Skip to main content

Posts

Showing posts with the label validity

Testing the Test

  Testing Features. We are constantly taking tests, giving them, or reading about them.   Let’s get the outcomes straight. That is, let’s learn the testing buzzwords.   In this way, you’ll know what to look for when somebody proposes a test for something, or gives you some test results.   For example, COVID-19 testing is replete with hidden meanings and outcomes. ·         Sensitivity : Ability to test correctly for the disease (or whatever). ·         Specificity : Ability to test correctly for no disease (or whatever). From these we have ·         True positive: the test is positive for disease and disease is present. ·         True negative: the test is positive for no disease and no disease is present. ·         False-positive: the test is positive but no disease is present. ·         False-negative: the test is negative but the disease is present.   ·         Validity : test measures what it is supposed to test. There are at least five types of validity: o    F

What is Validity?

Validity is a mighty big word. It seems to have its own meaning for each subject, though most center on the notions of correctness, consistency, conformity, and accuracy.   In fact, just Google, “whatchacallit validity” to get another definition.   Below are a few notions of important definitions. a.        As a state, as in being authentic or genuine. b.        As a force, for example legally. c.        As a measurement, as implying it does measure what is purported. d.        In research, as in the accuracy of measurements. e.        As proof, as in obeying the laws of accepted logic and premises. f.         As approved, as by an authority. g.        In statistics, as following and satisfying statistical tests. h.        In methodology, as in properly observing accepted procedures. i.          In prediction, as in how a test predicts another criterion. j.          In construct, as in the degree to which inferences can be made or how two tests closely compare.