One of the best criticisms of an article in @Quillette I have yet seen. On the (de)merits of the content (credibly argued as flawed and possibly biased, with evidence, not insult). This is how intellectual criticism should be done.http://ow.ly/5dJw50tSPS8
-
Yes. There is just not enough data to make the model really. So they shouldn't have tried.
That's a different argument than your parallel lines one. Sample size argument depends on what effect sizes you expect to find. They had enough to find a massive difference (d > 1.00), like often seemingly claimed. They didn't have enough to find d = 0.30 or whatever.
-
-
The lines are not based on data in each group, it's from the model fit, which assumes no interactions, hence lines must be parallel no matter the amount of data used to fit.
End of conversation
New conversation -
-
-
To be clear I'm not saying that I think they fudged the data or something. I just think they shouldn't have presented the model like this. If they want to do a test between two groups, do that. Don't plot what you don't really have data for, propped up with a model.
Thanks. Twitter will use this to make your timeline better. UndoUndo
-
Loading seems to be taking a while.
Twitter may be over capacity or experiencing a momentary hiccup. Try again or visit Twitter Status for more information.