Media bias is becoming quite the topic these days. As a matter of fact, media bias is supported by polling bias, and the 2008 election might bear that out. This is my own investigative report in brief, and is an opinion piece. It is not fully referenced, but may serve as a starting point for your own investigation. Please note that if some of the specific current events quoted as examples fall slightly outside of the sample poll's time frame, then any other objective event could be substituted to create an example of an objective, or less "loaded", question.
A Pew Research survey report of March 27, 1998, shows how pollsters can use subtle variations on questions to produce biased poll results. Here, The Pew Research Center did a survey to find out if polls actually are biased against Republicans. They determined that no such bias exists, but that (according to the Pew article) racial hostility by Republicans was missed.* However, look at the two questions that are cited in this article as having been posed merely to determine whether both groups of respondents "were equally well-informed about current events":
"Equal numbers in each sample could correctly identify Microsoft CEO Bill Gates, knew which party has a majority in the U.S. House of Representatives and knew that Bob Dole had recently loaned Newt Gingrich money to help pay off the House Speaker's ethics fine." [The Pew Research Center, March 27, 1998: http://people-press.org/report/?pageid=589 Link to Pew Research article.]
Note the political objectivity of the first question, as to whether both groups could identify Microsoft CEO Bill Gates. The second question, though referencing Democrats, is fairly objective in nature ~ though slightly positive since it refers to a "majority," which might have positive implication. Now note the negative implication of the data contained in the question that references Republican Newt Gingrich (exact wording not available). The question itself should be an objective question aimed at determining whether people are informed about current events; however, data contained within the question actually informs the respondent that Gingrich has had an ethics violation, and that Dole has paid to cover it. In other words, in a poll aimed at determining whether polls are biased against Republicans, and in a question aimed at determining the degree of knowledge of current events, pollsters have chosen a "loaded" question that contains negative information on a Republican, rather than an objective question. The negative information contained within the question could actually influence a respondent's opinion of Republicans in a negative way. One of the most damaging aspects of biased polling, if it is a trend, is this: if such biased questions are included as a matter of routine in the non-topical subsections of the poll ~ where the intent is merely "controls" to check the knowledge of the respondent instead of to gather subject data; and if a number of polls are done over a period of time; then the poll itself can be used to influence respondents and produce changed results in future polls. The copious number of political polls being used in the 2008 presidential election, if utilizing similarly "loaded" questions, could actually serve to influence respondents to vote a particular way. The pollster can claim to have been balanced in that they used one question referencing Democrats and one question referencing Republicans. However, the content of the questions bears a strong bias against Republicans.
Other current events of the same era, which might have been used to determine the respondent's knowledge of current events, included: the O.J. Simpson civil trial; the Timothy McVeigh (Oklahoma bomber) conviction and sentencing; the deaths, in skiing accidents, of both former entertainer, Republican congressman Sonny Bono and Democratic campaign manager, Michael Kennedy (son of Robert F. Kennedy). A question about any one of those events, or a combination of questions, would be less "loaded" than the question about the Gingrich ethics violation. Ironically, the most prominent current-events story of that era ~ one that would put Democrats in a bad light ~ was the Clinton-Lewinski sex scandal. That topic went unmentioned in this article (and probably in the poll).**
I believe that this same (or a very similar poll) was conducted again recently. I was a respondent in that poll, and noticed that the one current-events-oriented question that was asked of me was also weighted negatively against Republicans, as in the previously cited poll. A "loaded" question of this nature serves to inform the respondent of the negative event ~ which is not the implied purpose of the question. The ostensible intent of the question was simply to test respondent knowledge of current events, not to inform the respondent or include multiple questions weighted toward each party in a bi-partisan format. At the time that I answered the poll, I recognized that bias and was irritated by the question. All other questions in the same subsection were objective ones, such as questions calling upon the respondent to identify the majority house speaker and the minority house speaker. As a former survey-research interviewer, I recognized the question as being a biased question in an otherwise non-partisan subsection of the questionnaire.
*Some of the data and data analysis used by the pollsters to determine racial hostility is itself subjective in nature and could not be properly analyzed without more data to determine the facts of the subject within the context of stereotypes and whether specific stereotypes are true or false. (For instance, how much of minority poverty is caused by prejudicial attitudes in society, and how much of it is actually the result of the "self-fulfilling prophecy" syndrome caused by propaganda aimed at promoting minority status? A separate, objective survey would be needed to expand the topic in question.)
**I find this current event of the era particularly interesting: Former Democrat-turned-Republican Senator Trent Lott warned Congress of policy and leadership problems with the International Monetary Fund. The IMF had recently used over $100 billion to bail out Indonesia, South Korea, and Thailand. The Clinton administration asked Congress that week for an extra $3.5 billion "for an emergency IMF lending facilty and around $15 billion to cover the U.S. share of a quota increase to replenish IMF resources...drained by the three bailouts." However, in all fairness, the questionnaire used in the Pew Research poll may have already been compiled at the time of this event. [Source: Reuters: http://www.cnn.com/ALLPOLITICS/1998/01/23/reuters/imf/ Friday Jan. 23, 1998. Link to Reuters article on CNN.]
Southern Muse: copyright (2000, 2001, 2002, 2003, 2004, 2005, 2006, 2007, 2008 to present day) by D.K. Pritchett. All rights reserved. The URI of Southern Muse's domain is http://www.southernmuse.com.