Posts Tagged data
On Friday, the HBR Daily Stat posted a story about the link between high school club memberships and a person’s propensity to be a manager later in life. Later that day I ran across a tweet from Dave Zinczenko of Men’s Health Magazine, claiming that 7+ hours of sleep a night leads to a lower risk of heart disease.
The thing that strikes me about both of these articles isn’t necessarily the information that they convey – I think many would agree that getting more sleep is better and that the editor of their high school yearbook will make a good manager. What strikes me about both of these articles so completely confuse the concepts of correlation and causality.
Is seven the magic number of hours of sleep to ensure you’ll live a long and prosperous life? Or is the kind of person who sleeps at least 7 hours a night the kind of person who takes care of themselves, watches what they eat, exercises, and tries to limit the amount of stress in their life? Does being in a certain school club raise the likelihood of being a manager, or do people who gravitate towards certain school clubs (a la Freaks and Geeks) tend towards a certain kind of success?
President Bartlet: CJ, on your tombstone it’s gonna read ‘Post hoc ergo propter hoc.’
CJ: Okay, but none of my visitors are going to be able to understand my tombstone.
President Bartlet: Twenty-seven lawyers in the room, anybody know ‘post hoc, ergo propter hoc’? Josh?
Josh: Ah, post, after hoc, ergo, therefore… After hoc, therefore something else hoc.
President Bartlet: Thank you. Next? Leo.
Leo: ‘After it, therefore because of it’.
If you think I’m nitpicking over very fine distinctions, consider this: how many times has a finance manager asked you to quantify, in dollars, the benefits of your marketing activities? How many times have you been asked to predict – and later to show – causality between what you’ve done and some form of business result, no matter how multi-variate and subtly complex the interdependencies really are.
Did sales go up because of the advertising campaign you recently launched, or did they go up because the economy got better and people feel more comfortable spending money? Did market share go up because of they change you made to your product or the change that your competitor made to their product? Is your product not selling because people simply aren’t aware of your product, or is it not selling because people don’t see a need for your product. Or are there many more factors that contribute to your success or lack thereof?
While I think its important to set measurable goals and work towards attaining them, I also think its important to know when your thing did something and when you did something while something else happened at the same time. And I think its important for all of us to be honest about when it’s the former and when its the latter.
Admitting uncertainty means facing reality — and our own needs for security. But admitting uncertainty is not enough. We must learn to actively embrace uncertainty and work with ambiguity.
As I sit in a two day meeting about drawing insights from data quantification, this article is very timely. Obviously there’s merit in looking at your data and trying to draw as many insights as possible, but we as marketers can sometimes end up hiding behind the data, using them as an excuse to take no action in favor of gathering and analyzing more data.
There comes a time, however, where you have to accept whatever level of ambiguity you’re willing to accept, make a decision, and take action. The question then becomes: When is the right time? A better question might be: What do I get by delaying and gathering more data? Am I falling prey to the law of diminishing returns?
How do you overcome “Analysis Paralysis”?