Are we out of the woods yet?
Are we out of the woods?
Are we in the clear yet?
In the clear yet?
In statistics, as in life itself, we are often rather not so good at listening. In life, as in statistics itself, that is because we do not listen to understand. Rather, we listen to reply. I’ve mentioned this before in my posts, but I would like to return to this idea again, from a mathematical rather than an emotionally intelligent perspective.
I was watching a favourite data scientist of mine recently, Hans Rosling. He recorded a video a few years back that I often share with my statistics students, but it is his 2014 talk in Berlin that drew my attention again. It talked about how to not be ignorant. What struck me is that, to reply well, we must understand to listen.
What does this mean, understanding to listen? Well, in mathematics, as in most things, if one does not understand, it becomes hard to listen. The words becomes a sort of vague monster eluding understanding or comprehension. Some people can still listen; I was recently complimented for how well I listened to a lecture given in a language of which I have almost no understanding. And while I felt it only common politeness to listen avidly to the speaker, mine is a fairly useless-to-me skill. I understood very little of the lecture, although I’m passably good at mathematics.
You see, even if one listens avidly, intently even, it can become difficult to separate the signal from the noise. Forgive me, this is a technical term. The signal contains the actual message we want to receive. It is the knowledge we search for in data. The noise is all the messiness of poorly collected data, not enough data, or the fact that we are simply bad at hearing analysis (tinnitus anyone?).
When looking at data, there are often issues that interfere with a good, clean signal. Study participants may not understand the language of the instrument well. I was with a group once that asked students “What do you like most about your discipline?” You may be surprised at how many wrote about corporal punishment rather than psychology. I was also at the table during survey design when semi-heated discussion broke out about the nuances between professor, teacher, instructor, and lecturer. That was made all the more memorable as more than one person at the table was British – where those words lose all their American connotations entirely.
This happens outside of statistics, too, of course. When was the last time the phrase “with all due respect” or “I’m not a ____” actually meant what they ought?
The advantages of life over statistics is that there is usually more room to connect and re-transmit the signal. If the message you want to send is “I care,” then that is your signal. What noise is introduced by the differences between email versus foolscap? I’d argue both will introduce some noise, but in certain scenarios, one or the other may be less noise.
But we’re not talking about sending messages. We’re talking about receiving them. Listening.
Nowhere is it more clearly seen just how much understanding it takes to listen than in sentiment analysis or ‘text mining.’ One of the most challenging problems today in statistics is classifying human words into positive or negative sentiments. Talk about noise! Take a moment to stick a paragraph into the linked analyser. Do you think it worked?
What does it take to understand enough to listen? How do we prepare to listen well?