Dealignment, realignment, non-alignment? To what extent can these concepts account for changes in electoral support for parties?

In a time of alleged dealignment, realignment and non-alignment there is perhaps a strange observation to be made. In the UK the Labour and Conservative parties have been in perpetual power at national level since WW2, as they have almost been at local level. In Germany, the post-war CDU-CSU and SPD parties have continually been in power at national level, as well as being in government in every Länder for half a century. In the United States, since the 1860's the Republicans and Democrats have not only held power in almost all federal and state institutions (excluding the rare occasions such as the success of Jesse Ventura in the 1998 governor elections), but they have seemingly also had the monopoly of representatives in government bodies; in Congress, there is a single 'independent'. These facts themselves beg the tongue-in-cheek question, 'if that's dealignment, what does an aligned system look like?'. For part of this study, I shall therefore try and see if such an observation is fair. This would seem to deal with the concepts of dealignment and non-alignment. The next area to look at is that of realignment. Realignment would incorporate the observations of party continuity, but suggest that different voters are supporting the parties, giving constant results, but for changing reasons. A more interesting angle to approach this is to suggest why such trends seem to occur, but further, whether the assumptions that govern how we view the voters' relations to parties are themselves fair. This must be where this study must begin.

There is a danger when theorising about such issues to think of voters as sheep, dutifully making their cross on the ballot. There seems a dryness about discussion of voting behaviour, that when a trend seems to be identified, such as working class support for socialist parties, that there is an entrenched rule that political scientists have discovered, but which controls the votes of the people. A working class person does not vote for the leftist party because he wakes up, realises that he is working class and so, following the 'rules', votes left. The reason for such an identification is that it is thought that by voting for the leftist party, as a working class person, their interests will be better served and represented. Perhaps this point need not be made, but there is a tone in much of the literature to suggest that party identification is akin to support for a sports team. This, I feel (I hope) is false. This does not mean that such forces as peer pressure and family traditions will not be identified as being causes, but this must be seen to be because of perceived shared interests between peers or within families. Changes in people's interests would be expected to lead to changes in voting behaviour.

Following from this another point should be made. A political party is not independent of a changing world, it does not exist in a vacuum. A party is made up of and funded by individuals who face all the changes in society. The role of parties in competitive democracies is to govern or to represent a viable alternative to the governing party. There is therefore a need for cleavages to distinguish between parties. These cleavages must reflect cleavages in society that are enough to gain the support to win power. This must therefore mean that when a broad consensus arises where once an old cleavage stood, the parties must find new cleavages. Assuming that individuals vote for their perceived self-interests, if party cleavages change then the people that would be expected to vote for them previously would also be expected to re-evaluate their support, according to where they stand on the new cleavage. To clarify this point, statistical proof of deviation from traditional alignment could be seen as the result of changes in party as well as changes in the electorate.

From this, it could sensibly be asked as to why instead of the existing parties adapting, new parties do not come in their place. The answer of this is to a large extent institutional and specific to the different party systems. Excluding these institutional reasons, another point is clear. This is that although there can be seen to be changes in interests throughout a country, these changes are not going to be uniform throughout. Added to this, there are multiple cleavages within party systems, so whilst a party may lose support of a section supporting one cleavage, they would regain support over other cleavages. In order to gain meaningful representation a new party would have to exploit a cleavage large enough to displace these multiple cleavages. In a large country this is extremely hard. The result of this is therefore evolution of parties rather than their displacement.

This point therefore address the question asked in the introduction. It does seem fair to suggest that even in times of party system stability, there can theoretically be a loosening of alignment between voters and parties. I stress the word theoretically, because I believe that there has largely been stability of alignment in all three of these countries.

Before going further we must ask exactly what alignment is. The best explanation of alignment theory comes from the Michigan model put forward by Campbell, Converse, Miller and Stokes in the 'American Voter' of 1960. This model shows that alignment between voter and party can be seen both in its direction (who is voted for), and its strength. The conclusion of the study suggested that whilst there was fluctuation in voting behaviour, voters did align themselves to a particular party which they would 'come back to' in the long run. They also noted that partisan alignment was seen as contributing to a voter's personal identity and was durable through passing issues, events or personalities. This point of personal identity does suggest a degree of what I shall call 'sports team alignment', but I wish to propose that this still is due to identifying themselves as having shared interests with a party. Indeed, the study did identify key cleavages which divided the electorate to account for such alignments.

If, however, alignment was complete then there would seemingly be no variation in election results. V. O. Key suggests that there are three types of election. The first of these is the maintaining election in which alignment is adhered to. The second type is a deviating election, where a voter will go against a traditional alignment due to a particular issue, event or personality. After such an election, they might be expected to return to their aligned party. The last election Key identifies is the realigning election. Here there is a long-term change in partisan loyalties which sets up a new set of voter-party alignments. As was noted above, such deviations or realignment does not necessarily extend to all those who could be previously seen as aligned to a party, but instead could be a sector of supporters, such as regional, class or race.

To work backwards from here, we must now ask as to how voters gain these alignments. This was suggested above as being related to how voters perceive their interests tallying with the interest of a political party. This does go further. Once a voter aligns themselves with a party over the cleavage that they see as central, they are seen to also support the party-line on issues which are of less personal importance to them. There is an assumption of commonality between interests and so in areas where the voter is less-informed it seems justifiable to believe that the party with which they share interests on key issues will be capable of making decisions in other areas that the voter would support. This is part of identifying with a party. Such electoral choices are inevitable for a number of reasons. The first of these is lack on information for a voter, coupled with an apathy towards making time to research issues which are not of particular concern. The second important point is that a vote will go for an individual standing for a party, not an issue or set of interests. Candidates stand for a vast array of issues and it is impossible for a voter to select those issues or interests that they support from those they do not from a particular platform. For this reason an individual is forced to make a choice of who they generally support to carry through their interests. The individual's vote is necessarily a compromise, yet the result is a full mandate for the whole of the victor's program. By therefore voting for a party, rather than a policy, there is seemingly an obligation to support the actions of that party as their legitimacy to govern was based on your own choice. There is some room for descent in partisan alignment, yet there does seem a distinction between descent from within and support for another party. As was suggested by Key, even if this descent does result in voting for another party, the identification is often strong enough to make such trends deviations rather than an abandonment of party alignments.

There is still the question as to how a voter perceives their interests. A traditional explanation of this relates to upbringing. It is clear that voting behaviour is statistically as hereditary as hair colour. This, again, is not because of a sheep-like following of parents, but the political education that often comes from the home. The view that people often have of the various political alternatives comes from this education. The stronger the party identification of the parents, the more likely the children will vote for the same party. This must be seen as two stranded, firstly that parental education is important in forming opinions, ideologies and priorities and secondly because statistically children are likely to grow into the same social, religious and ethnic groupings as their parents. This means that those issues that affected the parents will be central to the lives of the child. This argument can be extrapolated further into social groupings outside the family which have both educational effects and shared interests. When individuals in such groupings make decisions and become aligned to political parties and these trends are multiplied to show that not only are individuals aligned to parties, but so are sectors of the population.

The important thing to note when looking at such relationships is what sparks off deviations and realignments. It might be expected that changes in social factors, such as the advent of the 'post-industrial age' would lead to realignment, whilst factors such as period specific issues, political personalities and political 'events' would lead to electoral deviation, expressed in sectoral as well as individual levels. There is a consensus that in all three of these countries, there has been increasing volatility and so possible shifting of alignment, so this volatility must be translated as showing either realignment or periods of dealignment. The last possibility suggested is that of non-alignment, in which we find ourselves in a world where everyone will make a choice at each election based purely on the issues presented to them. The problem with non-alignment theory is that it forgets that political parties are to a large extent dependent on the voters for their internal structures. Although there might be some areas of the electorate that are not aligned, if such action were to occur throughout a population, parties would lose resources and be ineffective. Although parties are increasingly seen in the media age as being less dependent on grass-roots support, it has to be acknowledged that they are still vital to success at a national level. Parties are not purely the centralised machines that fight national elections, but they also govern schools, local councils and act as welfare organisations in communities. Such action would not be possible in a non-aligned system.

It is now necessary to make some generalisations of the form that alignment would be expected to follow in modern states. Lipset and Rokkan identified four major cleavages which seem to act to form partisan alignments. These are, centre-periphery, church-state, urban-rural and worker-employer. In states where more than one of these cleavages are evident, the political parties would base around a broad church model. This would mean that people will align with the same party for reasons of different cleavages.

Another important general point to make is that elections are often won and lost by the votes of a small proportion of the electorate. This effect is exaggerated in more complicated voting systems, but even if there seems to be great volatility in overall election results, this does not mean that there has been a decline in alignment across the electorate. Campaigns are a combination of two strategies. The first is to persuade the 'floating voters' to support your party, the second is to make sure that your supports turn-out to vote. Turn out is an incredibly important factor, with in all three countries enough votes not being cast to change the result of the election. With regard to 'floating voters' we must see if there is a genuine rise in the amount of people who conform to this category. The view which I wish to put forward is that although there is an increasing number of 'floating' or non-aligned voters in the post WW2 period, this accounts for only a small proportion of the electorate. However, because of the balance between party's support, this small number of votes are often decisive at election time. It was suggested in the British 1997 general election that in an electorate of over forty million, the parties were really competing for the votes of seventy thousand key electors in order to win. The point is therefore that although deviation from alignment is important, to give undue weight as a measure of overall political activity in a state is unfair.

The case of the UK.

The UK is a useful starting point for this analysis because of its relatively small size and strong party system. There are also some interesting features, such as the goings on in Northern Ireland and to a lesser extent Scotland and Wales which highlight some important points in alignment theory.

One of the key premises to this study can come from the work of Butler and Stokes who in 1974 suggested that politics in the UK is largely peripheral to most people's lives. As such they do not seek to inform themselves of the major issues and so must gain their voting guidance from others. They again highlight the socialisation of voting by virtue of family 'loyalties' as well as through the weakening feature of class alignment. It is clear when looking at voting alignment in the UK, class does always seem central. As shall be seen later, even when looking at other factors, such as regionalisation or ethnicity much of the results can be related back to the class element. In a post-industrial age, with a changing workforce and breaking down of class structures, it might be expected that if class is the central feature of voter-party alignment, with the changing of class structures, there would be expected to be a change in alignment structures. This points to the possibilities of both realignment as well as dealignment/ non-alignment. Many, including Ivor Crewe have suggested this, but there is still some striking statistical information to show that voting is not a random process between regions, classes, genders or ethnic groups which points to some form of alignment. My view on the British system is that although there could be a small amount of non-alignment and dealignment, there is still a large degree of class alignment. To illustrate this point, Figures 1 and 2 (Page 6) show distribution of electoral successes in the last two general elections.

Throughout the following discussion of how there is continual dealignment of voting behaviour, it is important to constantly refer to these maps. The patterns that they show of variations between voting, particularly between North and South and urban and rural areas is strong. The two elections also show two interesting cases. Figure 1, the 1992 election shows a time where party balance between the Conservatives and Labour is relatively even, perhaps showing an approximation of a status quo of one of Key's maintaining elections. Figure 2, the 1997 election is most telling when looking at the Conservative party. The areas that remain blue in the face of an electoral landslide must highlight the high points of alignment towards the Conservative party. Looking at what kind of voters these are will show what the basis for such alignment is. Before looking at these in detail, it is necessary to discuss the literature about voting behaviour in the UK.

As was earlier hinted of, there has been a general trend towards weakening of identification with political parties. This has been brought out in the British Election Study (BES) surveys started in 1963. Findings have shown that in 1964, 92% of the electorate had a 'general sense of party identification', compared to 86% in 1992. Those with a 'strong sense of party identification' has dropped more strongly from 43.3% (1964) to 22.8% (1992). An important point to note about this is that the change has not been a linear one throughout this period. Instead, there was a major drop in the 1970s which stabilised to the current levels. Such a trend was identified across Europe by Herman Schmitt and Soren Holmberg. Their analysis of the non-linear movement went against the speculations of people such as David Denver. Denver suggested that increased education of the electorate particularly through non-partisan television and other social factors accounted for such changes. In response, Schmitt and Holmberg suggested that the blips that occurred in electoral behaviour were due to political factors. Such political factors include the emergence of new parties, content of ideological conflict, changes in political leadership and crises. This view does seem to have some statistical backing in the UK, although with some of these observations, such as emergence of new parties, there are some 'chicken and egg' questions being begged.

Further evidence of this is seen in political party membership in the UK which has declined from 9.4% in the 1960's to 3.3% in the late 1980's. Although there is some evidence that this decline is reversing in modern times (particularly due to recruitment drives by Tony Blair and now William Hague), this trend does seem to follow the results of the party identification surveys. Again, the drops are non-linear and coincide with political occurrences, particularly in the 1970's. One of the strongest explanations of these results was the rise of the single issue activist groups which allowed people to politically participate in more focused ways. It is clear also that simply because someone stops being a member of a party, this does not mean that their voting behaviour will change.

Based on the simple 'Alford' index, there has been a declining alignment of class voting in the UK. I call this a 'simple' index because it is based on the class distinction being between manual and non-manual workers. According to this criteria, the working class population of the UK has dropped from over 75% in 1911 to around 65% in 1951 to about 50% in 1991 (according to census data). Many now disagree with such criteria for class and based on the work of the sociologist John Goldthorpe, a new classification of class can be seen. This 'revisionist' approach, highlighted by Heath, Jowell and Curtice suggests that by using Goldthorpe's classification, there is seen to be no change in class alignment. A shrinking working class and dynamics of society can still account for electoral volatility, but if the correct classification is used, class-party alignment is shown to be of merit. Blips that can be seen in this relationship are related greatly to political factors. Notably, the times when there was seen to be a decrease in this relationship (according to Heath et al) was when the Labour party was in government. This seems to confirm to a deviating, rather than realigning views of such voting patterns. There is also a suggestion that class is a greater aligning factor with regards to the British Labour party than the Conservatives, and that when a relationship is seen to relate Conservatives to class it is by default rather than being that which is central to Tory voters.

In order to make this more precise, investigations into specifically what aspects associated with a class lead to partisanship were made. One such strategy was to look at the voting behaviour in relation to home ownership. Those that rented their homes from the council were seen to be consistently in support of the Labour party between the 1960's and the 1990's, with consistently reverse results for owner-occupiers. This is regardless of the ten percent drop in amount of council housing in the 1980's alone. This shows strongly that people will vote for their perceived interests. When this is multiplied into a sector (such as council tenants), it is shown as sectoral alignment with parties. Consistency of party's stances with regards to key interests is highlighted by alignment over time. The problem is managing to spot the interests which cause alignment. For example, it might be expected that there would be seen to be alignment between those working in the public sector and the Labour party. Evidence from the BES survey shows that in fact whilst there is a bias towards the Labour party, the changes in voting behaviour in this sector (unlike council tenants) is in parallel to the general population.

The next important difference is that of gender. It has been suggested that all things being equal, had the franchise not been opened to women, the Conservatives would have been out of power continually between 1945 and 1979 (ironically, under Thatcher this trend became insignificant). Explanations for this suggest perceptions of the Tories as being the party of more conservative family values. This seems to tally with attitudes of women of the centrality of the family to their concerns. Further investigation shows that this relationship is quite complex. In reality, younger women are more likely to vote Labour than are the same age males, and older women are more likely to vote Conservative. This has led to what has been described as a 'gender-generation' link. The reason for this is largely thought to be that of motherhood and was seen to take root again after the "no such thing as society" Thatcher years, with the link becoming again evident in the early nineties. Although the difference between men and women is not statistically huge, because women increasingly out number men and are more likely to vote, it is of electoral importance.

Much the same importance can also be given to evidence of alignment of age to electoral parties. The older a Briton is, the more likely they are to vote Tory. This can be shown again in perceived interests, with the very name 'Conservative' suggesting that which is not radical. Older people, more reliant on stable jobs and pensions don't like radicalism . Some revisionists of this theory suggest that it is not the age of a person, but instead the vote that they cast when they first entered political life. There is again statistical evidence for this, suggesting that the reason that the Labour party is not greatly supported by the elderly can be traced back to their lack of support when the elderly were not elderly. It must be accepted that the best basis for predicting future votes is to look at an individual's voting record, the very essence of alignment theory. This means that in age terms, there is an independent cleavage in some cases, but in others it can be attributed to other factors.

The problem with age and gender is that unlike the class and council-tenant analyses, it doesn't explain the striking evidence of the two maps above. As was suggested above, the 1992 election does seem to reflect the political status-quo of post-war British politics, with the Tories more likely to have a small majority. As is seen, the distribution of seats in this case is far from even across the country. There is a clear North-South divide shown in both maps, but more clearly in Figure 1. Another important trend is shown in the cities. These are both shown in the insets as well as the islands of red in the sea of blue in Southern England. Many have interpreted this as some form of regional alignment, but I believe that much of these patterns can be related to explanations already discussed.

Not only is there a partisan divide between the North and South of the UK, but there is also an economic divide. The North has a greater industrial emphasis and is, to a large extent, less wealthy, with a population more dependent on the state. To a large extent, this is not the area where greatest social changes are occurring and it has been suggested that where the working class is developing into a 'new' working class in the service industries of the South, comparatively, the working class of the North is more traditional. This means that the traditional bonds to political parties remains constant as the parties remain in the same position on the ideological spectrum. This can also be seen in the inner-city areas where there is similar class dynamics compared to more affluent suburban-rural areas. Tagged along to this explanation can be the alignment of ethnic minorities to the Labour party. This relationship is extremely strong and although has been accredited to the more liberal stance of the left-leaning party, it is more likely to be due to class. This is highlight by the decline of this link in the Asian community as it becomes more wealthy.

Another explanation for the regional trends that has been given is based on alignment around Lipset and Rokkan's 'centre-periphery' cleavage. The perception is of the Conservatives as the party of centralised government in London, uninterested in the UK above Nottingham. This, however, does tally with class centred models, which see the common economic interests of the South against those of the North. Further, the class model better explains Labour's standing in the cities.

To move this further is to look at the election map of 1997. It is important here to recognise the importance of using a schematic map as on a more traditional map of England, even with almost two-thirds of constituencies Labour, the Conservatives look stronger because of the size of rural constituencies. This map shows what must be the lowest possible result for the Conservatives in a deviating election. The Conservatives are wiped out of Wales, Scotland and the major cities completely. The blue on the urban inserts represent extremely affluent city areas, such as Alan Clark's constituency of Kensington and Chelsea as well as suburban areas such as Cheadle in Greater Manchester. The strongest Tory alignment can be seen with an English, rural or affluent electorate. Again, this does seem to be a deviating rather than realigning election.

What does seem conclusive regarding these maps is that there is certainly some form of strong alignment of the British electorate. A further claim can be made that this alignment is of stronger force within the Labour party. In 1992 (the more typical election) of the nineteen seats with a majority over 50%, sixteen were held by Labour and only one by the Conservatives.

Another strong indicator of strong alignment is that of net volatility. This can be measured by both the Butler swing and the Pederson index. Both of these show that there has been no progressive move towards greater changing of voting behaviour in the UK. Instead, periods of high volatility are related to events and personalities. Stability of voting for an individual did follow a trend, although not persistent. The 1960s and 1990s both show relative stability in individual voting patterns, compared to volatility in the 1970's. The fact that this blip corrected itself points away from factors such as those suggested by Denver for variations on alignment models in the modern world. The rise of the third party in the 1970's for example could be one more credible explanation.

Before moving on, it is important to note one case in the UK where alignment theory is unquestioned. This relates to Northern Ireland, where votes are clearly cast according to religion, although there is some variation between radicalism and conservatism within each group. The clear differentiation between parties around this cleavage is not so clear in mainland politics where both parties act towards a catch-all model and the differentiation is not so clear.

Regarding the question as to why new parties have not formed in order to cope with changes in society, there are a number of reasons. The first follows from the observations of Heath et al above, namely, that there has been no dealignment of class voting in the UK and so no need for new parties to realign to. The second relates to the broad-church nature of British politics allowing for enough cleavages within each party to allow for some dealignment of supporters without damaging their prospects seriously.

The Case of the United States

The variation of the stances of political parties across the United States means that an analysis close to the above on a national scale would be fruitless. The US party system has been described being one of a hundred parties under two names. This does not, however, mean that people do not align, but instead that sectoral alignments such as have been detailed above cannot be as precisely made. As was seen in the UK case, there are some characteristics of electoral behaviour in the US that seem impossible in an unaligned system. For example, before Gingrich's Contract with America in 1994, the Democrats had held Congressional majorities for over forty years. Similarly, there had been GOP domination of the presidency in the post-war period. There is also some striking regional variations in voting patterns, such as Democratic dominance in the South and cities of the North East.

Literature still points to continuation of the views highlighted in the Michigan model of Campbell et al in 1960. The view here was that since the 1860's there was regional alignment along the cleavages of the Civil War. This lasted until the New Deal realignment of the 1930's which moved towards a more class centred cleavage, as well as agricultural versus industrial. Through this time, there was still some traces of the Civil War alignment, especially in the South. Again Key's analysis was useful in explaining volatility, suggesting that unless there was a clear sense of realignment, voters would return to the party that they best identified with. There is now a third major realignment that has been identified, that relating to the race movement of the 1960's. By virtue of numbers, if nothing else, race is a far more divisive issue in the US (ethnic minorities account for only about 5% in the UK). After the 1960's, the Democratic party was seen as the anti-segregation, pro-black party. This upset the whites in the South, breaking down further the Civil War alignment and to an extent the New Deal alignment. This does not, however, account for some other important trends in US politics.

The first of these is turn-out. Interestingly, although for any one election, there is unlikely to be a turnout much more than 50%, this does not mean that half of the population do not vote. Instead, fluctuations in voting can be seen not in switching party, but instead in abstention. According to Flanigan and Zingale, over 65% of people do vote at some point. This means that partisan alignment could still be in place, it is simply that there is often not the will to participate, but equally, there is not the will for dealignment. This must be noted as an important point.

The second interesting characteristic of the US system is that of split-ticket voting. Because multiple elections occur at the same time (often from the President, all the way down to the local coroner), there is scope for voters to vote both for and against a particular party, and they often do. This indeed allows for the same electorate to give a Democratic House to a GOP President, the most common arrangement in this half of the century. Further, it has been noted that split-ticketing is becoming more common. In 1980 almost 60% of voters said that they split their vote, compared to 29% in 1952. There are various reason for split-party voting. The importance of the incumbent in US elections cannot be dismissed. In the 1998 House elections, incumbency returns were over 98%, with a typical figure being around 94% for the century. This shows some strong person support for candidates who many voters see as detached from the national party machine. Separation of powers means that it is hard to accredit success or failures to particular branches, so in times of general stability incumbents are greatly supported. This leads to another reason for split-voting. It could be perceived that separated government works by separating government. Voters might wish to split the branches between the parties to force compromise and to act as a constitutional check.

The main reason for split-ticketing is perhaps that which was hinted at towards the start of this section, that the US can be seen as a country of a hundred parties under two names. Representatives will be elected on more local issues or for their personality, rather than on national issues, as the President would be. Indeed, although candidates do stand for parties, unlike in the UK, the election is centred on the individual rather than a specific party platform. It would be hard to develop a party platform that would unite Mississippi and New York Democrats. Even analysis of Gingrich's success (seemingly the creation of a party platform with his 'Contract with America') showed that these were not the issues that gained success, but instead it was a protest against single-party rule. For this reason, the choice seems multi-dimensional. It does appear that individuals, especially incumbents, are central to the voting decision, but the only way that the individual can gain prominence and gain their first victory is through relying on the party alignments. Incumbency voting might just be showing continually voting for the same party, but when incumbent returns are compared to non-incumbent returns, the later are seen to be far more open races. This must point to this effect of the individual candidate.

Except for in the black community, there has been increased overall volatility in voting behaviour in the US. This has a few reasons, but has thought to be able to occur without removing the various alignment models that have been proposed. The first of these reasons has been seen to be that suggested by Denver of an increasingly educated electorate, through increased media penetration and other social causes. Another explanation is that of greater social fluidity created by the post-industrial age, allowing class alignment for individuals to change. A more interesting point is the lack of major cleavages in modern US society. As the younger generation grows to be further from the Great Depression, such alignment becomes less relevant. Also, if race is not a large issue in an individual's life, the realignment of the 1960's would not have had much of an effect. Whilst there does seem to issues such as abortion, gun-control or Reganite economic strategies which might cause some voters to differentiate, the cleavages are not as strong nation-wide to be seen to lead to a new alignment. For this reason, there is a suggestion that there could be increased non-alignment in the US as parties are chosen on period specific issues and personalities. The increasing number of people who register themselves as independents highlights this pattern which is coupled to political apathy and lack of participation amongst the young.

The Case of Germany

Germany conforms again to the model presented in the British case. The German parties are, however, exempt of much of the history which is evident in the UK and the US. The major political cleavages are class and religion, although the catch-all nature if the German party systems means that the divergence is not very strong. Because of much the same reasons as outline in the British case, the two major parties can rely on core support and it is the smaller proportion of floating voters who will be responsible for changes in government.

Perhaps the greatest indicator of stability in voting behaviour is shown in the Länder, where there can been seen to be dominance by particular parties, such as the CSU in Bavaria or the SPD in the more industrial Saarland. Although the coalition nature of government does confuse observations, there is still a North-South class divide in Germany following much the same pattern as in the UK. The results of the EuroBarometer survey into the relationship between class and voting in Germany to show 'insignificant and trendless fluctuations' in the lifetime of the Republic, showing that such alignment can be easily identified.

An interesting development occurred in the first unified election in 1994. Here the class voting alignments seen in the West were reversed. The working class increasingly voted for the CDU and the more middle class voters for the SPD. This reversal was explained in relation to the perceived association between the SPD with socialism. Those who benefited from the East German Socialist system, those in the middle classes, voted for the party they saw as being the more to the left. Those who had faired worst out of the old regime, the working classes, voted for change away from the old system, towards the CDU. Since then, mixing and a realisation of which party is more in line with working class interests has reversed this trend, shown in SPD strength in the new Länder.

There did appear to be some form of watershed in German politics in the 1998 Bundestag elections. For the first time, a party other than the CDU, SPD or FDP had gained power, the Greens. They did this largely by displacing the FDP as a radical, libertarian alternative party. Their success does not mean that there is any justification in suggesting an end to alignment in Germany, but instead it highlights the power of a small proportion of the electorate in a well balanced, well aligned system. The bulk of the support of the major parties was traditional voters, with a few diverting from their CDU alignment due to short-term factors (namely the economy and a long period of one party rule). These electors would be expected to return to their previous alignment. What is perhaps interesting is that this is the first time that electoral results, rather than party manoeuvring has caused a change in government. It is, however, too early to see if this trend can be repeated. There is much evidence that because of the time that the Kohl government had been in office that this election can be viewed in much the same way as the 1997 result in the UK. For this reason, it would be wrong to speculate too far ahead.

This stability must also be seen to be related to the religious cleavage. Although German society is becoming increasingly secular, and East Germans particularly, this cleavage is still powerful. As was seen in Northern Ireland, in the absence of stronger cleavages such alignment can give particular parties great success over time. Such alignment is particularly relevant in the rural areas where political activity is less and so easier for people to slip into more traditional patterns of alignment.

Conclusion

In all, to answer the question posed in the introduction, it seems like these are clear cases of alignment of the bulk of the electorate. This is shown in the Democratic dominance of Congress in the US and the clear North-South divides in both the UK and Germany. This does not mean that variation from alignment is not important. Indeed, it is clear that often the small proportion of the electorate who do not align may be those responsible for election results in well balanced systems. Evidence further suggests that when there are periods of deviation from alignment, these are because of period specific political factors rather than sociological changes such as those proposed by Denver. Further, even in a time of changing class structures, links can still be made between classes and support for political parties. These links can be seen to be largely unchanged since the 1950's (if the appropriate classification is used, such as by Heath et al).

The importance of political scientists recognising that politics is in the periphery of most voters worlds and that a simple form of identification in order to make a choice should also be recognised as showing the importance of alignment theory. People gain their perceptions of both their interests and the political parties from those around them, and so sectoral based alignment should be expected. To associate with a political party as a source of education for issues that are not well known gives a relationship that can develop through adulthood, shown by those with stronger identification (and so alignment) seen to be the in older generation.

The high point of decrease in alignment can be seen in the 1970's, since when there has been a restoration of stability. In all these countries this can be accredited to major political events, such as the rise of the Liberals in the UK or the race issues in the US. These periods do seem to show a partial realignment of sectors of the population. The large array of issues that are within a political party in all these systems allows for such realignments to occur while other, more traditional alignments can remain. This explains much of the consistence in voting patterns. It is clear that while much of the changing in voting behaviour is due to deviation from aligned patterns, there are some cases of realignment and amongst the young, non-alignment. This being said, it must be repeated that in the UK in 1992 86% of the electorate did say that they identified with a particular party. This must show that is there is to be major changes to alignment theory, there is still a long way to go.