Sunday, June 27, 2010

Q & A

The following is a message I received via Facebook. Identifying information has been removed.

I work at one of the Twin Cities Hospitals. Yes, I voted to strike. If I posted the question I am going to ask you on the MNA page I would get my head bitten off.....Out of 12,000 eligible nurse votes, less than 9,000 voted, something like 8700.Out of those 8700 (less than 75% of the nurses) 87% voted to strike. That is only 63% of Twin Cities nurses voting to strike. How is that overwhelming?!

My Answer:

If this was a survey, you would say you had a voluntary sample of 8700 out of a population of 12000 (roughly 73%). Although the Office of Management and Budget requires a sample size of >80% for surveys to be used for government policy-making, >70% response is considered adequate for most academic purposes, and reasonably reflective of the studied population. 63% of the population is still significant, beyond any margin of error, and would be a filibuster-proof majority in the US Senate. Using the same math, you can also say that 9.5% of the MNA nurses voted against the strike. In that 3300 members didn't vote, the actual support of the strike could be anywhere from 63% to 90.5%. We'll never know.
However, this was not a survey; it was a vote. That is, it's a way for us to contribute to a decision-making process, to directly instruct our bargaining teams in what we want them to do. Thus, only those who participate matter. A 73% turn-out for any election is well above the norm, although more would have been better.
I, also, despair that someone with a sincere question or declines to see things in black and white is ridiiculed, viewed with suspicion, or both. It may be the way things are in situations like this, but that doesn't mean I have to like it or support it.

Perhaps it's time to set aside the torches and pitchforks, and at least for now, call off the witch hunt. Such scourges against those lacking "ideological purity" always do more harm than good.


  1. Isn't it also fair to suggest that based on a population of 8200 you can extrapolate the remainder of the population that did not vote would have a similar pattern baring any unusual confounders.

  2. Yes, were, this a survey, a >70% response rate would likely correlate well with the overall opinions of the studied population. If demographic data was included, the case could be even further strengthened.
    While that's an important consideration and supports assertions concerning the strength of the results, there are important differences between a survey and a vote.
    Because it's a vote, only those who show up count. Even if the turnout was 10% and the resolution won by 55%, it would have still won (unless there are MNA bylaws that require a certain turnout and percentage). So while the turnout and percentages are politically valuable, they don't affect the validity of the outcome.