21.
What is the MOST important statistical attribute a special test must have to effectively limit Type I errors?

1. Sensitivity
2. Specificity
3. Positive likelihood ratio
4. Negative likelihood ratio

The correct answer is: (Click for answer)