Reliability and validity are core aspects of measurement. 8 Previously validated instruments may require adaptation for changed circumstances, but, whether utilizing an existing instrument or developing a new one, explicit attention to measurement is important to the conduct and reporting of research. 4–6 In other disciplines such as the behavioral sciences, there are bibliographic databases of measurement instruments, 7 and researchers are trained to use existing instruments with known validity and reliability whenever possible. Without this infrastructure, the health informatics evidence base will be weak and knowledge will not cumulate. 3 A robust library of reusable instruments creates an infrastructure for research that facilitates the work of study design, strengthens the internal and external validity of studies, and facilitates systematic reviews. It has been suggested that health informatics has a paucity of well-known and consistently used research constructs with established instruments for measuring them.
Just as poor study design or inadequate sample size can jeopardize the integrity of a study, so too can measurements that are to a significant extent unreliable or invalid. 1 If these are absent, investigators must proceed carefully based only on assumptions about what their measurements mean. Before using a new instrument, investigators should carry out measurement studies that explore whether the methods are acceptably reliable and valid. For this reason, researchers use preexisting measurement instruments wherever possible and typically only develop their own instruments when there is no existing suitable instrument or when they are measuring a new construct unaddressed in published research. Measurement is fundamental to empirical science and requires instruments that are valid and reliable: that provide reproducible results and measure what they claim to measure.
0 kommentar(er)
