Wanted: A Poll with a Real Sample

Ridenbaugh has the latest on a Greg Smith poll that has Larry Grant up over Bill Sali by 22-14% with Undecided at 64%.

I’m skeptical of this poll and I was skeptical of the last one. The last poll had a margin of error of +/- 6%. This poll regarding the Congressional race has almost unheard of Margin of error of +/- 7.4%.

What does that mean? Well, according to Smith, Bill Sali could really be ahead 6.8% or down by 23%. A good margin of error would be +/- 4% which would mean with an 8% Grant lead, it could be tied, or Grant could be up by 16. That’s a more tolerable variance.

The last poll used a sample of 400 voters for the whole State of Idaho and 218 of those were from the 1st district. This one used 25% less voters, so assuming he took the same larger percentage of 1st district voters, we probably end up with these numbers being based on answers from 163 or less Idaho 1st District residents.

In addition, the time frame of the poll may have something to do with it. The timeframe of the poll was from August 28th-September 1st-the week before Labor Day when a lot of people are away or getting kids back to school. It’s also curious that a five day poll only produced 300 results.

I’m sorry, but that doesn’t really create a relaible sample. This is a fluke poll. I’ll wait until we get some more solid polling in this race. I want to see something with around 600-700 voters statewide and 300-400 in the Congressional District. Something with a margin of error that’s not so wide as to make it absurd.

—–