Austin - Nate Silver told us here at SXSW
that his work in the science of prediction is popular because, as he put it,
there was so much “low-hanging fruit” in his first two fields, politics and
baseball. He believes those two domains
– served by his www.fivethirtyeight.com blog and PECOTA tool for forecasting
ballplayer performance – delivered “lots of marginal gains” from relatively
modest effort. Huh? I wonder what brilliant, successful pollsters
and sabermetricians think about their work before Silver’s arrival.
When asked about new arenas for study, Silver listed public policy matters such as urban planning, education and even prison recidivism. Why? Because “these are areas with low levels of competition,” he asserted. Silver said these domains offer the marginal gains he seeks. He added in this context that, “We need more data in education.” Really?
When asked about new arenas for study, Silver listed public policy matters such as urban planning, education and even prison recidivism. Why? Because “these are areas with low levels of competition,” he asserted. Silver said these domains offer the marginal gains he seeks. He added in this context that, “We need more data in education.” Really?
We had a chance to talk after his presentation, which drew
on themes from his best-selling book, “The Signal and the Noise.” I offered
what is painfully obvious to many folks working in higher education research
and prediction these days – institutions are drowning in far too much data with
too much of it of limited value, or worse. The argument should be for better
data, not more data. He agreed with this point, of course, since his central
premise is to discern useful signals from wasteful noise.
I added that what’s really needed are sharper, prioritized
objectives and strategies from colleges and universities – and all types of
organizations – that can help executive teams and institutional research
professionals understand what is most important. It seems to me that before we
can sort signals from noise, we need to know what matters most to our
organizations and why. It’s as if, and forgive this tortured verbiage, leaders
need first to separate wheat from chaff at the strategic level if their data
analysts are to know how to discern signals from noise in specifc, contextually
useful ways. Twitter @jessicamcwade