit – about voters, and such assumptions cannot be directly tested using evidence at the level of parties or election outcomes.
At the same time, the individual-level evidence collected in election studies and other political surveys has not been geared towards gauging public responses to different and changing party orientations. In this article, we make a first attempt at testing prevalent assumptions in political marketing theory on a representative sample of voters, thereby broadening our understanding of public attitudes towards political parties. It addresses questions that should be of interest to students of parties and elections in general, as well as those interested more specifically in political marketing. How do voters tend to respond to different and changing postures and strategies of political parties? And how do they see the relationship between these postures or orientations? Are product and market orientations viewed as mutually exclusive (Wellhofer, 1990) or as a zero-sum game, a continuum where increased market orientation entails reduced product orientation (Lees-Marshment, 2001a, b)? Or are public attitudes consistent with Henneberg’s (2006) assertion that parties can and should aim to lead at the same time as follow?
Beginning to test theories of political marketing on voters
As in any strand of political research, there are differences and disagreements among political marketing researchers. Whether or not a ‘political marketing paradigm’ can be identified, there is plainly no theoretical framework or set of assumptions that is common across all those working in the field. In our empirical work, then, we cannot therefore test some putative ‘theory of political marketing’ – there is no such thing. Nor can we confront voters with anything like the full range of assumptions and theoretical nuances evident in the literature. Even assuming infinite patience on the part of survey respondents, the problem remains that a closed-ended questionnaire approach is unsuited to capturing that kind of subtlety. In short, we have to choose which model of party marketing strategy to test. For three reasons, we opt for Lees-Marshment’s (2001a, b) orientations model. First, it is best known to students of electoral behaviour and so is an obvious starting-point for those seeking to bridge the gap between political marketing and that sub-discipline. Second, British parties were central to the conception and development of Lees-Marshment’s model, and these are the parties which are evaluated by voters in our research. Third, Lees-Marshment offers unusually clear descriptions of party orientations, descriptions which come in useful for our survey work.
These reasons are practical more than substantive. There is no reason why competing theories or models within political marketing could not be subject to parallel empirical work with voters. Ultimately, we have to start somewhere. That is not to say that the analyses here should be regarded as irrelevant for researchers favouring alternative interpretations and frameworks. First, the fact that a study is intended to test one model does not preclude it from generating evidence pertinent to other models. Survey design becomes crucial here: the questions must not be based on the assumptions underlying Lees-Marshment’s model, but should test them. For example, questions about product and market orientation should allow the two to be seen as wholly incompatible, wholly compatible and any point in between. Thus, the responses to such questions can bear on political marketing research beyond Lees-Marshment’s work. Second, and related, terminological differences should not obscure the fact that there is at least some shared understanding of core concepts. For instance, the distinction between product and market orientation runs pretty closely parallel to Henneberg’s (2006: 34) distinctions between leading and following, and between market-driving and market-driven orientations. Hence, again, there is scope for a study of Lees-Marshment’s orientations to have wider relevance. Next, we set out three core research questions that the survey was designed to address.
Q1. Do voters recognize the product/market distinction?
According to Lees-Marshment (2001b), a party’s electoral prospects depend (at least in part) on its orientation. That does not, of course, presuppose that voters actually notice and distinguish different orientations. To invoke a closely related example, a Downsian theorist could argue that voting decisions conform to a spatial logic even if voters are not necessarily undertaking the calculations (see Friedman, 1953: ch. 1). And Lees-Marshment (2001c: 700) argues that the success of the Conservatives in the 1980s was owed to the party’s ability to adopt a market orientation while maintaining a reputation among voters as product-oriented, thanks in particular to the established view of Thatcher as a conviction politician. The implication is that market orientation may prove highly successful even if – perhaps especially if – undetected. Indeed, ‘those in any party may deny being market-oriented for all sorts of reasons’ (ibid., p. 705).
On other occasions, however, Lees-Marshment implies that the benefits of a market orientation will only accrue if voters recognize that shift. She links market orientation to the fulfilling of promises in government, arguing that it can thereby help in overcoming public cynicism about parties, but acknowledges that this reputation for delivery depends on ‘the extent to which [parties] become market-oriented and voters become aware of it’ (2001c: 700, emphasis added). More fundamentally, the paradigm case of political marketing, ‘New’ Labour, is argued to have achieved its success precisely because voters did notice and responded to the party’s shift from product to market. The notion that parties seek to conceal their marketing orientations leads to some odd inferences, for example that Tony Blair was wincing with embarrassment after the very public conference battle over Clause IV of the party’s constitution, rather than delighting in that opportunity to spotlight the party’s change in approach. 3 Since political marketing theories are precisely about how parties present themselves to and are seen by voters, it seems a reasonable empirical question to ask whether differences in orientations are noticed by the voters.
Naturally, we would not expect voters to be familiar with the terminology of product and market orientations. However, when these notions are explained in straightforward language – about parties sticking by what they traditionally believe in, or changing their policies in line with public opinion – then there seems no major obstacle to understanding them. As Ferejohn (1995: 107) points out, the spatial metaphor is in common usage well beyond academic treatments of party politics. And in Britain, the empirical focus of this study, the key developments in electoral politics over the past quarter of a century have been often and easily framed in terms reminiscent of the product–market distinction. Both ‘New’ Labour’s triumphs and the healthy Conservative poll leads throughout 2008 and 2009 were interpreted as showing that those parties have successfully manoeuvred into the ‘centre ground’ of politics. Equally, both parties’ periods in the electoral doldrums are associated with a lurch to the extremes – or, less melodramatically, with a retreat into traditional values and priorities – and thus a failure to attract more than core support. We are interested not only in whether voters understand the terms but also, in particular, in whether they recognize the trade-off between the two orientations. As Lees-Marshment (2001b: 1082) points out, the different orientations are ideal types and most real-world parties will show a mix of the two. Nevertheless, it is inherent in her definitions that a party can only become more market-oriented by shifting away from a product orientation, or vice versa. Furthermore, the idea that parties face such a trade-off also underlies the narratives of British politics sketched above. Evidence that voters see no such conflict would certainly raise questions about their use and understanding of these ideas.
Q2. Is a market orientation more popular?
Earlier, we quoted Lees-Marshment’s assertion that a major party ‘needs to adopt a market orientation in order to win a general election’. This suggests a straightforward ‘yes’ in response to Q2. We should restate the caveat that adopting a market orientation is not the same as being recognized by voters as doing so, and that (as quoted above) ‘those in any party may deny being market-oriented for all sorts of reasons’. This does hint at the idea that an obvious market orientation may incur electoral costs, and Lilleker’s (2005) focus group data do suggest that some non-voters in 2001 would have turned out but for ‘New’ Labour’s shift away from its traditional product. More often, however, the clear implication of Lees-Marshment’s arguments is that market orientation is electorally beneficial, and that it need not be surreptitious to deliver these benefits. After all, as noted earlier, some of the key examples of market-oriented parties have sought publicly to make a virtue of their responsiveness to public opinion.
This is not to say that market-oriented parties are argued by Lees-Marshment to be nothing but slavish pursuers of public opinion. There are inevitably constraints on these parties’ responsiveness to voters, notably the need for promises to be deliverable and the need to keep party members happy. Nonetheless, the basic argument holds: to the extent that a party is able – within these limits – to respond to voters’ wants and demands, it will be more electorally successful. Any indication that market orientations actually prove less popular would certainly raise serious questions about Lees-Marshment’s core arguments. Certainly, if these arguments are more than a restatement of the Downsian model, parties should be rewarded not only for moving closer to the median voter but more broadly for adopting an approach to politics that places responsiveness above a commitment to existing stances (Lees-Marshment, 2001c: 696–700). In order to test this, we examine whether voters who see a particular party as more market-oriented regard it more favourably than those who see that party as predominantly product-oriented. Related, but introducing an element of comparison, we examine whether voters who see one party as more market-oriented than another tend to prefer the former. In both cases, we also examine the relationship between perceived orientations and attitudes towards the parties while controlling for a comprehensive set of other variables established in the electoral behaviour literature as significant predictors of party choice. We can thus assess whether orientation perceptions are able to make an independent contribution to our understanding of how attitudes to parties are formed.
Q3. Is preference among orientations moderated by partisanship?
Dunleavy (1996: 280–281) criticizes political scientists in general for their failure adequately to recognize heterogeneity in the electorate, in particular noting frequent use of the ‘archetypal singular’ when describing voters. The complaint applies to much of the political marketing literature as well. It is easy to refer to ‘the voters’ as if they all share the same needs and wants, the same demands of and reactions to parties. This is self-evidently not the case. For one thing, winning a British general election is not about winning the support of ‘the voters’; it is rather about winning support from enough of the swing voters in marginal seats (Lilleker, 2005; Ormrod, 2005). 4 Such ‘swing’ voters are by definition not party identifiers, and this presumably explains why researchers in the field have paid so little attention to partisanship. The ends of political marketing can be achieved without recourse to those voters who report such an attachment. Q3, then, does not derive directly from the political marketing literature. Often the only mention of party loyalties is a note that they are weakening – it is this trend that is held to have opened up the market for votes (e.g. Lees-Marshment, 2001a: 14; Newman, 1999: 126).
Yet, dealignment notwithstanding, self-identified partisans still make up around half of the British electorate. And we might expect their reaction to orientations – certainly to the orientation of ‘their’ party – to differ from the reaction of those swing voters who are the target of a market orientation. Put simply, partisans tend to like the existing product. This is partly because they are predisposed to endorse what their party does and says anyway. It is also because, at least on a Michigan-style reading, partisanship derives from social location, and social location influences the kind of product that voters want (Campbell et al., 1960; Green et al., 2002). The link is nicely encapsulated in the term ‘heartlands’, the areas in which a party is best liked and in which its existing product is most in demand. A familiar argument from political debate in Britain over the past decade is that ‘New’ Labour deserted these heartlands – heavily industrialized and unionized urban areas, predominantly in the north – and thus alienated those voters who favoured the traditional left-of-centre line of products that was discontinued (see Lilleker, 2005; Scott, 2000; Wring, 2005). 5 We would therefore expect product orientation to be relatively popular among supporters of the party in question, and market orientation (for all parties) to be relatively popular among non-partisans. If those expectations are confounded, that raises questions both about the nature of partisanship as sketched above and, more importantly for present purposes, about the need for a market orientation in order to win over swing voters and thus win elections.
Data and measures
The data come from two surveys of the British electorate, both conducted over the Internet by YouGov. The first was a YouGov omnibus survey, fielded between 29 May and 2 June 2008 and with a sample size of 2,461. The second was fielded between 24 February and 3 March 2009 as part of the British Election Study (BES) Continuous Monitoring Survey. The sample size was 951. In both cases, results presented here are based on data weighted to make the sample representative of the British electorate (in terms of a combination of variables including age, gender, social class, newspaper readership and past vote). Most of the data analysis in this article is based on the first survey, in which we explore the relationship between orientation perceptions as well as their association with partisan attitudes. However, only the second survey offers the kind of wide array of control variables necessary to test whether orientation perceptions have an independent impact on party preferences.
The core measures in both surveys are those gauging respondents’ perceptions of the orientations of the three main parties (Conservative, Labour and Liberal Democrat). In order to make these as valid as possible a test of Lees-Marshment’s model, the question stems were drawn almost verbatim from her descriptions of those orientations (2001a: 28–30). Since, as noted earlier, scholars in political marketing differ over the compatibility of these orientations, we were interested to see whether voters envisaged a trade-off. Hence, in the first survey, we asked about the two orientations separately. The first question was: ‘A product-oriented party argues for the policies it believes in, and sticks by them even if they are unpopular with many voters. Please say how accurately you feel this describes each of the following parties.’ A seven-point response scale was used, running from ‘not at all accurately’ to ‘very accurately’. There followed a parallel question, this time opening with: ‘A market-oriented party finds out what most voters want, and is willing to change what it stands for to meet those demands.’ 6
In the second survey, constraints on space meant that we were not able to ask about the orientations separately but instead asked respondents to place the parties along an 11-point product-oriented to market-oriented scale (with the orientations described as above and the extreme points labelled as ‘strongly product-oriented’ and ‘strongly market-oriented’). This practical necessity has a methodological advantage, allowing us to test whether the key findings hold across different operationalizations of the core measures. Specifically, this second survey imposes on respondents the same trade-off envisaged in Lees-Marshment’s approach.
It might be wondered why we included the specific terminology rather than simply describing what these orientations are. The reason is that, since we were asking the same questions about several different parties, repeating the full descriptions would quickly have become tiresome for respondents. At the same time, trying to condense the descriptions would have compromised our claim to be testing the orientations as described by Lees-Marshment. The same can be said of swapping in more idiomatic descriptions like ‘a party that follows’ and ‘a party that leads’. While these have the virtue of brevity, they also bring in a range of other considerations that would have seriously impaired the validity of the study. Indeed, the fact that respondents will not have heard of the orientation terms before, and would be unlikely to grasp their content without a description, could be argued to be an advantage. Responses are likely to be driven more by the way that the orientations are described than by the labels. The latter are simply intended as a short-hand to remind respondents of those descriptions. One implication of this is that the results can still be of interest to those who reject Lees-Marshment’s choice of labels for the orientations that she outlines.
In order to address Q2, concerning the popularity of different orientations, we needed an independent measure of attitudes to parties. We used the standard propensity-to-vote (PTV) measure that is argued by van der Eijk and Niemöller (1983) to reflect electoral utility and, in any case, provides a useful scale measure of attitudes to the parties. The question is worded thus: ‘On a scale from 0 (very unlikely) to 10 (very likely), how likely is it that you would ever vote for each of these parties?’ This is available in both surveys. The measures of partisanship required to address Q3 differ across the surveys. The second uses the traditional Michigan ‘Generally speaking…’ version, while, in the first, we use the ‘alternative’ party identification question which has been shown to elicit much lower proportions of identifiers in the population – around half as opposed to over 80 percent with the Michigan wording (Sanders et al., 2002). The ‘alternative’ wording is: ‘Some people think of themselves as usually being a supporter of one political party rather than another. Do you usually think of yourself as being a supporter of one particular party or not?’ 7 Having first addressed Q2 and Q3 by exploring the bivariate relationships between orientation perceptions and likelihood of voting for the parties, we then test fuller models of PTV, using the second survey and the wide range of control variables that it offers, in order to test whether those bivariate relationships hold in a multivariate context.
Political marketing and party loyalties – the evidence
Q1. Do voters recognize the product/market distinction?
Before we examine how respondents perceived the parties’ orientations, it is first worth looking at how willing and able they were to supply such ratings. A non-response (‘don’t know’) option was explicitly offered throughout the questionnaire.
Given that survey respondents generally display considerable willingness to answer questions, even if highly unsure of their views, the average 20 percent non-response rate on these orientation questions looks high (and still higher in comparison with the mean 5 percent non-response to the PTV questions). This could indicate that many voters simply do not recognize the notions at the heart of political marketing analysis; alternatively, it could be that those notions are understood by voters but were not clearly conveyed to them by our question designs. At this stage it is very difficult to adjudicate between the two possibilities, but we return to this issue in the concluding section. 8
Now we turn to the ratings of the parties supplied by those who did answer the orientation questions. Figure 1 illustrates the mean scores for the three main parties on both orientation variables. At the bottom of each pair of columns we report two statistics: a dependent-samples t-statistic for the difference between that party’s mean product-orientation and market-orientation scores; and the correlation between the two sets of perceptions; that is, the extent to which respondents see the two orientations as connected. The columns are remarkably similar in height with one exception: the Conservatives are clearly seen as being particularly market-oriented. It is impossible to know whether this is an enduring or temporary result, though it tallies with recent political history and Cameron’s overt attempts to reposition the party in the centre ground. Although, at the time of our survey, most political commentators would probably have regarded Labour as still occupying that territory, too, the party was not seen by voters as particularly market-driven or unusually remote from its traditional product.
Mean perceptions of product/market orientation for the three main parties.
Strikingly, the correlations are all more or less zero. Voters see virtually no trade-off between the two orientations. Only with Labour, and then only to a tiny extent, do responses reflect the negative correlation between product and market orientation that the political marketing literature would lead us to expect. As these negligible correlations imply, substantial minorities of respondents regarded parties as simultaneously both market-oriented and product-oriented. Eight percent of our sample rated the Conservatives at 5 or higher on both scales, and over 18 percent gave them a 4 or higher on both. The corresponding proportions for Labour were 4 percent and 12 percent, for the Liberal Democrats 3 percent and 9 percent. Clearly, quite a few voters think that a party can both stick by its policies no matter what and be willing to change in order to meet demand. This suggests that voters disagree with Lees-Marshment’s interpretation of the relationship between the two orientations, and instead endorse Henneberg’s (2006) argument that parties can indeed both lead and follow. (Or, for that matter, they can do neither. Non-trivial proportions of respondents rated the Conservatives (8 percent), Labour (12 percent) and the Liberal Democrats (8 percent) as ‘not at all’ product-oriented or market-oriented.)
Predictably, respondents rating a party highly on both orientations tended to be identifiers with that party, and those rating a party at ‘not at all’ on either orientation tended (via the PTV measures) to have indicated a strong distaste for that party. These respondents presumably regarded each orientation as a positive characteristic, and were keen to credit a preferred party with both or to accuse a disliked party for lacking either. It might then be thought that partisan bias conceals the ‘real’ and expected negative correlation between the two orientations. This turns out not to be the case, however. When the analysis was repeated including only those respondents disclaiming any party identification, the product/market correlation for Labour was basically unchanged (–0.04) and for the Conservatives and the Liberal Democrats actually became more clearly positive (0.15 and 0.14, respectively). Even these more ‘objective’ judges of the parties see no necessary trade-off between product and market orientation.
Q2. Is a market orientation more popular?
Here, we consider how voters evaluate product and market orientations by examining the relationship between orientation perceptions and willingness to vote for the party in question. Our initial finding from Figure 1, namely that the Conservatives appear more market-oriented than the others, tends to support Lees-Marshment’s core argument. At the time of the first survey, carried out before the ‘credit crunch’ triggered a brief dip in the party’s popularity, the Conservatives enjoyed a substantial opinion poll lead. This is reflected in our data: that party’s mean PTV score (5.1) was markedly higher than those of Labour (3.8) and the Liberal Democrats (3.7). To use classic marketing lingo, one might conclude that the Conservatives’ ‘unique selling point’ was their market orientation, and therein lay their advantage over the other parties.
Of course, this is a speculative finding based only on aggregate relationships. We need to investigate the individual-level relationships between perceived orientations and readiness to vote for a party. The left-hand graph in Figure 2 shows mean PTV scores for each party among those rating that party at each level of product orientation. The right-hand graph does the same but for market orientation. (Relationships for all three parties are presented together for reasons of economy of presentation.)
Perceived product/market orientation and propensity to vote for all parties.
Perhaps the most obvious feature of both graphs is the evident curvilinearity of the relationships between orientations and popularity. For all three parties, only up to a point do increases in perceived product-orientation or market-orientation serve to increase respondents’ willingness to vote for that party. However, the turning-point varies both across orientations and across parties. The left-hand graph is more obviously skewed, with parties given the maximum rating for product orientation proving much more popular than those given the minimum rating. The right-hand graph is more symmetrical, with voters unimpressed by parties rated as ‘very market-oriented’, although again the lowest PTVs are found at the lower end of the orientation rating scale. In short, the answer to Q2 is ‘no’. Although neither orientation is popular beyond a certain point, that point arrives sooner with market orientation, and the shallower upward gradient in the right-hand graph also suggests that market orientation has a generally weaker association with willingness to vote for a party. That said, there are hints that the relative popularity of orientations differs across parties. For the Conservatives, product-orientation is clearly more popular than market-orientation: in terms of PTV for that party, the optimal rating was 5 on product but only around 3–4 on market orientation. For Labour, the benefits of product orientation run out sooner, while the market orientation graph does not noticeably turn downwards until the maximum rating. The Conservatives are rewarded more for product while Labour is rewarded more for market orientation.
The wealth of information in the graphs above risks obscuring the central point about the relative popularity of orientations. Next, then, we approach that question from a different angle, calculating between-party differentials on each orientation for each of the three pairs of major parties. These differentials run from –6 to +6 (one party seen as ‘very’, the other seen as ‘not at all’, product-oriented or market-oriented), with 0 indicating that the parties were seen as equally oriented in that way. We then plotted PTV scores for each of the parties in that pair against these orientation differentials. The resulting graphs illustrate how the relative popularity of parties covaries with perceptions of their respective orientations. Put another way, they allow us to gauge the extent to which being judged as more product-oriented or market-oriented than one of its rivals gives a party a potential electoral advantage. In Figure 3 we present graphs for each pair of parties: Conservative versus Labour (Figure 3.1), Conservative versus Liberal Democrat (Figure 3.2) and Labour versus Liberal Democrat (Figure 3.3). In each case, the left-hand panel plots PTV scores against the product orientation differential while the market orientation graphs are in the right-hand panel.
PTV scores for pairs of parties by differentials in their perceived orientations.
There is a striking difference between left and right panels: the beneficial effect of being perceived as more product-oriented is consistently and markedly stronger than the corresponding effect of market orientation. For each pair of parties, the party that is seen as the more product-oriented is always more popular and its popularity increases as its advantage on this orientation widens. In contrast, being seen as more market-oriented than a rival does not help a party to anything like the same extent. At least between Labour and the Conservatives we see strong effects at the extreme ends of the scale: that is, those who see one of these parties as very market-oriented and the other as not market-oriented at all will prefer the former. But smaller perceived differences do not translate into popularity gains. And, from the two pairings involving the Liberal Democrats, an even more ambiguous picture emerges. That party in particular seems not to gain at all from being perceived as more market-oriented. Aside from a peak at –5 in the two right-hand graphs (i.e. those who see the Liberal Democrats as leading the other two parties by 5 points on the market-orientation scale), the Liberal Democrat PTV scores do not relate in any systematic fashion to how much more or less market-oriented they appear to respondents. On this evidence, the answer to Q2 remains a quite firm ‘no’. A perceived advantage on product orientation is exactly that – an advantage. By contrast, parties gain much less from being seen as more market-oriented, whether this is in absolute terms or relative to their rivals.
Q3. Is preference among orientations moderated by partisanship?
The results so far sit uneasily with some of the core assumptions underlying Lees-Marshment’s political marketing model. In particular, they suggest that a publicly perceived move from product to market orientation can incur a cost. However, it could be that the overall preference for product orientation is driven by party identifiers (and their greater attachment to parties’ products). In that case, Lees-Marshment could nevertheless be correct that a publicly recognizable marketing orientation is the key to winning an election. Even if not preferred by the electorate as a whole, it may still prove more popular among the non-partisan voters who swing election outcomes. To test this, we examine the relationship between orientation ratings and PTV scores among two groups: those who identify with the party in question, and those who identify with none of the parties. This relationship is the same as that explored in Figure 2, but to avoid a surfeit of graphs we condense the information into a table of correlation coefficients (Table 1). To account for the curvilinearity of the relationship between orientation perception and likelihood to vote for a party, we correlated PTV scores both with orientation perceptions (in the upper panel of the table) but also with squared orientation perceptions (lower panel). For these squared terms, the orientation scales were centred around zero, which means that high values indicate extreme (high or low) market or product orientation and low values more middling scores.
Correlations between orientation ratings (main and quadratic terms) and PTV for a party: (a) all voters; (b) identifiers with that party; (c) all non-identifiers.
NB. All correlations other than those in parentheses are significant at the p<0.05 level.
As a point of comparison for our subgroups, the first two columns show correlations for the entire set of respondents. Since the full sample was the basis for Figure 2, this first set of results can be used as a guide to how the correlations summarize the patterns illustrated in those graphs. Hence, the positive correlations in the upper panel reflect the broadly upward trend in the graphs, the stronger coefficients for product orientation reflect the fact that the product graphs had a more pronounced upward slope, and the negative correlations with the squared terms in the lower panel reflect the downturn in the graphs at higher levels of each orientation – that is, the negative reaction to parties perceived as too market or product-oriented.
From the remaining columns in Table 1, we can see that this public demand for moderation on both counts is strongest among non-identifers. It is not the case that floating voters reward market rather than product orientation – there is some sign of that with the Labour Party, yet if anything the reverse is true with the Conservatives. Rather, they are particularly inclined to punish parties showing an undue preoccupation with either product or market. Among party identifiers, there is the broad tendency – in line with expectations – for product orientation to win more favour than market orientation, but the differences are slight and vary across party. Only with Liberal Democrat supporters is there a markedly stronger preference for product than for market orientation. The same is true but much less obviously so for Labour identifiers; they also show a greater inclination to punish ‘excess’ market than product orientation. Meanwhile, there are only weak correlations for Conservative identifiers (partly because their mean PTV score for the party was so close to the scale’s maximum point that there was limited statistical traction for gauging its covariation with the orientation scores). The more conspicuous feature of the results among party identifiers is that they are much less concerned about moderation. Correlations between PTV scores and squared orientation perceptions only just reach statistical significance for Labour and Conservatives and fail to do so for the Liberal Democrats, and for all three parties the estimates are substantially below those we see for non-identifiers. In summary, the data in Table 1 provide some grounds for answering ‘yes’ to Q3, and thus argue against the tendency in Lees-Marshment’s model to homogenize the electorate. Yet the moderating effect of partisanship is neither clear-cut nor in quite the direction that we forecast. In particular, the simple ‘partisans prefer product; non-partisans prefer market’ model receives little support.
Multivariate models
In the discussions above, we were prone to draw causal inferences from the correlational data. For example, if perceptions of a particular orientation were positively correlated with propensity to vote for a party, we referred to voters ‘rewarding’ parties for that orientation. The implied causal link from orientations to party preference is, of course, the crux of Lees-Marshment’s model. Yet this causality is inferred rather than demonstrated. Moreover, there is reason to suppose that the causality is spurious: that is, orientation perceptions are driven by prior attitudes to the parties that are also the real drivers of PTV scores. (We noted this possibility earlier when considering those who scored a party very low or very high on both product and market scales.) This kind of endogeneity problem confronts any researcher seeking to estimate the impact of party evaluations on party choice (see, e.g., Wlezien et al., 1997). However, it can be mitigated considerably by holding constant as many as possible of those prior variables that might influence both orientation perceptions and propensity to vote for a party. We used a crude method of control in the previous section when analysing partisans and non-partisans separately. A more sophisticated approach is to test multivariate models of PTV for the three parties, in each case including various controls in addition to the orientation perception variables. The second survey, fielded within a BES questionnaire, offers a comprehensive set of such controls: socio-demographic variables, newspaper readership, party identification, party leader attitudes, economic expectations and reactions to the financial crisis, perceived party competence on valence issues, and respondents’ stances on position issues. We can therefore run very well-specified models of likelihood to vote for the three parties, closely replicating those shown by the BES researchers to account for much of the variation in party choice in Britain (Clarke et al., 2004, 2009).
The full list of controls used can be read in Table 2, in which we present the three regression models. The lower rows contain the orientation perception effects that are our main concern here. Note that, since we have switched to the second survey, we no longer have separate variables for product and market orientation; instead, these are based on the bipolar scale running from product to market. For each party, we report coefficients for two orientation variables: the basic ratings and the squared ratings. 9
Coefficients for orientation perceptions from regressions of PTV for each party.
These results show that the patterns identified earlier in this article are robust against multivariate controls. Most notably, all three parties are being rewarded for striking a balance between product and market orientation. This is indicated by the strong and significant negative effects of the quadratic terms. Parties are penalized if perceived as located close to either end of the scale. In the multivariate context, we find less consistent evidence of a preference for product over market. On the one hand, there is a significant main effect for the Liberal Democrats in that direction, with those perceiving the party as more market-oriented being less likely to vote for them. On the other hand, we find that respondents who perceive Labour as more product-oriented are slightly (but not significantly) less likely to vote for the party and also (and significantly) more likely to vote Conservative. This recalls our previous finding that Labour is substantially less likely to benefit from product orientation than the other two parties. It also highlights that probability to vote for a party can also depend on how you perceive its rival(s) to be oriented. The Conservative Party proved more popular among voters approving of its product-market balance but also among those feeling that Labour had strayed too far towards the product end of the scale.
These significant effects confirm that voters’ perceptions of party orientation matter. They are clearly significant even when controlling for a host of predictors of party choice (and in models in which – at least for the Conservatives and Labour – a large proportion of the variation in PTV is explained). They are also powerful – often more so than the effects of core valence judgements (such as a party’s performance on issues such as the NHS and crime). Aside from party identification, the squared orientation effects are the most powerful in the Liberal Democrat model, the second most powerful in the Labour model, and far from trivial in the Conservative model.
In order to address Q3, we added interaction terms between the orientation variables and party identification. Rather than repeating the full models, in Table 3 we report only the orientation effects along with these new interaction terms. The dearth of significant effects among the interaction terms argues for a ‘no’ in response to Q3. In none of the cases is there the expected significant and negative interaction between party identification and the main orientation variable, which would have indicated that identifiers placed particular value on that party’s product. And only for Labour identifiers is there a significant interaction with the squared orientation. This, at least, is in the expected direction: the positive sign denotes that party identifiers are less likely to demand that their party strikes a balance between product and market orientation.
Main, squared and interaction (with party ID) effects of orientation perceptions on PTV.
NB. In order to avoid collinearity problems, we do not estimate all the interaction terms for all the parties in the same model. The interactions with Conservative identification are therefore included when modelling the probability to vote Conservative, and so on for the other two parties.
We conclude the paper by illustrating the net effects of orientation perceptions on party preferences. The graphs in Figure 4 are based on predicted values of PTV from the full regression models. In each case, we calculate the mean predicted value among those rating that party at each point of the orientation scale. Three features of the graphs are worth noting and serve as a useful summary of our key findings. First, all parties are clearly penalised when perceived as strongly oriented in either direction. Second, even with extensive controls, the effects of orientation perceptions are striking. The slopes become quite steep, especially towards the extreme points (a point that is of substantive significance given that non-trivial percentages of respondents located parties at those extremes). Third, there are marked differences across the parties. For Labour, the graph is not obviously skewed in either direction. However, for the Conservatives and Liberal Democrats, the peak of the graph is clearly to the left of the midpoint, signalling the relative popularity of product orientation.
Mean predicted PTV values for each party by its perceived orientation (calculated from regression models in Table 2).
Conclusions
The ideas at the heart of Lees-Marshment’s political marketing model – that parties face a trade-off between sticking by their traditional stances and modifying these to please a greater number of voters, and that a party favouring the latter approach will perform better in elections – are not abstruse academic notions. They dominate popular discourse about electoral politics. It seems unlikely then that, once explained in more everyday terms, the product/market distinction will have proved too unfamiliar or difficult for survey respondents. Yet the straightforward predictions generated from the Lees-Marshment model were confounded when tested on individual-level data, both in bivariate and multivariate analyses. Voters did not recognize the trade-off between the two orientations, and they did not on the whole prefer market orientation. Even non-partisan voters, the main targets of market orientation, tended on balance to prefer product orientation.
One possible explanation for this is that our survey measures were confusing or misleading. The hefty proportions of non-response on the orientation questions point in that direction. However, those who did respond do not appear simply to have supplied ‘non-attitudes’, answers delivered more or less at random in order to fulfil their obligations as respondents. It is not that the responses lacked structure, but rather that their structure differs from that which Lees-Marshment’s model leads us to expect. Moreover, the fact that respondents identified the Conservatives as the most market-oriented party suggests that – as in the case of ‘New’ Labour and many other important examples for marketing scholars – significant shifts towards a market orientation do not go unnoticed. The question is whether those shifts in themselves prove popular with voters. Judging by our data, it is far from clear that they do. Rather, voters – more or less across the board – want a bit of both orientations. They endorse a party that will stick to its core principles without being intransigent, and a party that is responsive to public opinion but again not indiscriminately so. Most importantly, they see these demands as compatible. A party can both stand by what it believes in and shift in response to public opinion, according to many voters.
This will not come as a surprise to some political marketing scholars (e.g. Henneberg, 2006; Henneberg and O’Shaughnessy, 2009), or those researching marketing more generally (Connor 1999), who have argued in detail that there is no such trade-off between leading and following. Follow-up work to this study could put their rival theoretical arguments to parallel testing of voters’ reactions. Meanwhile, once the assumption of an automatic trade-off between product and market orientation is relaxed, Lees-Marshment’s claim that a marketing orientation is necessary in order to win elections starts to look more accurate. It is a matter of degree, however. Judging by our data, a party will not be popular with voters who see it as wholly inflexible in the face of public opinion. Hence parties probably do need to be somewhat market-oriented in order to win elections. But they risk unpopularity if they seem too ready to bend to public demand, and they certainly will not be rewarded for appearing heedless of their core principles and policies. (The maxim for parties should be ‘everything in moderation, including moderation.’) On this reading, electoral success requires parties to compromise with the median voter rather than to stalk her. Again, this finding is strikingly consistent with the arguments of Henneberg (2009) about the need for parties both to lead and to follow. One task for political marketing scholars – a task in which this kind of survey data can and should play a role – is to identify how parties can convince voters that they are doing both.
Footnotes
Funding
We are grateful to Research & Innovation at the University of Strathclyde for funding the first survey reported here and to the British Election Study team for fielding the second survey via their NSF-funded Continuous Monitoring Survey.
Acknowledgements
Thanks to Joe Twyman and Kate Davies at YouGov for their help and support during the data collection process. Previous versions of this research were presented at a Greater Glasgow Politics Group seminar (February 2009), at the inaugural conference of the University of Strathclyde’s Centre for Elections and Representation Studies (March 2009) and at a Political Marketing Group panel at the annual meeting of the Political Studies Association in Edinburgh (March 2010). We thank participants at all three events for their helpful comments and suggestions. We also thank anonymous reviewers of this and earlier drafts of the paper.