No, they're not. Bad analysis => cannot draw conclusions either way.

I have focused on Simpson's paradox in this thread because somebody else brought up controlling for education level, but it's not the only problem I noticed. I don't have any desire to go into a deep technical discussion of p-values and their interpretation, but I'll leave you with the thought that even with purely random data a proportion of them will be below your "critical threshold" alpha due to sheer chance - by definition alpha is the false positive rate for classifying effects as significant. If you try out a whole bunch of models at random, some of them will meet the alpha threshold even though they're not actually significant. The Yale professor strongly inferred that this was his methodology - he took a data set gathered for other purposes and tried things out until he got an interesting "significant" result. The fact that it's a "controversial" result is getting him lots of media attention. In the long run he may or may not turn out to be right, but this isn't good science.

Bottom line, since the analysis was done improperly (in several ways), you can't actually draw conclusions either way.

Because it is possible to have both of these statements be true at the same time: "Tea Party people on average know more science than non-Tea Party People" and "At every education level the Tea Party people know less about science than the non-Tea Party people". The correct conclusion would then be that if you want to know how much somebody knows about science, look at their education but then adjust downwards if they are Tea Party people. Note that I'm not saying that that is the case. I'm saying rather that we can't tell whether the Tea Party identification has a positive or negative effect because the stupid social scientist did an improper statistical analysis. The more dominant education is in determining the outcome, the more important it is that you take it into account when considering the impact of other factors.

I'm not condemning anybody except the Yale professor. I'm just a professional statistician who refuses to get suckered into drawing a conclusion one way or the other at this point based on shoddy analysis, and I'm trying to alert other /.'ers to the fact that the impact could actually still go either way if what you and I both agree is a major factor, education level, is properly accounted for.

Then you're not understanding Simpson's Paradox - this is exactly why you want to control for education.

Consider something floating on the water. Its movement will be a vector sum of wind effects and current effects. Suppose you happen to get a sample where the wind and the current are in opposite directions, and you try to estimate the effect of wind only, i.e., you leave current out of your model. At a minimum you will underestimate the impact of the wind, and if the current happened to be dominant in a large proportion of your sample you might even draw the false conclusion that free-floating objects move in the opposite direction from the wind! That's Simpson's Paradox - by omitting an important effect, you can actually end up drawing a conclusion that is the opposite of the truth.

Because of the potential for Simpson's Paradox.

Pretty unstable, that is. The thing has almost no ground clearance, and seemed to really be wobbly on takeoff. I'd be pretty nervous for takeoffs and landings.

Reminds me of the anecdote about Prof. Rota when he was asked by a reporter why MIT didn't really have an applied math department. He responded, "We do! It's all of those other departments!"

Most authors try to avoid gaping plot holes, while Hollywood seems to consider them mandatory.

Failure to negotiate a license and failure to negotiate for a license are two different things.

So if some author wrote an article discussing finer interpretations of the Bible, citing sections as quotes in the original Latin, Hebrew, or ancient Greek, and you couldn't understand them, that would be willful ignorance on your part?

Because of path dependence. Even if everybody agreed that Dvorak was so much better than Qwerty, and they don't, the transition cost of migrating to Dvorak by replacing all the hardware and software and retraining everybody is deemed not worth it. That's why it's often better to be first to market than to be technically superior -- if you can get a high enough early adoption rate your competitors will have a real uphill battle.

For personal use I like Tower, but that's OS X only and costs $. SourceTree, in addition to being free, works on both Windows and OS X so that's what I recommend for my students.

The real irony was that Lotus killed off the Windows port of Improv because it was overtaking 1-2-3, which they regarded as their flagship product at the time. Improv was a dream to use and stood head and shoulders above Excel and other 1-2-3 knockoffs of the time. Every once in a while I dust off and fire up my old NeXT cube just to blow friends away with the fact that spreadsheets don't have to suck, and we've known how to do it right for over 20 years now. The closest modern equivalent is Quantrix, but that costs a small fortune so it's never going to see widespread adoption - besides which, it got really uglified when they ported it to Java.

If the reputation scores weren't normalized by how long people have been active on the site, it invalidates the study's performance measure. It takes both good answers and time to accumulate high reputations.

Thus guaranteeing that nobody with a lick of sense in the corporate world will touch your stuff. If you didn't specify a license option of some sort, then (at least in the US) you hold copyright and anybody who uses your stuff without your explicit written permission is wide open to a law suit. A lot of folks think copyright is something you have to apply for, but registering a copyright is not necessary - that's merely a convenience to help support your claim if it ever gets contested.

I don't think that means what you think it means.

Who'da thunk this was such a hard concept? Look, if there are primitives in a language, then pretty much by definition not everything is an object.