Apologies
Before I start, I apologise for this post:
- For those of you of a six sigma nature I apologise, you might find this a bit Janet and John.
- For those of you of a non six sigma nature, I apologise for your six sigma brethren. They really ought to have told you this, it is quite important.
In God we trust, all others bring data
That is a quote from the quality guru W.E. Deming. What did he mean? If you want to be sure what is going on get some data, don’t rely on opinions and hunches, get the numbers.
How do you get the data? I bet you were taught the easiest way when you were at primary school, to use a tally chart or 5 bar gate. If that is rather longer ago then you would care to admit to, here is a short refresher course:
Stage 1. Think about what you are going to count
Be clear why you are counting
Some people count things for pleasure. They are often seen standing at the end of railway platforms counting trains. (I used to do this when I was 14, let me assure you it is better for your social standing if you have a sound reason for counting things.)
If, however, you work for the Highways Agency, counting the number of vehicles that are using a road is perfectly acceptable. But even the guys in yellow jackets need to have a hypothesis that they want to prove, or question that they want to answer.
- Are they worried that a car park is too small?
- Are they trying to prevent accidents at a black spot?
- Are they thinking of investing in a bypass?
Being clear what the problem that you want to solve is, is all important. As with most things in life it is better to think a bit about what you are trying to achieve before you start.
Be clear what you are counting
The minute you get into serious counting you are going to need some help, somebody who is going to count with you. And the minute you get help is the minute you start to have communication problems.
Technically speaking what you need is operational definition. It might sound a little obvious but if you plan for somebody else to count with you a little definition is all important.
- Is that a lorry or a juggernaut?
- Does that MPV thing count as a car or a van?
Write a list of the different categories and make sure that everybody is clear what goes into each category.
- Try to keep your list of categories short so that whoever is doing the tally can’t mess it up. If you have more than 10 categories then they will mess it up (I promise)
- Always have an “other” category. Even if your categories are male and female, you never know what you might be missing, it is a big world.
- Discuss the categories with everybody who is counting so they are absolutely clear.
If you end up with lots and lots of “other” that is a good clue that something has gone awry in your categorisation, time to have a rethink.
Count facts not opinions
If you want to know why there are so many traffic jams count vehicles, speeds, sizes or obstacles, not opinions. People might well think that the reason for the traffic jam is: inept councillors, stupid neighbours or an erratic lollipop lady, but now you have strayed into the realms of root cause analysis.
Only count opinions where opinions count, maybe before an election, otherwise stick to the facts.
Stage 2. Count
How long do you need to count for, how much data do you need?
Pragmatism is a virtue
If you want to know how busy a road is you could:
- Look at it for 10 minutes
- Stand by the road side from 4:30pm to 5:30pm every day for a week
- Buy a tent and invest 24 hours a day, 7 days a week for 4 weeks wearing your best anorak
It is fairly obvious which is the pragmatic answer.
Sometimes you need a little more science
There are statistical models that work out how much data you need. The sad truth though is that many of your stakeholders won’t understand them (when pushed I will admit to being a six sigma black belt and I find them a bit of a struggle).
If you must use them then there are 2 things to think about:
- How easy is it to spot the thing you are looking for? Do you want to know the proportion of lorries on the road, or Rolls Royces?
- How sure do you need to be in the answer? Are you having a quick look-see or are you investing millions off the back of the data?
Think about the answers to those questions and agree how much counting you need to do. (Though if the answer to the last question is millions I suggest you get your stats book out)
Stage 3: Analyse the data
Don’t mix and match
The traffic outside your local cinema on Saturday night is very different to that on the nearest lorry park on Tuesday morning. It is fine to compare the two, but it is probably not so good to add them up and average them out.
Show the when and where and what you are counting. It will avoid confusion.
Be clear in your presentation
Present your facts in a clear unambiguous way that ties back to your problem. State your result
“Lorry numbers suggest that we should invest in a bypass for Greater Snortingbury”
Then show the data from your tally to support it. If your words and data don’t match it is usually better to change the words and not the data (though this may not always be politically expedient).
Stage 4: Do something with the data
What ever you find is likely to lead to more questions, operations analysis is like turning over rocks, something will scurry out. Think about what the data has told you, refine your problem statement or hypothesis and then start again.
Happy counting.
If you enjoyed this post sign up for updates by e-mail
Read another opinion
Image by Duncan~’s
Leave a Reply