IPPCR 2018: Measurement Part 2: Validity

>> David Luckenbaugh: Hello, my name is David Luckenbaugh. I'm a statistician with the Office of Equity, Diversity, and Inclusion in the Office of the Director at NIH. I want to welcome you here for a conversation about measurement validity in clinical research. Our objective today is to try to enhance understanding of the key principle of validity for clinical research.

So, imagine a lot of people do this, first thing in the morning, some people do it in the afternoon, you might hop on a treadmill, maybe you just go out running. But there's a lot of ways that you can think about what's happening going on -- what's happening with what you're doing. You could think about the time, how long have you spent running,

or walking, or doing what you're doing? How long have you run, how fast have you gone? How many calories have you burned? All of these are different ways of thinking about performance. So, with performance, we can talk about performance as a construct. A "construct" is a theoretical concept. In looking at the measurement we just talked about,

performance in running or in walking, the theoretical concept could be, in fact, performance. But you have to decide what your measurement is going to be. Your measurement is some way in which you can look at that theoretical concept. So, if performance is what you're interested in, or speed is what you're interested in,

you can develop a measure of speed. On a treadmill it measures it for you so it's really easy to do that. So, your operational definition for speed could be, "Whatever the measure is that's shown on your treadmill." For distance it could be the amount of miles that you ran. So then, if you were studying the speed that somebody ran then