n The 1996 election continued the exodus of nearly 21 million voters out of the ranks of the two major political parties that began in 1992. Most of these migrating voters supported Ross Perot in 1992 then stayed home in 1996. The GOP has been especially hard hit by this migration, losing 16 million voters over the past two Presidential election cycles (a loss partially offset by new, first-time GOP voters) — the result of the fracturing of the coalition formerly known as Reagan Democrats.

President Clinton lost the support of 2 million who voted for him in 1992, a loss that was offset by his solid showing among first-time voters (4.5 million of whom voted for him in 1996). His 6 percent increase in total vote (1996 versus 1992) was the second most anemic second term election performance this century, following Eisenhower in 1956 (the average increase in the total vote for a president winning a second term is 25 percent).

n While Clinton attracted no significant new constituency to his 1996 vote tally, his vote became decidedly more economically upscale, losing 2 million votes from households with less than $15K income, gaining 3.5 million votes from $50K+ income households.

n In light of these facts, the post-election analyses by Democratic pollsters Mark Penn and Stan Greenberg have a surreal quality; each endeavors to explain an illusory gain in the president’s vote which is detectable as a percentage but not in actual votes.

n Women fled the electorate: 6.4 million fewer women voted in 1996 than in 1992. The President’s purported success in appealing to the female vote actually was quite narrow. The 2 million additional female votes he received in 1996 came from women who were: first-time voters; Hispanics or Asian-Americans; over 45; unmarried. President Clinton gained far more votes from married men (1.3M) than from married women (500K). The president and Bob Dole both attracted additional votes from women 60+ (versus 1992); both candidates hemorrhaged female voters under 45. Because of these diverse vote dynamics, no generalization can be made about what the female vote did in 1996.

n Seniors (60+ years of age) bucked the national trend by increasing their turnout. Bob Dole was the beneficiary of this effect, attracting 3.9 million additional 60+ voters versus an increase of 2.8 million for Clinton. The 60+ cohort was Bill Clinton’s best in 1992, his worst in 1996.

n Union households turned out in greater numbers (up 2.3 million) doubtless due to Labor’s $40 million "voter education" effort. But 47 percent of these new union voters went for Dole — although only 30 percent of union household voters overall went for Dole.

n The turnout fall-off among younger voters, 18-29, was especially pronounced; all three candidates experienced dramatic declines.

n Turnout fell among Democrats, Republicans, and Independents; Liberals and Moderates. Only Conservatives increased their turnout (by 1.2 million) between 1992 and 1996.

The political task facing Republicans is not to deny Democrats any of their current support, but to lay claim to Perot’s eight million voters and to reengage those disaffected voters who despair for the non-material quality of life in America.

 

---

The Meaning of Clinton’s Reelection

Political observers have been treated to the post-election spectacle of dueling Democratic pollsters, who are battling to define the meaning of President Clinton’s reelection victory. At issue is the historical and strategic lesson the Democratic Party should take away from the 1996 Presidential election as the Party contemplates its future.

On one side of this debate is Mark Penn who, with his partner Doug Schoen, served as the President’s pollster during the 1996 campaign. Mr. Penn conducted a post-election survey for the Democratic Leadership Council (DLC), the analysis of which he titled, "The Mandate for the Middle." On the other side is Stanley Greenberg, President Clinton’s 1992 pollster. His post-election survey was conducted for The Campaign for America’s Future. Playing a supporting role on Mr. Greenberg’s team is Ruy Teixeira of the Economic Policy Institute.

The disagreement between these two accomplished analysts of public opinion is in essence this: did President Clinton win because he is a "new Democrat," who successfully appealed to middle class, MOR (middle of the road) voters [Penn]; or did he win because of his uncompromising defense of traditional entitlement, environment, and education programs, which appealed most to a "downscale electorate" [Greenberg & Teixeira]. If Penn is correct, the implications are that his party ought to embrace the reality of limited government, balance the budget, reform Social Security and Medicare to preserve solvency, and use government to "equip people to solve their own problems." If Greenberg is right, Democrats ought to eschew entitlement reform, reintroduce universal health insurance, and maintain their essential faith in government activism.

Both analyses are, in their way, stunningly specious. Each interpretation hinges on an analysis of the increase in the Clinton vote. But there was no increase in the Clinton vote worth writing home about; their squabble is rather like pulling the drain plug on a full bathtub and then arguing over what made the tub ring move higher.

President Clinton won reelection with 49 percent of the popular vote, an increase of 6 percent over his 1992 showing. But his 1996 popular vote was a modest 2.5 million more than 1992 (coincidentally, also a 6% increase). Since 4.5 million of his vote came from first-time voters, the president lost the support of 2 million folks who voted for him in 1992. This substantiates anecdotal data obtained by traveling about the country during 1996 conducting focus groups and chatting with voters: one never encountered voters who said, "I voted for George Bush or Ross Perot in 1992, but in 1996, I’m going to vote for Bill Clinton." One could occasionally find, on the other hand, voters who did not intend to repeat their vote for Clinton in 1996.

To be sure, some unknown number of persons did vote for a candidate other than Bill Clinton in 1992, then switched to Clinton in 1996. But whatever this unknown number of voters was, 2 million more defected from Clinton, voting for him in 1992 then mostly staying home in 1996.

The data in figure one present the average increase in total vote experienced by a President winning reelection to a second term. The significance of this data is in the rarity of such a candidate winning without adding substantially to his vote total and establishing new constituencies of support. President Clinton did neither.

The Pew Research Center for People and the Press noted a similar phenomenon in the trend of Presidential approval ratings: while Bill Clinton’s current approval rating is among the highest he has yet received, the post-election bounce he experienced is the smallest ever recorded for a President winning a second term. "It is the lowest rating for any president since World War II recorded immediately after reelection to a second term" [Pew].

In a year in which 8.2 million people fled the electorate and another 3 million or so stayed home who "normally" would have turned-out (leaving aside the other 90 million of voting age who did not participate), for Clinton to have maintained his vote total — indeed, to have augmented it a bit — is a pretty substantial accomplishment.

While acknowledging Clinton’s accomplishment, his performance does not represent a building of the Democratic Party for the future. The Democratic Presidential vote remained less than a majority — and some of the groups experiencing the greatest turnout fall-off are traditional bedrock Democratic supporters.

There is some proportion of the electorate that votes reflexively for the incumbent. These voters often say they "vote for the best candidate," but "best" means to them the most experienced or tested candidate — viz, the incumbent. We don’t know how great this incumbent advantage was in 1996; in 1984, Ronald Reagan running for reelection improved his popular vote by 24%, some portion of which were these "automatic incumbent voters." If Bill Clinton had done as well, his vote would have been 55.7 million, improving on his 1992 showing by 11 million votes, versus the 2.5 million gain he actually experienced.

In 2000, an incumbent will not be on the ballot, so these incumbent-defaulting voters will be up for grabs. The implication for the 1998 congressional election is that the decline in core Democratic voter groups experienced by Clinton (described below) will impede Democratic congressional prospects. The first-time voters who went lopsidedly for Clinton in 1996, on the other hand, are notoriously unreliable as voters in off-year elections.

Bob Dole would have won with 47.5 million votes, ceteris paribus. This goal represents an increase of 8.3 million over George Bush’s 1992 vote total. Instead Dole received just 100,000 more votes than Bush — even though there were 11.7 million former Perot voters up for grabs, the vast majority of whom stayed home election day. Since 2.8 million of Dole’s tally came from first-time voters, Dole lost the support of at least 2.7 million 1992 Bush voters. This is again a net figure; we don’t know exactly how many 1992 Bush voters did not support Dole. Dole may have lost 2.7 million Bush voters and picked-up no 1992 Clinton or Perot voters; he may have lost 3.7 Bush voters and picked-up 1 million Perot voters. All we know is he received 2.7 million fewer votes than Bush from 1992 voters, and that he received 2.8 million votes from first-timer voters.

The Methodology of this Analysis

The real story of the 1996 election was the exodus of 8.2 million from the electorate; and other dramatic migrations occurred within the electorate as well. In order to capture these dynamics, we reconstructed the total vote of key demographic groups for major party candidates in the 1992 and 1996 presidential elections as implied by the published exit polls of Voter News Service, then examined the change in real votes that occurred between 1992 and 1996.

In a year when election participation declined as precipitously as it did in 1996, exit polls reporting the percentage candidates received from various voter groups do more to obfuscate than illuminate what is actually afoot in the American electorate. An example: the percentage of female voters who selected Clinton increased from 45% in 1992 to 54% in 1996. So one might conclude from this the president was unusually successful in appealing to female voters. But in real terms, the president attracted a fairly modest 2 million additional female voters while 6.4 million female voters (net) stayed home. If Clinton’s electoral magnetism with women was strong, would he not have kept them in the electorate?

Post election surveys also tended this year to obscure what was going on among the electorate. Surveys are not dynamic — they are snap-shots, and do not as a rule focus on the change in the behavior of voter groups from one election to the next, hence they do not consider the net effect of voter shifts. For example: Stanley Greenberg contends that 10% of the electorate are "new Clinton voters." This translates into 9.3 million voters. We know 4.5 million are first-time voters who voted for Clinton, so by inference 4.8 million people (to Stan’s delight) voted for someone other than Clinton in 1992 then switched to Clinton in 1996. But using Stan’s figure, we also know that 6.8 million people voted for Clinton in 1992 and not in 1996. So while Stan’s static snap-shot analysis dissects the 9.3 million "new Clinton voters," it ignores the 6.8 million former Clinton voters brought to light by our "net" analysis.

References to shifts or movements in the electorate in this report should be understood as net shifts. Going back to the Dole vote for an illustration, we don’t know the actual number of 1992 Bush voters who did not vote for Dole, nor do we know the actual number of 1992 Clinton or Perot voters who did vote for Dole in 1996. But we do know that Dole received a net of 2.7 million fewer votes from 1992 voters than did Bush.

The flaw of this analysis, which is inherent because it relies on exit poll data, is that it considers the electorate to be a patchwork of demographic groups. But demographics are a very poor way of understanding the electorate; demographic characteristics do not begin to explain why the voter shifts described here occurred. The on-going "homogenization" of American society results in politically relevant attitudes, experiences, perceptions, and behaviors being ever more finely dispersed throughout society, with the result that demographics are progressively less politically significant. The only way to really understand the American electorate is as landscape of constituencies with shared attitudes, experiences, perceptions and values. Exit poll data are insufficient to paint this landscape.

The Penn Thesis: Donning the Mantle of a New Democrat

Mr. Penn’s central assertion is that President Clinton won because "voters recognize and approve of [his] New Democrat Centrism." This centrism consists of successful alignment with "regular Americans" or "the forgotten middle class" on the issues of a balanced budget (for), welfare reform (for), taxes (for cuts), crime (against), and immigration (nuanced: for stronger enforcement, but not so as to alienate Hispanics and Asian-Americans). The "New Democrat Centrism" also consists of "balancing the competing desires of the people for, on the one hand, a limited government that works within the constraints of new fiscal realities and, on the other hand, an effective government that continues to fulfill the roles that only government can fulfill." The "voters’ preference," writes Mr. Penn, "is for government that neither smothers nor ignores, but rather empowers and equips. President Clinton represents this balance."

But let’s be clear: Mr. Penn’s centrism is really just slang for a willingness to entertain entitlement reform (read cuts), and this is the principal disagreement between Mr. Penn and Mr. Greenberg.

Mr. Penn contends that as the result of his "new centrism" President Clinton won back key voter groups such as independents, married voters with children, and the middle class, while attracting support from new key constituencies: youth, women, Hispanics. The difficulty in substantiating Mr. Penn’s analysis is that there is no necessary or logical reason why these groups — and not others — would respond disproportionately to Clinton’s "New Democrat Centrism." Indeed, Mr. Penn does not offer up evidence that these groups dis-proportionately recognized or approved of Mr. Clinton’s "centrism." As the graph below illustrates, only half of American voters overall consider Bill Clinton to be a "New Democrat," and this is the thin reed on which Mr. Penn’s analysis rests. Recognize also that the dichotomy of "centrist" versus "traditional liberal" is an artificial construct, conceived of by Mr. Penn and imposed upon public opinion — these are not terms of reference voters themselves use.

Figure 3 reveals what actually occurred among those voter groups highlighted by Mr. Penn. First off, notice that every one of Mr. Penn’s "key" components of the electorate lost voters between 1992 and 1996, except for Hispanics (married males don’t count — they weren’t specifically cited by Mr. Penn). This decline was driven principally by the non-participation of ’92 Perot voters, but abetted by the failure of Clinton and Dole to divert much more than a trickle of that flood into their vote totals. In this light, it is difficult to argue the Clinton strategy was unusually successful in appealing to any of these groups.

Hispanics are an exception, of course: they gained substantially in turnout, and nearly 90 percent of the new Hispanic voters went for Clinton. Also an exception are Catholics, most of whom stayed in the electorate: the gains of Clinton (2.4 million) and Dole (500,000) roughly equal the decline in the Perot vote (3.1 million). But Catholics are not one of Mr. Penn’s key groups — he dismisses the increase in the Catholic vote for Clinton as part of the larger Hispanic phenomenon. To be sure, part of what sustained the Catholic turnout was the increase in Hispanic voters, but note that of the 2.4 million additional Catholic votes Mr. Clinton received, only 1.5 million can be attributed to the Hispanic vote (and then only if all Hispanic voters are self-described Catholics).

President Clinton gained a substantial number of votes among women, but for every new female voter he attracted, three decided to stay home; this is not evidence, as we have argued, of electoral magnetism. Plus, the story of gender in the election is nuanced, and elaborated upon later in this report.

Young voters certainly showed no allegiance to Clinton: he attracted 1.2 million fewer 18-29 voters, even while attracting 4.5 million first-time voters. One could argue, we suppose, that since Dole lost 2.2 million young voters (versus Bush’s 1992 total) Clinton actually improved his margin by 1 million. But this is a relative phenomenon; Clinton did relatively less poorly among youth than either Dole or Perot. In the aggregate, the youth of America turned thumbs-down to all three major party candidates.

Clinton performed weakly among Independents, increasing his vote by a scant 200,000 versus the 3.1 million leaving the electorate. Among married voters (Penn actually cites voters married with children, for whom data are not yet available), both Clinton and Dole did well, so that Clinton’s comparative advantage is modest.

Clinton did fairly well among the lower middle class ($15K - $30K household income), attracting nearly 1 million more votes from this cohort, while Perot gave up 3 million and Dole gave up 800,000. The "above average" middle class ($30K - $50K) was not attracted to Clinton in any significant numbers, but was repulsed by Dole, so this group too was relatively good for Clinton, but arguably not because of Clinton’s efforts.

Why Democrats Did Not Win the House

According to Mr. Penn, it was because they did not shed their liberal bias. "Congressional Democrats are perceived as still clinging to old-style liberalism" [undefined by Mr. Penn]. Congressional Democrats made two mistakes, Mr. Penn writes: they failed to join Clinton in the center, demonstrating fiscal moderation. Second, they failed to join him in riding the wave of good economic news in 1996, owing to the fact "Congressional Democrats invested [too much] in wooing the ‘downscale’ voter." Penn is right about the "downscale voter:" they fled the electorate.

Implications for Republicans

Although he has the evidence at hand, Mr. Penn profoundly misinterprets the historically low turnout of 1996. Almost none of the non-voters with whom he spoke said they were "trying to send a message of discontent" through their non-vote, so Mr. Penn concludes there was "not any significant sense of dissatisfaction with the candidates." He is wrong.

As Ben Wattenburg would say, "values matter most." This is evident from Mr. Penn’s own data. It is evident in our surveys, which find 80 percent of the American people agree with the proposition, "there is a crisis of declining individual morality in the country today."

The American people wanted in 1996 a serious discussion about how we are to arrest the declining moral character of American society. What they got instead was a conspiracy of silence between the three major candidates so as to avoid a substantive and potentially divisive debate. This is why so many voted with their backsides.

The Pew Center post election survey finds that Republicans are the voters most pessimistic about the societal health of our country. But Perot voters are not far behind in their extent of pessimism, and a substantial number of non-voters are similarly concerned.

Mr. Penn incorrectly asserts that the Clinton campaign did invoke a values debate with a series of "family values initiatives" concerning education, tobacco, family and medical leave, violence on TV, children’s health, curfews, school uniforms, keeping teenage mothers in school, and drug testing teens who are seeking driver’s licenses. "What they [the Dole campaign] utterly overlooked is that while parents want government increasingly out of their own lives, they welcome government help in their fight against outside influences corrupting their children."

But these initiatives are just panderings; taken singly or as a package, they do not constitute a coherent program to arrest the declining character of American society. And this is what the American electorate hungers for: a thorough-going debate about how we are to restore the non-materialistic quality of life in America; a credible program by our government to rebuild our communities; an opportunity to regain hope about our future.

 One of Mr. Penn’s points of advice to fellow Democrats bodes ill for Vice President Gore’s 2000 campaign. With no apparent unease, Mr. Penn concedes Democrats need to embrace the era of smaller government, and his survey finds that "reinventing government" — a signature Gore initiative — is not what the public yearns for; they want government to be shrunk.

This is an important finding, but don’t over-interpret it. While most voters are disinterested in "user friendly" government, they will not be patient with reductions in benefits to achieve smaller government. The prevailing view is, "there is so much fat in the budget, in the form of waste and foreign aid, that it is unnecessary to cut the programs I like in order to reduce the size of government."

Finally, Republicans should ignore Mr. Penn’s advice on the matter of bipartisanship: "Clinton and Congress must continue to forge solutions in the vital center that address our most significant problems and are compatible with smaller government." This observation trivializes the magnitude of our current problems, the depth of the public’s ennui, and the role of leadership. We’re not at the Kennedy School here: in Washington, in the real world of politics, serious problems require not technocracy, but the resolution of serious underlying competing values. The more important the policy issue, the less likely it can be resolved with resort to bipartisan consensus. The American public burns today with indignation, and will not long be patient with Washington’s "fiddling" in the "vital center."

Interestingly, Mr. Greenberg, Mr. Penn’s predecessor as the President’s pollster, made the same mistake as Mr. Penn, advising Democratic congressional candidates in the summer of 1994 that legislative gridlock — what occurs when there is insufficient bipartisanship — was the Republicans’ Achilles heel. It was not

The Greenberg Thesis: Fidelity to that Ol’ Time Religion

Mr. Greenberg’s central assertion is that President Clinton’s victory resulted from his stands on education and Medicare, above all others. As evidence, he cites the results of an open-ended question, in which 59% said they voted for Clinton because of his support for domestic programs (education, Medicare, and the environment), versus 31% who voted for Clinton as the result of his "centrist" positions (moderation, welfare, balanced budget, crime). Readers should know that these are Mr. Greenberg’s characterizations, his distillation of a thousand individual responses. The distinction he draws between defender of programs and centrist is not likely an authentic distinction of popular perceptions.

The focus of his analysis is on the 10% of the electorate who are "new Clinton voters," and the 20% "new Presidential year voters" (Mr. Greenberg’s poll was bang-on this latter figure; the actual percentage of first-time voters was 19%). These voters — without whom, Mr. Greenberg concedes, Clinton would still have won — came overwhelmingly from lower middle income families. "It is the downscale, not the upscale electorate," he writes, "that gave the Democrats the opportunity to win in 1996." And among these "downscale" voters, the Clinton defense of entitlements, not centrism, was especially decisive. "The popular agenda does not encompass the kind of entitlement reform contemplated by elite opinion." But Greenberg further concedes Clinton owes a lot to another type of voter: "among swing voters who may have pushed Clinton’s margin upward in 1996, the centrist issues are somewhat more important, but still not as important as Clinton’s defense of domestic programs." This leaves the reader to wonder which constituency was really decisive to Clinton’s reelection.

Of course the answer is neither. Clinton was not elected by swing voters, first-time voters, or new Clinton voters. He was elected by the same folks who voted for him in 1992. This leaves Greenberg in the position of making a tendentious argument from pretty thin data.

Exit polls reveal the precisely the opposite occurred: the Clinton electorate became decidedly upscale in 1996, belying Greenberg’s thesis. Far and away the group which gave Clinton his biggest boost were households of $50+K income, awarding Clinton an additional 3.5 million voters — a far greater boost than seniors (+2.8 million), Catholics (+2.4 million), union households (+2.1 million), females (+2 million). And 2.2 million of that increase came from households with better than $75K income. Meanwhile, the Greenberg downscale Democrats actually decreased their vote for Clinton: Clinton received 2 million fewer votes from households with less than $15K income.

 The Penn data (above) suggests Mr. Greenberg is wrong regarding the public’s willingness to engage in serious entitlement reform. But the fact that Mr. Greenberg, in the face of all the available data, is able to maintain his confidence that the public really likes big government is an important lesson for Republicans. It means our confidence in the public’s antipathy to government may be a tad imprecise.

So long as the question posed to the public concerns government in the abstract, we are on solid ground. But our Democratic friends have a knack for recasting policy debates as referenda on the current level of benefits, and in this arena we are vulnerable. Take Medicare: the fact is, recipients of Medicare know more about the program than their representatives do. They see the waste first-hand; they know government administration drives up the costs of delivery. So they have concluded the current level of Medicare benefits can be delivered at a lower aggregate cost to the taxpayer. This is how to interpret the previous graph reporting public perceptions of the necessity for major reform in Medicare and Social Security. But woe to Republican representatives who think this survey finding means the public is resigned to reduced benefits at higher user fees — the public knows all too well that their sacrifices are not required.

The Mythology of the Gender Gap

Selwa Roosevelt wrote in a recent Washington Post op-ed that all of her female friends voted for Bill Clinton, owing to the GOP’s hostility to women. This pretty much sums up conventional thinking on the gender gap, that there occurred a mass exodus of women out of the Republican Party, into the arms of a female-friendly Bill Clinton.

There was, of course, no such exodus, and if Ms. Roosevelt’s friends all voted for Clinton, this says more about her friends than it does about the female voter in 1996.

Bob Dole received 1.9 million fewer votes from women versus Bush in 1992, a loss offset by a gain of 2 million male votes. But Dole’s losses were not across-the-board: he did well among women 60+, gained a few votes among women 45-59, lost 2 million votes among women 30-44, and lost 1.4 million among 18-29 females. Dole’s losses among the 18-29 year olds were about equal to Clinton’s, but Dole’s losses among the 30-44 year olds far exceeded Clinton’s losses. Also, while Dole was gaining a few 45-59 votes, Clinton was picking up 1.4 million votes from this cohort. Dole lost equally among married and unmarried women, but since there are 2 married female voters to every unmarried female voter, his losses among single women were more significant on a percentage basis. Clinton’s gains among unmarried women were far more significant than his gains among married women. In the end, the female groups among whom Clinton scored his gains were not those among whom Dole suffered his losses.

Meanwhile, 6.4 million fewer female voters turned out to vote, dwarfing the fluctuations between the candidates. Owing to this exodus, the relative importance of the female vote diminished dramatically. In 1992, 54% of voters were female; in 1996 52% were.

Clinton’s gains occurred among a particular and narrow group of female voters. Bill Clinton’s 1.6 million increase among unmarried females, coupled with his 1.4 million loss among younger women (under 45), indicates he received substantial increased support from widowed, divorced, or never married middle aged (45+) and senior females. These are not the soccer moms of lore.

It also looks as though all of President Clinton’s gains among women can be attributed to his gains among non-White, non-Black females (he lost 400K black female votes, gained 200K white females — but picked up 2.2 million voters elsewhere). The ethnic "other" category includes some Hispanics (many Hispanics self-identify as white or black), plus Asian-Americans, Native Americans, and mixed ethnicities.

In light of these dynamics, it would be wrong to take Clinton as an archetype of how to appeal to female voters. Six million women who voted for Ross Perot in 1992 were up-for-grabs; neither Clinton nor Dole were effective in their appeals to deter this mass migration.

The Aging of the Electorate and Other Trends

While the principal story of 1996 was the decline in turnout, several groups bucked the trend. Far and away the largest was the 60+ age cohort. Seniors contributed an additional 6.4 million voters, a growth which far outstripped the natural aging of the population. One factor behind this growth is that seniors were relatively less enthusiastic for Ross Perot in 1992, and the greatest turnout melt-down occurred among former Perot voters.

The growth in the senior vote occurred equally among men and women, and disproportionately benefited Bob Dole: Dole received 58% of this additional senior vote. Seniors are clearly not one-issue voters, but if their vote constituted a referendum on Republican Medicare reforms, the GOP won.

Meanwhile, as seniors were increasing their representation, youth turnout plummeted by 6.6 million. These dual trends resulted in a very substantial aging of the electorate. In 1992, 22% of the electorate was under 30, and 16% was 60+. In 1996, 17% was under 30, and 24% was 60+.

Another group which bucked the turnout down trend came from households with 75K+ income. An additional 3.7 million voters in 1996 are from such households, over half of whom voted for Bob Dole. Offsetting this growth was the loss of 6.5 million voters from households with less than $30K income (roughly the median family income in the U.S.). Bill Clinton picked-up an additional 2.2 million votes from $75K+ households, and lost 1.1 million from below median income households. So both the electorate as a whole and the Clinton vote became significantly more affluent between 1992 and 1996.

The final important dynamic was the growth in the ethnic vote. Eleven million fewer self-identified whites voted in 1996, while 3 million additional Hispanic and African-American voters turned-out. Most of these additional voters went for Bill Clinton, but Dole also participated in this growth, gaining 500,000 votes versus 2.6 million for Clinton. Among new voters in certain ethnic subgroups — younger Blacks, Western Blacks, Midwestern Blacks — Dole actually bested the President.

While the distribution of the African-American vote did not shift much, Hispanics became decidedly more Democratic this election — perhaps Clinton’s only enduring structural success (Hispanic voters are 5% of the electorate). And while Asian-American voters did not improve their participation as did Hispanic voters, the shift away from the GOP was noticeable (Dole lost and Clinton gained 100K Asian-American voters over 1992). These two migrations may be the legacy of Republican positionality on immigration and alien issues.

Other than Hispanics and Union households — the latter of which did not on the margin go disproportionately for Clinton — 1996 saw the erosion of all traditional "core" Democratic voter groups. Liberals and Moderates together declined by 9.4 million, Democrats declined by 2.2 million, those with a high school degree or less declined by 3.5 million, Jewish voters declined by 1.3 million (one of the hardest hit of all groups as a percentage of turnout), voters from below median family income households declined by 6.5 million. This litany catalogues real damage to the bedrock Democratic vote, and a impediment to Democrats aspiring to challenge Republican Members of Congress in 1998.

Even in the face of the Dole defeat, the same cannot be said of the Republican Party vote. Conservatives increased their presence in the electorate by 1.1 million, and Dole won among the marginal or "new" vote of every demographic group which experienced a substantial growth in turnout, except for Hispanics. Dole just did not win by enough among these groups.

Clinton has probably established the high water mark of the "security" agenda (what Republicans uncharitably call the nanny state or government in loco parentis), and it is less than 50%. From here, Republicans have more apparent avenues for electoral growth than do Democrats. The 1996 Perot voters — 8 million strong and looking for a new candidate in 2000 — largely share with Republican voters a critique of American society in decay, as do many if not most of those who voted in 1992 but not in 1996.