Chapter 2 Social Choice

Introduction

Policy requires choices.  Very often adopting one course of government action means choosing not to do something else.  This is often because resources are scarce, but it may also be because one course of action is incompatible with another for other reasons.  The objectives of public policy were explored in the preceding chapter.  Objectives inform the rationales or reasons for specific policy choices, but in and of themselves say little about what causes society to pursue policies directed at particular objectives and why society chooses particular ways of doing so.   In this section I will explore some causal explanations of policy choices.  The first part of the chapter will present theories based on general characteristics of society, leading to a discussion of the social welfare function, structuralism, group processes, and systems theory.  These theories have been largely used to try to explain both policy stability and periods of major policy innovations, such as the shift to highly regulated societies, the shift to high tax societies, the adoption of the welfare state, and the shift to deregulation and privatization, for example. Many of these theories are deservedly referred to as belonging to ‘grand theory’ in which causes are looked for within very broad and general hypothesis about society.  The second part will critique these approaches and present an alternate way of looking at theories  of how policy comes about.

Society and the Social Welfare Function

Policy is in some sense made in the interests of society. But what is society?  How do we know what the interests of society are?  Does it have a being of its own, beyond the totality of individuals within it?  And how does society function and make decisions?

Society is obviously complex, with a variety of elements to it.  One way of looking at society is to separate voluntary engagements from ones based on compulsion or coercion.  Voluntary exchanges can never result in changes that make some individuals worse off, no matter how much overall societal improvement might be possible from such changes.  These kinds of changes require some capacity to see the bigger picture and impose improvements for which it is possible that the gains are much larger than the losses.  One possibility is that the government or the state could impose such changes.  The state uniquely has the power to make changes that benefit some at the expense of others,  since it can use its power to make individuals accept changes they might otherwise reject.

But on waht basis might it conclude that the changes on net bring an improvement in society?  At one time it was popular to talk of a social welfare function.  A social welfare function represents the relationship between overall societal well-being and the well being of individuals within society proiveded that efficiency is realized in production.  Society’s welfare is then a function of the aggregate well being of the individuals within it.   A society that is efficient in the use of its resources can have a wide range of possible distributions of things of value between its members.  A change of distribution takes away from some and gives to others.  A well behaved social welfare function will exhibit increasing amounts of loses of well being for those who sacrifice well being for any given increases in well being for those who gain.  In other words the marginal utility of aggregate amounts of goods and services declines as the endowment of individuals increases, and vice versa.  This being the case, there is some optimal distribution where overall well-being is maximized.  This is where the marginal utility of incremental endowments is equal for every individual.

The social welfare function thus expresses the trade-offs involved in redistributing well-being in society for any given level of overall societal well being.  The social welfare function makes it possible to overcome the limitationsof the Pareto criteria in defining improvements in classic policy situations where some people are made better off and others made worse off.  It becomes possible to determine wether such a change improves overall societal well being.

A functional relationship of this type requires that changes in the value of dependent variable with respect to each of the independent variables be cardinal measures.  This means in this case that there must be a common measuring stick to express the magnitude of changes in well being for each and every individual. The problem is that there is no comparable measure of well-being for different individuals. There is no aggregation principle that can be used to determine aggregate social well-being based on individuals own valuations of choices and situations.  Individual sovereignty would have to surrendered to a despot in order to express a social welfare function that made any sense.  The idea of a social welfare function is unfortunately largely conceptual with no way known to make it operational.  It is not possible to pre-detemine a social welfare function of the kind described, and it cannot practicably be inferred or extrapolated from data from individuals.

Preferences are the expression of values by individuals, and as such reveal how individuals will rank choices in terms of desirability or contribution to well being.  The individual is the only and thus the final arbiter on such questions, a fact that finally cannot be squared with the outcomes sought from the social welfare function.  Optimization is defined in terms of getting the greatest value from a set of choices under given constraints.  Optimization is fundamentally tied to rationality, which means choosing the most highly valued available option.  The idea of optimization requires an optimizing decision maker with the needed information.  In  a society where only individual’s valuations are relevant, there is no such decision maker behind the social welfare function.

The motivation behind constructing a social welfare function is to get the maximum value for society out of policy decisions.  The idea of the social welfare function fails because it is impossible to bridge the gap between individual valuations and aggregate valuations.  It is not possible to add values together to get a sum if there is no common measure among individuals of the values of goods and services. There are further problems associated with getting individuals to state honestly the values they place on baskets of goods and services if the more general problem could be overcome; something to which I will return.

There are some analysts who attempt to by-pass the problem by assuming that dollar values of bundles of goods and services can be used as a proxy common measure of value.  This in effect assumes that a dollar has the same value or worth to each individual.  By the same we mean the same underlying value such that individuals all get the same objective benefit from a unit of something of value.  But this is something that is beyond our ability to know.  An individual may have some sense of how much value she gets from something but she can never know how it compares with someone else’s valuation.  This the claim that a dollar’s worth of something has teh same value to all people is artificial, and if it is done, the social welfare function itself becomes trivial.  It is inconsequential how society’s wealth is re-distributed.  When some change makes some better off and others worse off by the same dollar amount, it doesn’t change a thing in terms of society’s well being.  Taking from Mary and giving to Paul has no normative consequences.  No ethical consequences arise from a larcenous state that takes from the poor and gives to the rich, even though it runs counter to the most widely shared understanding of ethics.

Voting and Social Choice

Voting involves the aggregation of individuals’ preferences.  One apparent way of attempting to get the highest valued choice for society based on individuals’ assessments of the worth of different choices is to get them to express their preferences through voting.  Voting is after all a clear and objective way for people to express how they rank one choice relative to another in terms of desirability.  Ranking of choices in terms of desirability is what is known as preference ranking.  It is an expression of a person’s underlying preferences regarding choices presented to her.

However, upon examination voting clearly has its problems.  It depends upon members of society participating, a rule is needed to determine when an alternative wins (majority support, 2/3 in support, unanimous consent, as examples), a determination needs to be made whether teh result is consistent and independent of the way the choice is offered, and someone needs to determine what the choices are and how the ballot is worded.  And it can also run into problems if the choices are public goods or services.  And what is the ethical basis for determining outcomes by votes if they are to empower the state to take away from some to make others better off?

Let us look first at the problem of public goods.  With public goods, indviduals can be expected to provide misleading information about their preferences through voting.  Suppose the vote is on building a new public park that will be open to all to users.  A public vote might lead some to abstain or vote no, even though they want and will benefit from the park, for fear that those who vote yes will then be subject to a tax on use.  Once again the free-rider problem arises.  In many cases the simple device of a secret ballot can help to resolve this, but not completely.  Nevertheless a simple vote cannot measure the intensity of the benefit to those in favour compared to those who oppose.  Those in favour might be families with children who strongly want and need park space and/or people who put a high value on outdoor activity and natural environments, while those opposed may simply be grumpy tax payers for whom the cost is very low, but who simply object in principle to any more spending.  As long as the latter make up the majority, the park will not proceed, even though the preferences of those in favour are much more strongly held than those opposed, and for more legitimate reasons.

Let us now turn to the rule that determines which choice wins.  The majority voting rule is the most popular and the one most often associated with democracy.  But majority voting really only determines that more people favour a course of action over another offered for consideration.  It doesn’t ensure that aggregate values have been maximized.  Suppose 51% of the people each get a very small net benefit from the policy, and 49% each experience quite substantial loses.  The policy will proceed, even though net benefits are negative.  This problem is not made any easier to because of our inability to develop a measure to compare quantitatively different individual’s benefits and losses, but it is certainly plausible that such outcomes could occur.  There is no reason to suggest that a two-thirds voting rule would be any better, other than it reduces the risk of imposing a loss on people who while smaller in number feel much more intensively about an issue.

Generally, when we think of voting, we think in terms of the decision at least being a clear, unequivocal representation of aggregated preferences.  In fact, voting does not necessarily give rise to clear, stable and unequivocal results.

This can be illustrated by a simplified example.  Suppose we have three voters, and they are asked to make a choice through majority voting between three different policies, A, B, and C.

Suppose also that their preferences are as follows:

Voter 1 prefers A to B to C.

Voter 2 prefers B to C to A.

Voter 3 prefers C to A to B.

If we ask them to vote by choosing first between any two of the alternatives, and then choosing between it and the third alternative, we get following peculiar results:

If we have them vote on A vs B, A wins.  If they then vote on A vs C, C wins.  So C is the preferred choice.

Now suppose we first ask them to vote on C vs B, B wins.  If they then vote on B vs A, A wins.  Now A appears to be the preferred choice.

Now suppose we first ask them to vote on A vs. C.  C wins.  If they then vote on C vs B, B wins.  Now C appears to be the preferred choice.

There appears to be no clearly preferred policy.  The one selected depends on the order in which the choices are presented.  Because they are asked to vote yes or no, which is standard in voting on questions, the votes must be ordered in sequence.  But the ordering of the sequence matters, and indeed the result is that any one of the choices can be chosen depending on the sequence.  The social choice is neither stable or consistent, and it is thus not possible to refer to a stable equilibrium choice.  This is completely inconsistent with the idea of maximization, where there is a best solution and it will always be chosen provided preferences do not change (i.e. are stable).

This problem, called the voter’s paradox, does not always arise in a three person, three choices vote.  Consider the following:

Voter 1 prefers A to B to C.

Voter 2 prefers C to B to A.

Voter 3 prefers B to C to A.

On a vote of A vs. B, B wins.  B wins over C on the following vote.

On a vote of A vs. C, C wins.  B wins over C on the following vote.

On a vote of B vs. C, B wins.  B wins over A on the following vote.

In this case there is a clear and unequivocal result that will be stable.  B is the clear winner

The reason for the difference in result is the structure of the underlying ordering of the preferences.  This can be explained by reference to a spatial mapping of the preferences.  In the latter case, each individual has a single peak to his or her preferences, when the preferences are mapped into two dimensional space, and when the choices from A to C can be mapped along a continuum which expresses the choices as more or less of some aggregative underlying attributes with A representing the least and C the most.

In the latter case, the middle choice never assumes last place in any of the preference orderings.  In the former case, individual 3 ranks B, the middle choice, last.  It is this that makes the difference.  In the case where the middle choice is ranked last, the voter prefers the two extremes to the middle.  This is referred to as the absence of a single peak in the preference ordering when it is mapped into two dimensional space.  Rather it can be thought of a trough, where the middle is in a lower place on the map then the two outside choices.  If such a trough exists, the preference order is not single peaked.  The analogy is to a roof, which either has the shape of  /  or

However, if the ordering appears as V shaped, it does not have a single pea

This can be presented diagrammatically as follows:

Figure 1

Single Peaked Preference

Figure 2

Non-Single Peaked Preference

Where single peaked preferences are absent, the outcome of voting depends in part on the way and the sequence in which the alternatives are presented.  The party controlling the balloting can determine which choice to present first, and thus what the eventual outcome will be.  In addition, it opens up the possibility of policy cycling, in which different and inconsistent policy choices are made simply because the ordering of the ballot questions is changed.  This inconsistency has nothing to do with the rationality or inconsistency of individuals.  Each individual has perfectly consistent preferences, and behaves rationally in expressing their ballot choices.

The problems with non-singled peaked preferences do not suggested that these conditions prevail in most cases where majority voting could be used to determine society’s choice of policy.  The important point is that the idea of maximization leading to a single, stable optimum appears to make no sense, as a general proposition.  We have already seen how the inability to get people to reveal their preferences can create problems, and also how the absence of no observable and measurable data on individual values that can be compared to others and added up results in difficulties for optimization models.  These are practical problems that arise in applying an optimization model to social choice, which could possible be solved by yet to be discovered indirect estimation techniques.  The voters paradox is an entirely different kind of problem.  No new or better estimation techniques could ever resolve it.  Rather it opens up the possibility of a new kind of problem, suggesting that it simply doesn’t make sense to think in terms of optimization models when a number of individuals are involved, since there are some cases where there is no clear and unequivocal best choice.

The Median Voter

In the case of a large number of voters with non-peaked preferences, and where there is a choice between a large number of alternatives can be ordered in ascending or descending order in terms of greater or lesser amounts, the a simple majority vote will settle on the preferences of the median or middle voter.  This is because in such a case, exactly the same number of people will want less as who want more, and the median voter will determine the outcome.

The sense in which this is the preferred outcome of society is simply that, excluding the median voter, as many support it as oppose it as a choice.  The fact that the median voter rules is simply as a result of the majority voting rule.  A different rule would bring a different result, but there is only one real normative basis for supporting this rule as opposed to another.  The only sense in which the majority voting rule can be said to be preferable is that it tends to minimize discontent.  It means that at least half the people will support any option chosen.  And  under a higher standard, such as 2/3 support, the number of frustrated voters will be higher because it rules out a number of alternatives which could draw support from between half and two thirds of the voters.

There is no assurance that the majority voting rule provides that only potential Pareto improvements will be approved.  It is possible that the total net costs to those opposed is much higher than the total net benefits of those in favour.  The simple majority rule cannot be depended upon to avoid this possibility.  The self-interest of the median voter will rule supreme, given sufficient knowledge of her preference or sufficient time for the system to discover and place her preferred position on the ballot.

For instance it is conceivable that the median voter may prefer to have everyone with more wealth than hers stripped of their wealth until they have nor more or less than her, and that she might gain majority support for this, particularly if that wealth was to be re-distributed to her and all the others with less wealth until everyone had equal wealth.  This is particularly plausible if we accept the proposition that everyone is motivated by simple self-interest.  In fact we know that this could and would not happen.   The reasons among other things are that customs, conventions and laws prevent such a question from being put on a ballot and from being implemented, individual values about confiscation of other people’s property are much more complex that the simple self-interest motive suggests, and people are concerned about such things and investment and growth, and not simply distribution.  It is more plausible to suggest that the amount of re-distribution policy that society gets or adopts is close to that preferred by the median voter, but the amount might be will depend on the factors indicated above, and the impact and relevance for her.

Democracy and the Institutional Basis for Social Choice

Regardless of the merits of ballot voting on individual issues, democratic societies are much more complex than is pictured by this type of decision making.  Democratic societies are much more complex, having a number of features which are relatively common and provide the basis for their functioning.

First, voting in democracies is primarily directed at electing representatives, who when elected are members of legislatures, parliaments or other representative forums, or  assume executive positions such as President.  The voter delegates decision making and authority to representatives, whose votes or decisions may in turn determine policy, among other things.  This is sometimes known as representative democracy.  Those who are elected may in turn delegate decision making power and authority to others, in the form of executive power.  Many of those with executive power are not elected, including those in the public service or bureaucracy.

Second the representatives in democracies are chosen through competitive elections.  Elections are fair, relatively frequent, and competitive.  Voters chose their representatives by a secret ballot, and can be confident that their vote will not be interfered with, and is without direct personal consequences in terms of harm.  Those who seek to be elected are free to associate and compete on the basis of political parties, and parties are free to contest elections and communicate with voters without interference.

Third, there are certain over-riding rules and conventions that are adhered to within society, and that provide a foundation for effective voting, participation, and representation.  These include freedom of association, freedom of expression, freedom of the press, freedom of speech, and protection against arbitrary interference by the state in their lives and their affairs.  The most fundamental rules and conventions form part of the constitution, and can only be changed under conditions of a high standard of consent of citizens.

Fourth, the societies function in accordance with the rule of law and judicial review.  What people can do, and more particularly what they can’t do, is set out in laws.  Laws are either contained in the constitution (in some cases unwritten as is the case with common law) or in formal expressions of the will of legislatures or parliaments.  Citizens can have a high level of trust that the governments will respect and enforce laws, and in cases of a dispute that the matter will determined by an independent court or judicial system, in a fair and unbiased way.

One of the things that people often find difficult to understand is that the protections and freedoms in a democracy are not absolute.  Protecting one person’s rights or freedoms may conflict with the full exercise of the same right or freedom of another.  Governments can and do pass laws and take actions which interfere with some people’s rights and freedoms, but the rule is supposed to be that the infringements must be solely to advance the rights and freedoms of others, the trade-offs must be reasonable and justified, and the interference must be minimal under the circumstances.  Everyone has the right to a judicial review of whether the government’s interference is within acceptable limits, given the protections provided.  Unfortunately, there is a possibility that the determination of what is acceptable can be quite subjective and not always as clear as one might think given the principles.  One example is the interference with the rights of protestors to post signs at international conferences which might be within sight of and embarrassing to foreign leaders.  Another is the infringement of aboriginal rights solely because they interfere with the interests, rather than the rights of others.   The acceptance that rights may be infringed because of other peoples’ interests rather than rights opens very wide the door to infringements of rights.

Public Choices in A Representative Democracy

In a representative democracy, it can be argued that it is those who are elected by voters who are responsible for choices made by the government, not the voters themselves.  Further, modern governments are highly complex, with many officials and agencies, most of whom are not elected but who are in one way or another authorized to take responsibility for what government does.

This does not mean that voters do not make choices.  Their most obvious choice is made at the time of elections.  Electoral choices involve quite complex decisions by voters, in which various attributes associated with the choice are involved.  These include the personalities and character of the party leaders, the personality of the individual candidates, and the various party policies.

Elections are an opportunity for candidates, usually organized in terms of parties political, to put choices before the electorate.  Thus in elections the choices that the voters have available to them are limited to those put before them.  In elections political parties are agenda setters, each placing before the electorate in an electoral district, one particular choice.  The choice available to voters at particular election is between  alternative agendas, as represented by competing political parties.  If each political party is competing in terms of agenda that contain essentially the same attributes, differing only in the degree of intensity (e.g. more or less government spending, more or less tough on crime, more or less change in health care provision, etc.), then an index representing the amalgam of these attributes will express dimensions of choice as expressed by a party agenda.  This is similar to the proposition that people “combine the attributes of a multidimensional choice situation into a ‘total score’, such as ‘net gain’” (Mintz 1993, p5).   The idea is that people are able to trade-off attributes; if one is low on an intensity scale and another high, the latter can compensate for the former.  The result is with intensity being a monotonic function of the intensity of each of the separate attributes (or dimensions, as Mintz calls them). With a large number of voters, this provides a well-behaved functional transformation of intensities of individual attributes into the overall intensity index (the function will be well-behaved if, for each attribute, holding all other intensities constant, an increase in intensity of one produces an increase in the intensity index).  This intensity index then provides the dimension of choice in terms of which voters arrange their preferences.  The parties present their agenda as a basis of choice for the voters, directed toward meeting voter preferences.

Under these conditions, a party with complete information and complete flexibility in the choices it can put before the electorate, and which has the sole objective of winning, will present the choice of the median voter.  The extent of the win will depend on how far away from the choice of the median voter the other party or parties position themselves.  In terms of the electorate, the prefererred position of the median voter, which divides the voters into two equal sized groups, will attract at least a simple majority.  So if a party is able to present the choice of the median voter, its position is unassailable, and all other parties will crowd toward the position of the median voter in subsequent elections.  With complete information, parties all crowd toward the centre, and the votes of those at the extremes become more and more irrelevant, reducing the incentive for them to vote.  As long as the peeling away of voters is symmetrical relative to the dimension scale, this will have no affect on the outcome.  Thus a world of reduced voter turn out could simply represent efficiency on the part of the individual voters, since the only vote that counts is the median voter.  The withdrawal of others from voting, provided it happens more or less symmetrically from either end of the spectrum, will have little impact on the outcome, with efficiency (i.e. not devoting time to activity with no expected return) being the explanation.

In a actual experience, it is appears that the crowding to the center and the dropping off of voters from the process is becoming common.  However, if this description is the whole story, than we should also have stable election outcomes over time at the voting equilibrium as described.  But this does not appear to be the case.  Rather in most countries there are frequently elections that parties win by a large majority, only to lose badly the next time around.  Why does this happen?

First, there is the possibility that the voters preferences have changed.  The response of each voter to the index which provides the dimension of choice described is rooted in their preferences.  The proposition is that voters can rank order their preferences in response to each of the index values, that the preferences of each voter are single peaked, and that each voters responds rationally to the choices presented.  Stability or a continuation of the equilibrium from one election to the next also requires that preferences be stable.  Political parties are not necessarily passive in their relation to voters preferences.  Political parties committed to a certain view of society and the need for fundamental change will attempt over time to get voters to understand the world differently, and thus aim of a transformation of preferences.  This approach is essentially rooted in a belief in the individual as capable of development and learning, out of which comes a fundamental changes in preferences.  Contemporary political scientists, perhaps under the influence of economists, put little credence in such a belief.  And indeed, numerous empirical studies suggest that preferences are quite stable in the short and medium term, and thus that the volatility in voting results from one election to the next is unlikely to be due to shifting preferences.

A second possibility is that not all voters have single peaked preferences.  The preceding result implicitly assumed single peaked preferences.  Suppose voters can be grouped into three groups, each defined by relatively similar preferences, and each roughly equal in number.  If the preferences are single peaked, the winning combination is that which represents the median group of voters. However, if one of the group’s preferences has a trough rather than a peak, the voter outcomes in sequential elections could be highly unstable.  In this case, one of the groups of voters prefers either of the extremes over the middle of the road.  An example might be social conservatives who, compared to a liberal middle of the road view of state power and intervention, might prefer either extensive state intervention in accordance with their social views, or if that isn’t possible, minimal state involvement along libertarian lines.  What they like least is middle of the road liberalism, in which everything is a trade-off.  The parties, in trying to put together a winning coalition of voters, will attempt to construct programs that will attract the middle voters and some significant number who prefer the extremes.  The choices presented by the parties from one election to the next may shift quite significantly, in that now what becomes important a strategy that will cobble together a program that attracts the middle and enough of those who favour the extremes over the middle to pull off a win.  As has already been shown, where there are non-single peaked preferences, the outcome of voting is not going to be a stable equilibrium.  Rather there is likely to be instability and cyclical results as the parties jostle to shape the agenda and thus form a winning coalition from one election to the next.

A third possibility is that the agenda changes from one election to the next, but for different reasons.  The choice of the agenda set by a party may not be entirely determined by coalition politics.  It is possible that voters or parties make changes in their judgments about the attributes of the choices that are important.

Up until now, preferences have been referred to as belonging to individual voters.  Political parties and candidates develop strategies that involving putting choices before the voters.  These strategies yield a complex election program, which up to now have been treated as objects of choice presented to voters.  Voters’ choices at any one time are limited to the alternatives actually presented, and they are small in number.  In a system with strong party discipline and party based voting, the number of alternatives presented is essentially equal to the number of parties contesting the election.

The choices presented by parties can also be thought of as preferences, but of parties now and not of individuals.  As such they represent the aggregation of individual preferences that they believe will gain the greatest number of votes, perhaps subject to certain constraints placed on parties by party members, ideology (unless dead), and historical circumstances.  Parenthetically, it is clear that a typical party platform will not be some kind of additive aggregation of the preferred positions of all voters, since such an aggregation cannot be presented as a voting choice unless preferences are identical.  Prior choices have to be made in the process of assembling the choice to be presented to the electorate, and trade-offs of a fully informed vote maximizing party will be made on basis of the assumed or underlying preferences of the members of electorate, according to  the trade-offs between attributes by the voters that lead to their mapping of their preferences.  The choice presented to the electorate by a party is thus a revealed preference itself, but of the party.

The choice presented by the party is the preferred position of the party on the derived index scale referenced earlier. One of the possible reasons for the instability of electoral outcomes can be instability in party preferences, and not changes in the preferences of individuals.  But why would party preferences shift if individual preferences are relatively stable?

There are a number of possible reasons.  Previous constraints may be relaxed, as party members cast over-board relics of the past, whether people or positions on specific issues.  New constraints may also be added.  A term in government may embed in the voters mind certain perceptions of a party as to make it impossible to present its preferred agenda as clearly or as unambiguously as it would prefer.   Or information about voter preferences may improve, if for no other reason than that each successive election adds information.  The processing of such information may also improve through such things as better leadership, more representative membership, stronger and better advisory processes, and better research.Another possibility is a shifting external environment.  As we have already indicated, the program presented by a party can be seen as an agenda.  There may be agenda shifts originating outside of the neat structure of voter preferences and the resulting structuring of constrained voter choices by parties, of the kind we have been discussing.  (Jones 1994).

Endogenous shifts, or ‘shocks’, may cause an agenda shift.  Even vote maximizing  parties may not be able to present a stable program to the electorate election after an election because the environment changes.  An example might be an unexpected change in circumstances in society, such as a major environmental disaster.  Another might be a shift in media attention so that some issue suddenly receives huge attention.  Yet another might be where some organized interest mounts a highly effective campaign to shift the agenda.

Such cases may affect voters, or they may impact on parties as they formulate positions.  The new circumstances may cause voters to evaluate choices differently, or they may cause parties to evaluate choices differently.   Parties may simply respond because of uncertainty about how voters will respond, or because they themselves tend to respond to media and events as part of an overall attention shifting process.  The latter may happen simply because how they reason about policy may in part be affected by what is current and about which there is new and widely available information and discussion.

Whatever the underlying reason the salience of an issue changes.  This can be either the cause of or an effect of parties evaluating choices differently because of an external shock.

The Policy Process

Up to this point, the agenda of the winning party has been treated as a choice within which are embedded a number of policy choices (one for each attribute).  Thus the winning parties agenda becomes policy.

This in itself is not a realistic position.  Just as voters preferences are interpreted through parties and elected representatives, so too party preferences may be interpreted through Ministers or Secretaries and public agencies.  We are left to describe a mechanism has yet been presented that describes how the preferences of parties and most particularly winning parties are transformed into policy.