Validity and Reliability

When researchers design studies and interpret the results, two important concepts are validity and reliability. Simply stated, research validity is concerned with the question, “Are we measuring what we think we are measuring?” Research reliability is concerned with the question, “Are we measuring this consistently?” This entry will review both internal and external validity in quantitative research, threats to validity, common methods of measuring reliability, and the importance of validity and reliability in qualitative research. An example will demonstrate how validity and reliability function and why they are so important for research.

Lucy buys a new scale. She steps on it for the first time before breakfast, and the scale reads 132 pounds. She starts a diet. Five days later, she weighs herself again before breakfast. The scale reads 132 pounds. She continues to diet, and 5 days later she weighs herself again. The scale still reads 132 pounds. This scale is producing reliable, or consistent, results. One could draw the conclusion that Lucy's diet did not work. However, are the results accurate or valid? First, one would have to know how and when she was weighing herself. The first day Lucy stepped on the scale it was warm outside, and she was wearing shorts, a t-shirt, and flip-flops. Five days later, when she weighed herself again, it is raining, and she was wearing jeans, a sweatshirt, and sneakers. Five days after that when she weighed herself the final time, she changed pattern and weighed herself late in the evening, after eating dinner. She was also wearing jeans, sneakers, and a sweater. As anyone who has dieted can attest, though these results are reliable, they are not valid—what Lucy was really measuring was the changes in her clothing and the timing of her weighing-in sessions, not her real weight.

To draw conclusions about a phenomenon, researchers have to be certain that they are measuring the phenomenon correctly—they want to measure the actual weight and not the changes in clothing. This is the issue of validity. Additionally, it is important that researchers can replicate their results. If a researcher only gets a certain result one time, she or he cannot trust the results. Conversely, just because results are reliable does not mean they are valid—one can consistently measure the wrong concept. These issues are so important that researchers must guard against the threats to validity and work to enhance reliability. Researchers have also developed multiple ways to measure validity and reliability