Friday, June 26, 2009

No Such Thing As A Simple Poll

I conducted a poll in my office yesterday to select a name/codename for a project. We had 5 options and 13 people on the electoral roll. The idea was to come up with the best name which was agreeable to the most number of people. Given the small numbers involved and the intelligence and absence of malevolence in the population, I thought it would be a simple enough decision to be made democratically. Boy, was I wrong!

The choice was to be made from a set of acronyms:
* These names were thought up by a smaller group from different perspectives on the planned project.
* All of them were acronyms with cheesy expansions - explanatory notes were provided along with the options.

We had to guard against the situation where less people thought that the chosen one was inappropriate (there is a name for this, forgot, couldn't find it on wiki), so I defined some rules aimed to prevent this scenario.

The rules were as follows:
1. It would be an open vote (no anonymity) and had to be sent in by email to a designated pollster(me) before a cut-off time.
The vote was an open one because people could possibly behave irresponsibly when given anonymity.

2. People could vote for the 2 options that they liked the best and rank those 2.
Two choices were allowed to factor in groupthink - people voting for something other than what they truly want because they think the group would want another one.

3. Formula for Calculation of the NAI (Name Appropriateness Index) for was as:
NAI = N(FirstChoice) + 0.5 * N(SecondChoice)
Weightage for the second choice was to respect the decision-making process. In retrospect, maybe that factor should have been a little higher.

4. The option with the highest NAI would be chosen.

5. Ties on NAI would be resolved by an Honorary Adjudicator.
Repolling would be invalid because of the high influence of groupthink. Plus, the title Honorary would incline the Adjudicator to make a responsible choice!

However, it turned out to be a typically microcosmic model of more complex democratic polls
1) A clear winner - There was no tie
2) One abstainer - Hated all the names
3) One absentee - Was not reachable
4) One postal vote - Over the phone
5) One invalid vote - Voted only for 1 option

The winner turned out to be a sub-optimal one and we had to have an intervention from our boss! It left me wondering how sub-optimal the results for a naiver scheme for a larger and less informed population, had to be!

No comments:

Post a Comment