It was a county with the highest labor non-participation rate in America and an embarrassment to the state’s Governor. Labor non-participation measured the people who were unemployed and not looking for employment. The Governor asked one of his administrative assistants to give him ideas of how to increase the labor participation rate. The result of the assistant’s work was a training program for remote workers since the county was very rural and unlikely to attract outside business. The cost of the program was reasonable, and the Governor proudly announced it at a press conference.
One year later, the training program was a miserable failure. No one signed up for it. The Governor asked another assistant to visit the county to find out what the problem was. Here’s what he found out. Virtually everyone who could work was working. The county’s economy was largely cash only employment. People worked at the local ski resort and hunting lodges for cash (no checks, no taxes withheld, etc.). They made hay and planted crops in the summer for the corporate farms in the area. Again, for cash only. Virtually everyone in the county had their own home that they built for themselves. They lived on at least one-acre plots of land. They had huge gardens and canned most of what they grew for winter eating. They had well stocked deep freezers from the hunting season and from butchering of their own animals. What the assistant concluded was that the citizens of the county had a life that many would envy.
Touching the data refers to the need to gain a first-hand understanding of the data through direct observation. Without touching the data, you will have missed the real insights the data provides.
There are various ways to touch the data.
- Do the job yourself, if possible, if you are observing work being done.
- Personally interview people who are the source of the data. Their stories will provide incredible insights.
- Look for outliers that are not representative of the rest of the data. These can often provide useful insights on what could go wrong or conversely what best practices might look like.
- Look at conformity or lack of conformity in the data. Both fs these can be very insightful. If the data is very consistent, explore how this might have happened and whether this is good or bad. Do the same thing if the data lacks conformity.
- Look for distinct populations in the data. Again, ask why this might be the case and what this might suggest for the analysis you are doing.
Touching the data takes longer than just relying on existing data. The analysis will take longer. Then how do you justify touching the data? You might answer the justification by asking these questions: How important is it that we get real insights for our analysis rather than just mindless number crunching? How can we get insight on the intangible and largely human stories that data alone can’t reveal? While we can measure the extra time and cost for touching the data, how do we measure the time and cost of a wrongly formed analysis?
No matter where a person is in an organization, there is an incalculable benefit from touching the data.