Melted Dead Dinosaurs or Super Heated Sand?

Dishwasher full and running — I need a cold drink after working in the yard. My wife offers me one of the kids’ plastic cups — I choose a half-size juice glass telling her I prefer super heated sand over melted dead dinosaurs.

Married readers will be quite familiar with the look she gave me.

It did get a laugh (eventually) and a suggestion that it was a great article title. So here it is. This kind of labeling is fundamental to data analysis. What you call something influences your perception. I won’t get into the current political circus, but you can see where names and labels in politics drive opinions and emotions across the spectrum. Democrat or Republican — Liberal or Conservative. These cause people to react viscerally, instantly disagreeing with or defending the viewpoint. Some percentage of people overcome that reaction, examine the data more closely, then confirm or adjust their position. As a data analyst we don’t have the luxury of letting people get a second look. Emotion can stomp on our most carefully crafted reports. That emotion causes a lack of trust in the data, and without trust there is no value, or action.

Vulcan Clarity

 Do we strip all emotion from our analysis? Use only titles like “Column 1” or “Attribute Z”. Absolutely not. We construct the data to convey selected emotions. Not bias, or an intent to sway our audience -but to convey confidence, trust, and reliability. We do that by using emotion — and ADMITTING IT. Show the origin of our labeling. Be clear about methodology.

A Better Name

Let’s look at my title: melted dead dinosaurs or super heated sand. Clear bias in labeling. I wanted the glass, so labeled it such. This could have been: colorfully formed specialty cups or well-stirred burned dirt. Much different messaging. Put this in the context of an analysis.

Report Showing Lack of demand for dead dinosaur products

100% of sweaty husbands in the study prefer super heated sand over the alternative.

Knowing the story, you can see it is ridiculous. It is also entirely factual. To present this professionally, there are a few key points we could change.

  1. State the purpose of the study.
  2. Indicate your population.
  3. Show your yardstick (Measurement Systems Analysis).
  4. Use consistent names for data elements.
  5. Be honest about gaps.

Report on Consumer Demand for Types of Drinking Vessels

Produced by National Glassblowers Association

Study Population: 1

Data obtained via verbal survey. No repeatability or reliability tested.

Test products: Plastic 12 oz drinking cup and Glass 8 oz drinking cup.

Study inconclusive due to MSA and statistically insignificant population.

Why it matters

Repeat business. Bottom line, you might make a single customer happy with the marketing spin on the first report but not only will he not trust your integrity, you’re not building a reputation that will drive additional business your direction. Here’s the key: offer next steps. The study was not sufficient. Clear, factual, but so what? The business owner can’t act upon that. Your report needs to identify actionable next steps. Something like: suggest 10 study groups comprised of 15 people each, ages 28–35, male, ethnic and income diversity to reflect population of southern Miami. Even better — give a cost estimate and a time line. Biased data should not drive biased decisions. . . it should drive better data.

For the record, the kids chew on the edges of the melted dead dinosaur cups and it really grosses me out. No offense intended T.Rex.


Warm hugs or cold data?

For those of us that have been involved in studying business processes for a long time, we’ve seen the pendulum swing back and forth between abstract data driven decision making to a humanistic approach where opinions are valued, people are validated, and warm fuzzy hugs matter.

Conflicting Guidance

Study after study shows each of these as successful. Or a failure. Really depending only on which one you’re attempting at the moment. Why the confusion? The study authors (or book sellers) will tell you about tremendous numbers of variables or the changing culture of the millennials. Things that are accepted as truth and very difficult to disprove.

My experience as a business leader and data analyst tells me that they’re both right — and both wrong. Data matters. It matters very much -but ONLY when it is related to people.

Let’s look at an example. Churchgoers. Depending on your choice of denomination you might argue the warm hugs/cold data side of the equation, but still a very important data element tracked by churches is attendance. Another is weekly donations. Most churches have a sign visible and are proud showing these numbers.

You might try to draw a correlation between the number of attendees and the donations. For short periods of time, there will probably be a correlation. Taken as a whole, it is NOT a good indication of the health of the church. It doesn’t address the people. The data alone is not enough. Which people attend matters. How they FEEL about attending matters. For churchgoers in the industrialized world a significant increase in their individual donation would have a negligible impact on their personal finances. The National Center for Charitable Statistics shows that for people earning under $200,000 per year, they give from 2.6% to 4.0% of their income . . . in descending order. That is — the higher earners give less. Why does this matter to us? It shows room for growth in giving. Another interesting item noted by the The Center on Philanthropy (COP) at Indiana University was that for about half the population, giving increased around the holidays. So when people have more of their own expenses — they also give more. Financial pressure clearly not the root cause. Finding the sweet spot is a combination of people and data. Why don’t people give more?

Anecdotal data has been given a bad name by analysts. I think it is key to good judgement. Taking the church example — I could track and trend every possible item like weather, sermon topic, regional economy, and try to correlate to attendance and donations. Not a great use of time. A better approach is to look for the variations in the data that you do track then apply it to people by talking to them. Be direct. Attendance was down for two weeks — start calling people and ask how they felt about the sermon three weeks ago. Use the data to ask the people. This does multiple positive things: it gains engagement in a solution – lets them know you care; it is much cheaper than additional data gathering; and it creates a spotlight effect that causes some improvement WITHOUT additional action.

Most of us don’t run churches, which makes this a great abstraction. Now let’s apply it to our own businesses. Say you run a call center and average time per call has increased (usually not a positive trend). Option 1: berate employees. Option 2: Gather and analyze multiple variables and look for correlation/causation — extremely difficult with any degree of reliability as I’ll discuss another time. Option 3: Engage key team members (not managers) and ask them for input on causes and solutions. No process mapping activities or death by PowerPoint justification for existence — but direct questions of people.

The same approach can apply to your customers. Sales are down. You have some data as a starting point. Ask customers. Don’t try to be cagey or shrewd. Direct, honest questions with a sincere interest will get you the additional data that you need to make good decisions about people.

Cold data should only be a pointer — a place to start when deciding what questions to ask. People are the key to a complete understanding.

Data is the beginning of a solution, not the end.

Hosting Recommendation

I want to recommend SiteGround to anyone thinking of getting their own web hosting service. They were recommended to me recently and have performed exceptionally well. New to having to deal directly with a hosting company instead of self contained firewalls / servers, I was a bit surprised when a security issue came up.
Not only did they notify me immediately, but when I submitted a help request online, they had responded with an answer in under five minutes . . .on a Saturday afternoon. Exceptional customer service and great rates. If you’re interested, follow this affiliate link: 
Web Hosting