Article on the general problem of measurement error in epidemiology intended for the intelligent layman. Michael Wallace explains the potentially severe consequences of this often overlooked issue, and how statistics can help bring us back - or at least a little closer - to the truth:
"Analysis in an imperfect world" (Significance)
| ⇾ |
An inconvenient truth: When we observe the world, we sometimes make mistakes, which can invalidate the findings. |
| ⇾ |
Errors can come in many forms:
|
| ⇾ |
A convenient untruth: Due to the myth of attenuation the impact of measurement error tends to be overlooked and trivialized in the general belief. |
| ⇾ |
In the study of measurement error, more data is needed to characterize the size and structure of the measurement error itself:
|
| ⇾ |
A whole host of techniques are available for addressing the problem of measurement error in practice:
|
| ⇾ |
Problems that measurement error might cause can be minimized or avoided by proper study design and analysis strategy. |