Using statistics well

I recently wrote a rant about the use (and overuse) of statistics, and pointed to the Twitter Influence Calculator (now Twitalyzer) as a working example of statistics of questionable value.

I should have paused.

It turns out that (a) they are quite thoughtful about the metrics they are using and (b) more important those metrics can in fact teach someone (me for example) how to use Twitter more effectively.

Twitter is a funny thing, and suggesting it can be used effectively may strike some as risible; but the truth is that in my job, making frequent high:signal contact with the community is valuable, and, it turns out, Twitalyzer zeros in on some of the more interesting aspects of that contact.

There is the obvious, such as  Influence which they define (in more detail than I will here) as “number of followers, number of times you are retweeted, your generosity (see below), times you are referenced by others and number of updates you publish in a week.  One can argue for different metrics, but it would be hard to make the case that these are absurd.

Mr. Rogers  In addition to giving you your “score,” they go on to tell you your change, which is helpful, and then if you like, how you might improve that score (e.g., you need more followers and friends…. Won’t you be my neighbor?)Signal

The second number they provide is Signal, and they use it exactly in the same way I do when I write “More Signal/ Less Noise” – that is do your tweets contain useful information. They define useful information in this case as any of the following: references to other people, links to urls, hashtags and retweets (given that they must do this by machine and semantics are hard to interpret, this seems reasonable). They even break these out for you in TwitPie a handy pie chart and another chart showing change over time.

The third value is Generosity, defines as the number of retweets you provide. I was skeptical about the importance of retweeting (in fact, I worried it was just noise) but asked about it on Twitter and received many amazing, thoughtful (brief) responses. Best of the lot was from Bob Martin (@uncleBob) whom I’ve admired for many years, who wrote “twitter is a loose network of tight clusters. RT is the means for communicating beyond your cluster”  Not only do I find that compelling, it is almost a haiku.

The penultimate measure on Twitalyzer is Velocity, defined as  the rate at which you tweet. Velocity + signal/noise gives you a good sense of your contribution.

Finally, Twitalyzer measures your Clout: that is, the number of references to you divided by the total number of possible references (that latter number is a bit fuzzy to me, but I think there is an upper limit set by the Twitter API). In any case, the more folks reference you the more clout you have.

Charts, Tables and Analysis, Oh My

Once you’ve absorbed all that there are endless charts and analysis, but many are targeted at helping you increase your effectiveness, which makes this exercise worthwhile  For example, one analysis offered to me was to compare my average values for number of follwers, retweeting,e tc. with the “11,343 people we’re tracking who have about 25% more influence (on average) than you…”)  Nice.  What stands out in the chart, right away, is that on average, those folks have more friends than I do (story of my life). Where is Mr. Rogers when I need him?

All in all, more information than you can shake a virtual stick at, and much of it useful. And a wonderful distraction when you should be working.

It will be interesting to see (a) how the metrics are refined over time and (b) to what degree the very act of measuring influences that which is measured (The “Liberty’s Group Behavior Uncertainty Principle – measuring the activity of a group causes the members of the group to change their activity” (I’m certain I’m not the first to say it.)

A police officer pulls Werner Heisenberg over for speeding. “Do you know how fast you were going, Herr Doktor?” asks the trooper.  “No,” replies the physicist, “but I know exactly where I was.”

Related Resources

Some Twitter Social Network Analysis
My Experiences with Twitter Part 1
Scott Hanselman on How To Use Twitter
Perl Script for Twitter Analysis

Previous We Love to Measure Everything

This work is licensed under a Creative Commons Attribution By license.

About Jesse Liberty

Jesse Liberty has three decades of experience writing and delivering software projects and is the author of 2 dozen books and a couple dozen Pluralsight & LinkedIn Learning courses. He was a Senior Technical Evangelist for Microsoft, a Distinguished Software Engineer for AT&T, a VP for Information Services for Citibank and a Software Architect for PBS. He is a Xamarin Certified Mobile Developer and a Xamarin MVP and a Microsoft MVP.
This entry was posted in z Silverlight Archives. Bookmark the permalink.