1. Looking at the data, a linear regression is fit to data that looks anything but linear, calling into question the validity of the model
2. The units in some of the table are unclear, but overall it's not clear to me how much the brain shrinks in this data. As in, how significant of an effect is this? And I'm not talking about statistical significance.
3. This is correlation not causation. Maybe if your brain shrinks you can't focus long enough to move around.
Yes. I also see problems in other areas:
-they didn’t define the threshold for sedentary behavior beforehand, so it could be they just move the threshold to hack the p-value. At least, they could analyze with different thresholds then publish the results open for interpretation.
- they didn’t control the effect of normal aging in older people. 7 years is a long period and by older people we can expect a decline in cognitive performance regardless of their life style. What we interest is however how much the sedentary live style contributes additionally to this decline.
- because the authors rely solely on the smartwatch data and put all kinds of movements together (sitting, walking around, laying down), we can not see whether changing posture and moving around can help.
- most importantly, they didn’t isolate and control for the amount of cognitive work. Doing cognitive demanding work is known to delaying worsening symptoms in AD and related diseases.
1. Looking at the data, a linear regression is fit to data that looks anything but linear, calling into question the validity of the model
2. The units in some of the table are unclear, but overall it's not clear to me how much the brain shrinks in this data. As in, how significant of an effect is this? And I'm not talking about statistical significance.
3. This is correlation not causation. Maybe if your brain shrinks you can't focus long enough to move around.