For example, if we administer a measure of depression to a sample of participants diagnosed with Major Depressive Disorder, the reliability of such scores does not apply if we administer this instrument to the general population. For the reliability coefficient to be relevant to a given population, the population must be similar to the sample initially used to assess reliability. The reliability of the scores depends largely on the use of the instrument and the population to which it will be administered. Reliability is generally lower for research purposes than for clinical use. Researchers can afford a rough estimate of reliability due to the abstract nature of their work, while scores from assessments used in a clinical context have a direct effect on an individual's life. Reliability can also be too high, for example when elements of an instrument are excessively redundant or too redundant