http://www.wvwnews.net/story.php?id=3948
by Bethany Stotts
Are young American voters becoming increasingly http://www.wvwnews.net/story.php?id=4107? That’s what Campus Progress, a liberal activist group, is arguing in their newest study, “The Progressive Generation.” The reports’ authors, David Madland and Amanda Logan, base their analyses on the General Social Survey (GSS), the National Election Study (NES) and Pew Research Center data.
Defining the “http://www.wvwnews.net/story.php?id=3575” as Americans age 18-29, they conclude that this group:
– believes the government should ensure them good jobs and fair wages (45%);
– believes that the government should provide more services (61%);
– is less likely to perceive the government as wasteful (51%).
In addition, they connect this generation to a higher appreciation for government services.“On many economic issues, Millennials are more progressive than any previous generation,” they write, claiming that this new demographic will likely keep its progressive values as it ages and shift the future political climate toward more socialist practices. They dismiss the idea that voters become more conservative as they age as “an overly simplified and often incorrect assumption,” and little more than a “stereotype.”
The problem is, Logan and Madland’s analysis is based on shoddy methodology and their NES data has a very small sample size of around 225 voters. When Accuracy in Academia asked Scott Rasmussen whether a sample size between 150 and 350 was scientifically sound, he responded “Practically speaking, I would never do that.” Rasmussen is the President of RasmussenReports, a respected public opinion polling firm and the top polling site for both the 2004 and 2006 election cycles.
AIA has identified the following errors within the CP report:
– consolidating categories to exaggerate results,
– omitting pertinent information,
– claiming trends where none exist,
– standard error as high as 14.6%
– inconsistencies in measuring,
– not actually providing the data upon which conclusions are based,
– not listing the confidence level.
Most errors spanned five of the eight categories and can be recognized by those with only a basic background in research methods. Yet the authors claim grand significance for their questionable findings, arguing that the survey results “clearly indicate the strong progressive leanings of Millennials today, and the likelihood these trends will endure over the coming decades.”
While our critique is restricted to the study’s extensive use of NES data, Campus Progress’ touting of faulty data based on non-representative samples calls into question its scholarship and professional ethics.
Our Analysis
Statistics is not an exact science. In fact, polls are often estimates. Statisticians have developed a concept called “margin of error” which indicates the range within which study results are considered reliable. For example, if a poll places President Bush’s support at 40% but indicates a standard error of 3%, then Bush’s approval rating likely lies between 37% and 43%. (Double the standard error to calculate the margin of error).
AIA’s analysis used a 99th percentile confidence threshold—a stringent scientific standard to ensure that results will be accurate 99 out of 100 times.
http://www.campusreportonline.net/main/articles.php?id=2327