I am working to get a copy of the poll memo to analyze, but the Post is reporting that an internal poll from Garin-Hart-Yang for the Van Hollen campaign has him leading Donna Edwards by a margin of 45-40. Here’s the story, more hopefully to come after I get a chance to review the poll memo.
Van Hollen garnered 45 percent to Edwards’s 40 percent among likely Democratic primary voters, a margin that is not statistically significant but differs from recent independent polls showing the race nearly tied or favoring Edwards.
* * *
Both candidates are viewed positively in the newly released survey, which comes less than four weeks before the state’s April 26 primary. But Van Hollen again comes out slightly ahead, with a 60-percent positive rating to Edwards’s 53 percent. Van Hollen also leads in the key Baltimore media market, according to the poll, which includes both the city and surrounding suburbs.
* * *
The firm Garin-Hart-Yang conducted the poll for the Van Hollen campaign by surveying 604 likely Democratic primary voters on a mix of cellphones and landlines from March 28 to 30, including 294 in the Baltimore area. The margin of sampling error was plus or minus 4.1 percentage points for overall results.
Note to Rachel Weiner: a single survey result, standing alone, cannot be said to be (or not be) “statistically significant.” It’s only when comparing one poll result to another that statistical significance has any meaning at all. The more commonly used phrase “margin of error,” on the other hand, is an attempt to measure the fact that an entire population (Maryland registered Democratic voters, for a purely hypothetical example) is not being measured, but only a small, representative sample of the larger pool. MOE is used with respect to individual surveys, and is lately based on the number of survey respondents – the more respondents, the lower the margin of error. MOE is also used to calculate and assess statistical significance when comparing different poll results. See here for even more super nerdy analysis on these two concepts.