Interesting research that seems directionally true, but I wonder if this overstates the case. The sample set is people who are willing to download an Android app and share data with Stanford researchers. Those people likely have more info online than the average person.