Rhetorica: Press-Politics Journal

Gallup has released information that challenges the predictive model of primary elections by Dr. William Mayer. I have sent Mayer e-mail asking him to take a look at the release. I will also be comparing it closely with the Mayer essay that I have relied on to make my contentions about the press-politics of primary campaigns.

The Gallup release considers a much longer period of time than the Mayer data because Mayer’s purpose is to study the effects of the McGovern-Fraser reforms since 1980. Further, Mayer includes data from two elections–1980 and 1996–that the Gallup release omits. Mayer’s work, based on his intentions, would have been incomplete had he left out these dates.

I am attempting to account for the apparent differences now. And I’ll post the results of my analysis as soon as possible. If I must make corrections to the work I’ve done so far on Rhetorica, I will do so publicly and immediately.

UPDATE (6:45 p.m.): Earlier today I sent e-mail to Dr. William Mayer alerting him to the Gallup article. I have not heard back from him yet. If/when I do, I will report what he tells me here.

I have reviewed the data and compared it to the Gallup article. I find the discrepancies may be explained this way:

1- Gallup omits data from 1980 and 1996 because incumbents were running nearly uncontested campaigns. The Mayer model treats all years and candidates from 1980 to 2000 equally.

2- Gallup considers election years not included in Mayer’s model. Mayer is studying the effects of the McGovern-Fraser reforms.

3- Mayer’s data and the Gallup data may be from different polls. Here is what Mayer says in his recent article in Political Science & Politics (April 2003, citation here, emphasis mine):

“My interest here is on the last poll before the start of delegate selection activities–meaning, in most years, the last poll before the Iowa caucuses. As shown in the top half of Table 1, the candidate who was leading in this poll went on to win the nomination in five of the six cases.”

The table Mayer refers to considers the elections from 1980 to 1992. The anomaly is Hart in 1988 for reasons we may attribute to the Donna Rice affair. Mayer’s data from 1996 and 2000 conforms to the data given by Gallup in today’s article. The divergence, then, is in the data prior to 1980 and the data for 1992. Here’s what the Gallup article says:

Although then-Arkansas Gov. Bill Clinton had by January 1992 risen from relative obscurity to the point where he was in second place in Gallup’s poll of Democrats around the nation, he was still running behind former California Gov. Jerry Brown. Only later did he move to the top of the national poll’s list.

It is not at all clear that Mayer and Gallup are referring to the same poll. Further, some of the data in the Gallup article clearly comes well before the Iowa caucuses. So we are comparing apples and oranges.

Mayer is specific about which poll is crucial to his model. If Gallup is attempting to challenge Mayer, then an attempt should be made to match the polling data.

I am satisfied that Mayer, a Harvard Ph.D., is quite capable of reading a poll. Further, his research has been used by political scientists, without challenge to the data, since 1996. You may check the data for yourself. Here are the citations for Mayer’s earlier work:

Mayer, William G. 1996. “Forecasting Presidential Nominations.” In Pursuit of the White House: How We Select Our Presidential Nominees. ed. William G. Mayer. Chatham, NJ: Chatham House.

____. 1997. “The Presidential Nominations.” The Election of 1996: Reports and Interpretations. ed. Gerald M. Pomper. Chatham NJ: Chatham House.

____. 2001. “The Presidential Nominations.” The Election of 2000: Reports and Interpretations. New York: Chatham House.