COMM 101 Lecture Notes - Lecture 7: Airport Security, Criterion Validity, Observational Error
Document Summary
Goal: determine how well the variables measure the concepts. Two basic principles: we want consistent measures, we want measures that assess the ideas we are hoping to test. Data is never perfect: perfect would mean that we measure exactly the concept we think we are without error: the goal is to minimize error. Imperfect data comes in two forms: random error: we don"t know when/why the measure sometimes screws up. Ex: airport security (on a non-holiday weekday) how early should you be: 30% of time, no line, 50% of time, short line, 20% of time, long line. No matter what you guess, you"ll be wrong some of the time: systematic error, or bias: some groups of people have different measures. Time everybody and see how fast they can run a mile: problem: a taller person will typically run faster than a shorter one even if they are equally t.