Pollster.com

Michael McDonald

Department of Irony: Markos Allows Himself to be Influenced by Rasmussen

Topics: Approval Ratings , House Effects , Obama

Recently, Markos Moulitas of DailyKos questioned if Rasmussen's polling was designed more to set a narrative than to accurately predict election outcomes.

I was thus highly amused by his appearance as a guest on ABC's This Week on June 6. Commenting on Obama's reactions to BP's oil spill, Markos stated: "I mean I don't think there's any doubt that the polling is slipping for Obama."

Really? Let's suppose that Markos is using Pollster's trend line averages as the evidence for this statement. The most current trend line as I write this on Sunday, June 6 does indeed show an apparent recent tick down in Obama's approval rating.
Obama_approval_6_6_10.pngHowever, as I have pointed out previously, movements in Obama's approval rating can be sensitive to the ubiquitous polling of Gallup, YouGov/Polimetrix and - you guessed it - Rasmussen. For whatever reasons - and there are potentially many - these pollsters tend to be the least favorable to Obama. I do not want to argue about the accuracy of these polls, I'll leave that to others. It is sufficient to note that these pollsters poll often and tend to be the least favorable to Obama and Democratic candidates.

The other national pollsters often run their polls on a regular monthly or so schedule, absent any major event to stimulate them to run an additional survey. At times, this causes lulls in the polling among these national pollsters. When a lull occurs, the Pollster trend line becomes dominated by the frequent pollsters and tends to move in a negative direction.

We are in one of these lulls right now. And as a consequence, the Obama trend line has moved in a negative direction. You can see this by using the "Tools" to remove Gallup, YouGov/Polimetrix and Rasmussen from Pollster's Obama job approval plot.

Obama_approval_6_6_10_modified.png

In fact, if you look closely at the recent polling, Obama's approval rating from all three of these polls has recently trended higher.

Obama Approval
Rasmussen: 5/23-25/10 (43%), 6/1-3/10 (48%)
Gallup: 5/25-27/10 (45%), 6/1-3/10 (48%)
YouGov/Polimetrix: 5/22-25/10 (44%), 5/29-6/1/10 (45%)

If these pollsters didn't consistently have a lower job approval rating for Obama compared to the other national pollsters, Obama's job approval trend line would actually be increasing!

The punch line is that Markos allowed himself to be influenced by what he claims to be a Rasmussen narrative. And the moral is that care should be taken in interpreting the Pollster trend lines.


Multi-Partyism in American Politics?


Is America on the verge of a European-style multi-party democracy? A May 12th Wall-Street Journal/NBC poll finds that 31% of American adults' view on the two-party system is that, "The two-party system is seriously broken, and the country needs a third party." This sentiment is consistent with my analysis of partisan voter registration, which shows a slight rise in the number of people who are eschewing the major political parties to register with minor political parties or affiliate with no political party.

What is the cause of discontent with the major political parties? The most accessible answer is that it is a product of the times. Voters' attitudes are tied to the economy and they are expressing their displeasure with those in power through multiple measures of low trust in government, low approval of the political parties, and a desire for an alternative.

The economy is likely a major factor for voter anger towards the parties, but there are long-term historical trends that also shed light as to why minor parties may be poised for modest electoral success. In the figure below, I plot from 1870-2006 the effective number of parties elected to the US House (black line) along with a measure of the ideological cohesiveness of the two major party's caucuses (blue and red lines). McDonald Effective Number of US House Parties.png

Source: Michael P. McDonald. 2009. "Mechanical Effects of Duverger's Law in the USA." In Duverger's Law of Plurality Voting: The Logic of Party Competition in Canada, India, the United Kingdom and the United States, Bernard Grofman, André Blais and Shaun Bowler, eds. New York, NY: Springer.

The effective number of political parties weights the number of parties by their relative strength. When two major parties hold all the seats and are near parity, the effective number of parties is close to two. When one party dominates the other, it is less than two. When minor parties win seats, it is greater than two. For it to be close to three, essentially at least three parties must be near parity.

Since the American Civil War, the effective number of political parties in the US House has generally only been substantially greater than two in the period before 1920. Two significant minor party movements during this period, the Populist and Progressive movements, account for the minor party candidates elected during this period. Indeed, the success of the minor parties during the era is understated, as major party candidates would often run "fusion" campaigns with minor parties, by running under such labels as the "Democrat Populist" and "Republican Progressive" parties.

What is also distinctive during the period of minor party success is the ideological cohesiveness of the major political parties. I analyze the ideology of US House members using a widely-used measure by congressional scholars is known as DW-NOMINATE scores, which identify the ideology of members from the votes they make in Congress. A measure of ideological homogeneity is the standard deviation of these ideology measures. When it is low, the congressional parties are more ideologically cohesive.

Minor parties have greater electoral success when the major parties are more ideologically cohesive. If the relationship is not visually apparent, the correlation coefficient between the effective number of US House parties and the standard deviation of the Republican ideological voting scores is -0.39, a strong relationship. The correlation for the Democrats is a weaker -0.20. This is due to the greater ideological dispersion of the congressional Democrats when the conservative Southern and liberal non-Southern wings of the party existed in the uneasy New Deal coalition formed by FDR during the Great Depression, and which began fragmenting during the Civil Rights movement of the 1960s. Since 1982, the congressional parties have again become more ideologically cohesive.

Conventional wisdom in political science has until recently been that single-member district electoral systems, like that used in the United States, tend to foster a party system composed of two major political parties. Recently, there has been a resurgence of minor political parties in democratic countries that use single-member districts systems like Canada, India, and the United Kingdom. The United States has been resistant to this trend. One reason may be that the American political parties are like willows in the wind. They are able to bend to local conditions within districts to effectively squeeze out minor party candidates.

As American parties become more ideologically rigid, more space is provided to minor party candidates to flourish. If so, then we may be on the verge of at least some electoral success for minor political parties. However, this success is most likely to be fleeting. Since the dissolution of the Whig Party prior to the Civil War, the major parties have proven themselves quite capable of absorbing these minor parties into their electoral coalitions.


Voter Registration Trends, May 2010


Trends in party voter registration since the 2008 presidential election suggest that a small, but perhaps meaningful, number of registered voters are abandoning the major political parties in favor of minor political parties or are forswearing any party affiliation.

Twenty-eight states, plus the District of Columbia, allow persons to register with a political party. Among these states, the reported number of registered voters has declined by 2.6 million or 2.6% since the November 4, 2008 presidential election. This decline is expected. Election administrators remove people who have moved from their address or are otherwise no longer eligible. Absent an interesting election to stimulate new registrations, the voter rolls are not replenished as fast as they are purged of these defunct registrations.

McDonald National Party Registration Table.jpg
The Democratic Party has experienced a slightly greater absolute loss in the number of party registrations, at 1.2 million, compared to the Republican Party, at 1.1 million. However, since there are considerably more registered Democrats (a partial consequence of the universe of states that permit party voter registration, see the state numbers below) the percentage loss since 2008 of 3.5% is greater for Republicans than the 2.7% loss for the Democrats.

The number of persons registering with a minor party is actually increasing, by 52,810 or 2.4%. Further, although the number of those unaffiliated with a political party is decreasing, the pace of this decrease of 389K or 1.6% is less than that of the major political parties.

America is a long way from having a viable multi-party system at the federal level, like we are currently witnessing in the United Kingdom. However, these trends are consistent with the notion that some American voters are willing to express their frustration with the major parties by registering with a minor political party or affiliating with no party. Indeed, the increase in unaffiliated registrations is a long-term phenomenon observed since the 1970s.

Now, it should be noted that those who self-identify as "unaffiliated" tend to align themselves with a political party at the polling booth. But, this begs a question that scholars of political behavior have not adequately addressed, which is why these independents act like partisans but do not want to associate themselves with a party. Registration-based sample surveys could be used to address this by asking probing question of unaffiliated registrants. I hope that this is a research agenda someone would be interested in exploring.

Returning to the increase in minor party registrations, as discussed in Paul Herrnson and John C. Green's edited volume Multiparty Politics in America, people who identify with minor political parties tend to be more sophisticated than those who are unaffiliated with any political party. These people tend vote and volunteer for campaigns more often than the general public. Their absence from the major political parties may adversely affect major candidate campaigns, particularly where a minor party candidate is on the ballot.

This is not simply a Tea Party movement. There are a number of different minor political parties that range across the entire ideological spectrum. For example, in Maine the only state-recognized minor party is the Green Party, which has seen an increase of 8,790 or 34.1% since the 2008 presidential election. In North Carolina the only state-recognized minor party is the Libertarian Party, which has seen an increase of 3,685 or 101.3%. Maryland may demonstrate how this trend is an expression of frustration. The increase of 21,167 or 29.2% is entirely among the 23,897 new registrations with the Maryland Independent Party. A check of the Maryland Independent Party website shows little activity to account for a grassroots groundswell that trebled the party's support.

There are a number of interesting trends worthy of mention among the state statistics. In the two states where voter registration increased, Delaware and Colorado, only the number of registered Republicans decreased. The Colorado trend may be worthy to watch in the coming months since statewide elections are expected to be competitive. The only states where Republican registration increased are Louisiana and New Jersey. In 2009, New Jersey Republican Gov. Christie won a closely-watched election and Louisiana held some state and local elections. This suggests that Republican registration may yet recover during the fall 2010 elections as the campaigns gear up.

McDonald State Party Registration Table.jpg
Now, for some data notes. In the state table, I note the date of the most recent voter registration report from each state. In some cases these reports are quite recent, while for a few the last report may have been for a fall 2009 election. I report active plus inactive registration statistics, where available. Inactive registrations are people who have not voted recently, but because they are registered to vote, I have chosen to include them. There are eight states that do not have statistics for minor party registration, because the state did not provide separate statistics for minor party and unaffiliated registrants in 2008 or their current report. Among these states, the unaffiliated registration tend to be either increasing or is not experiencing as steep of a decline as the major parties, suggesting that minor party activity is hidden within these numbers.

Finally, some caveats. Partisan registration is not a perfect measure of the state of national partisanship for several reasons. We cannot know from these statistics what is happening in the twenty-two states that do not have party registration. People who identify themselves with a political party in surveys may not register with a party because they do not intend to vote in primaries. They may wish to vote in another party's primary because that party dominates in the general election and they want to cast a more meaningful vote in that party's primary. They may not run out and change their party registration whenever they change their party self-identification. Despite these limitations, none satisfactory explain why we would observe an increase in minor party registrations and, in my opinion, do not adequately explain why unaffiliated registrations would show less of a decline than major party registrations.


 

MAP - US, AL, AK, AZ, AR, CA, CO, CT, DE, FL, GA, HI, ID, IL, IN, IA, KS, KY, LA, ME, MD, MA, MI, MN, MS, MO, MT, NE, NV, NH, NJ, NM, NY, NC, ND, OH, OK, OR, PA, RI, SC, SD, TN, TX, UT, VT, VA, WA, WV, WI, WY, PR