Specificity
Specificity is about diagnostic tests identifying healthy people
Specificity gives you a measure of the ability of a diagnostic test to identify healthy people or people who do not have a disease state. In order to set up an experiment related to the specificity of a diagnostic test, you must have a current "gold standard" to use as a reference for detecting disease states.
In regards to the 2x2 table, specificity is focused on this specific part of the table:
Specificity is the number of people without a disease (as rated by the "gold standard") that have negative test results for the diagnostic test. In other words, specificity is the proportion of people that were healthy and tested negative for the diagnostic test.
Another interesting aspect of specificity is that in some cases, it can "rule in" a given disease state. If a test has high specificity (> 95%), then it can detect healthy people. This means, at the same time, that a positive test result is a strong indication that a person does HAVE the disease. With high specificity, the ability to "rule in" disease states will increase.
Another interesting aspect of specificity is that in some cases, it can "rule in" a given disease state. If a test has high specificity (> 95%), then it can detect healthy people. This means, at the same time, that a positive test result is a strong indication that a person does HAVE the disease. With high specificity, the ability to "rule in" disease states will increase.
Here is a 2x2 table that depicts the calculation of specificity:
Specificity
There is a quick and easy calculation that can tell you if a diagnostic test is useful in any capacity whatsoever, the Youden index. The formula is ((Sensitivity + Specificity) - 100). Any Youden index value below 50% means that test is essentially worthless and should be discarded.
Click on the Receiver Operator Characteristic button to continue.
Hire A Statistician
DO YOU NEED TO HIRE A STATISTICIAN?
Eric Heidel, Ph.D., PStat will provide you with statistical consultation services for your research project at $100/hour. Secure checkout is available with Stripe, Venmo, Zelle, or PayPal.
- Statistical Analysis on any kind of project
- Dissertation and Thesis Projects
- DNP Capstone Projects
- Clinical Trials
- Analysis of Survey Data