Comments on: Lies

By: david

david — Sun, 27 Jul 2008 19:09:45 +0000

and what’s the right way? Therin lies the rub…

By: Stephen

Stephen — Sat, 26 Jul 2008 23:41:11 +0000

And that’s probably the most difficult part to do.

By: Tom Osborn

Tom Osborn — Sat, 26 Jul 2008 03:45:58 +0000

AND looking at the data in the right way.

By: Stephen

Stephen — Fri, 25 Jul 2008 12:29:33 +0000

If I’m watching the pixels that register your behavior and making key decisions based on that, how do I know you are just not bored and moving the mouse in circles?

I assume that’s why companies like Nuconomy are measuring “engagement” instead of “attention”.

In Blink, Gladwell makes an important point. It’s not the volume or range of data that helps you make the right decision, it’s selecting the right data to look at in the first place.

By: david sable

david sable — Thu, 24 Jul 2008 17:00:33 +0000

Blink moments….http://www.gladwell.com/blink/

By: nick

nick — Thu, 24 Jul 2008 08:41:43 +0000

I think Clarke pulls out a good point. Any one with .5 of a brain can use the same statistics to point to differing conclusions – BUT bringing your own personal experience to the issue can be invaluable – gut feel, intuitive thinking,instinct, whatever you want to call it – we spend every second of our waking lives taking in information. We should make the most of what we’ve personally experienced.
Sometimes it’s better to take a step back and say “hey that just doesn’t seem right based on my experience, I would expect….”

But always keep those stats in the back pocket armoury.

By: david

david — Tue, 22 Jul 2008 07:17:19 +0000

I forgot about the wandering eyes…….

By: Tom Osborn

Tom Osborn — Tue, 22 Jul 2008 06:40:20 +0000

A message I like to use with my clients about simulators and modelling: Think of the computer (or simulator, or OLAP tables and dashboards) as another voice in the room. It can answer questions that may be hard to answer in any other way. But as a voice in the room, you have to decide whether you want to believe it. There might be something else going on…

Also, statistics can come out wrong if they are abused, or wrong because they are analyses of the past (where data comes from), while consumer behaviour is in the future. Good experimental design lets you reach into the future (but not very far).

Also, statistics and modelling often are right, and their impact has been in lower costs, more effective reach, channel innovation, etc. Why isn’t data-driven commerce perfect then? Partly because everyone is using it (to some extent) and competing with each other (raising the bar), and partly because the future isn’t the past.

Tom.

PS: I’ve seen eyeball tracker studies. They cost heaps, and pity the poor guy with wandering eyes.

By: Clarke

Clarke — Tue, 22 Jul 2008 05:52:41 +0000

That quote is from a politician – people often accused of being economical with the truth. So I think the statement is often taken out of context – really, it is an inditement of the use of statistics by politicians in a time (19th Century England) long before research codes, bodies and standard methodologies.

Sure, interrogate and understand the research methodology before you trust any statistics, and take them in the context of personal experience too. But if you don’t try to put rigorous data insight a the core of your decision making process you’ll get beaten by your competitor who does. 99% of the time anyway