What’s with all the fives? In five minutes (ahem) I wrote down the references I could find or think of that had a “five” in them:
- Your five-day weather forecast.
- 95% confidence interval
- 25 years of age to enter the general risk pool for auto insurance
- 95% of donations go directly to our clients
- High-risk period for motorcycle riders is 0.5 year experience or less
- (In many cases) 5 years after being cancer free, for your doctor to tell you’ve re-entered the general cancer risk pool
- 15 years after being cancer free for your insurance company to agree with your doctor.
- Luther’s 95 theses
- 5th, 25th, 50th, 75th, 95th percentiles
- 50-50 chance
- 50 goals in 50 games (hockey)
- 50th anniversary
- 95% success rate
- Five books in the Torah, five Islamic Pillars, five wounds of Jesus Christ
- A perfect fifth (chord)
- Slaughterhouse Five
- Five-year plan
- Five Olympic rings
We have a predilection for the number five. In some cases the importance is technical or symbolic, as in a 50-50 chance or the five Islamic Pillars.
However, if you look at these and think “Some of this seems arbitrary,” I’d agree. You’re right to be suspicious. If an answer looks arbitrary – for any reason – the underlying question may be arbitrary, or something other than what it should be. Or, the answer itself could simply be made up.
Are we just going to let this happen? I think not… When we control our questions, we’re a long way towards controlling our question-and-answer process. And the best weapon against poorly-formed questions and fraudulent answers is inquiry itself. To get started, we can just turn these answers on their heads:
- “Why 95% confidence interval, and not some other number?”
- “What’s with the five-day forecast, when today’s forecast isn’t right?”
- “Is it really five years? Why does my insurance company insist on 15?”
- “Why 25 years? Why not 23, or 26?”
- “Do you really give 95% of my donated money to your clients?”
- “What’s so special about 75th percentile? Would the 82nd be just as good?”
Questions like this probe rarely-evaluated assumptions and conventions, so the initial answers may be heavily laden with intellectual baggage. We should insist on a good English-language answer, other than appeal to a higher authority. These proceedings unfold within the temporal plane.
Ultimately we’ll learn something. Perhaps the answer and its associated question was arbitrary, or a different question entirely was being answered, or the answer itself was arbitrary. Maybe everything was OK and the answer really was 50% (it happens). All of those outcomes help us understand of what we’re asking, and its support in our data.
Personally, I don’t think we can pursue this enough. (It’s a great topic at a weekly Q&A context meeting.) Anyone can participate and bring a relevant issue to the table. (Digging up the answers can be a technical endeavor.) Besides, there is something about rooting out the unneeded conventions and assumptions that can inhabit our questions. It’s well, fun.