The more data that I explore the more I’ve come to realize that truly objective data is a rarity. There are omissions, unquantifiable variables and simple biases that all play a role. Understanding the strengths and weaknesses of a particular data series is vitally important to good analysis. I’ve seen many a researcher come to bogus conclusions because they failed to understand that such-and-such was an artifact of the data and not of the phenomenon being studied.
So what does this have to do with ice? In this week’s Monday Morning Sea Ice/Global Temp report (link leads to a video), Joe Bastardi wonders aloud if U.S. arctic/antarctic ice data is not tainted by ideological bias in support of the global warming hypothesis. This is not a charge that anyone should make lightly, but Bastardi has clearly done his homework with a plethora of additional data in direct conflict with the U.S. data. As a fellow data-geek, I can appreciate how the dots can slowly begin to connect themselves as one studies one ream of data after another.
This is why I follow one cardinal rule with data work–the researcher must collect and analyze the data personally. This may seem a tedious waste-of-time when so much free labor, i.e., interns, may be floating around, but these types of anomalies take time to ferment in the brain before you have that “A-Ha” moment. I simply don’t understand how anyone can put their name on an analysis and not have done any of the data work . . . at best, you are leaving knowledge on the table and, at worst, opening yourself up to trouble.
Is Bastardi right? Who really knows . . . since there is another cardinal rule of data work: if you mess-up (purposefully or not), no one will notice. Or so you hope 🙂