Interesting to read @jean_twenge's critique of @OrbenAmy & @ShuhBillSkee's work (tinyurl.com/y75zg5dt) recently published in Nature. Although Twenge et al touch on the importance of how we measure digital tech use (DTU), I think the issue deserves more attention [1/9]
Twenge et al title their critique “Underestimating digital media harm.” But it’s the base of that first word (ESTIMATE) and how it relates to measurement that has important implications for the findings in this field. Most studies examining the link b/w DTU and well-being…[2/9]
rely on retrospective estimates of use. This is the method used in representative studies like Monitoring the Future and YRBSS. BUT, people are bad at estimating DTU. Which makes sense. It’s hard to estimate behaviors you do all the time! [3/9]
tinyurl.com/ycjd46v8
tinyurl.com/ycjd46v8
Studies have found that estimates of DTU don’t correlate strongly with actual use (see @davidaellis review here: tinyurl.com/ycacdrb8). And, importantly, well-being outcomes have stronger correlations with estimates of DTU than more objective measures. [4/9]
This is what @OrbenAmy & @ShuhBillSkee found in their Psych. Science article comparing retrospective estimates of DTU against EMA: tinyurl.com/ybyhjl4k. And we found this when comparing estimates of iPhone use to objectively tracked iPhone use w/ the “Screen Time” app [5/9]
Providing an estimate of DTU depends heavily on various cognitive processes. Thus, things that impact those processes (e.g. depression) may impair your ability to provide an accurate estimate. HOWEVER, objective measures of DTU don’t rely on these cognitive processes [6/9]
This leads to measures of estimated DTU that are systematically biased. And, crucially, this bias may be predicted by factors that are fundamental to the associations being examined, such as higher amounts of DTU and higher levels of depression [7/9]
tinyurl.com/ybau2xy7
tinyurl.com/ybau2xy7
If those who are more depressed and/or using DTU the most are the ones providing the least accurate estimates, that has concerning implications for findings in this field. As most studies have relied upon some sort of retrospective estimate of use to measure DTU [8/9]
Make sure to read @OrbenAmy & @ShuhBillSkee's response to Twenge and colleague’s critiques here: tinyurl.com/y94tan8a [9/9]
