Statistics Flashcards
Clinical Significance
Practical importance of a treatment effect - whether it is real, genuine, palpable, noticeable.
Clinical Trial
- any research study that prospectively assigns human participants or groups of humans to one or more health-related intervention to evaluate the effect on health outcomes.
Confidence Intervals
Range of values so defined that there is a specified probability that the value of a parameter lies within it
Data
Recorded, factual material
Effect Size
- magnitude of an intervention, reflected by an index value
- independent of sample size
- calculated from data in a clinical trial
- most interventions have small to moderate effect size
Effectiveness
Performance of an intervention under “real-world” circumstances
Efficacy
Performance of an intervention under ideal and controlled circumstances
False Negative
A test result which incorrectly indicates that particular condition or attribute is absent
False positive
A test result which incorrectly indicates that a particular condition is present
Fidelity
- The extent to which delivery of an intervention adheres to the protocol or program originally developed
- How close the intervention reflects the appropriateness of the care that should be provided
Implementation Science
The science of putting (executing) a project or a research finding into effect
Minimally Clinically Important Difference
- smallest difference in score in the domain of interest, which patients perceive as beneficial, mandating a change in patient’s management
P-value
The probability, under an assumption of no difference in groups of obtaining a result equal to or more extreme than what was actually observed. Usually depicted at 5%
Reliability
The degree to which the result of a measurement, calculation, or specification can be depended on to be precise.
Statistical Assumptions
Characteristics about the data that need to be present before performing selected types of inferential statistics
Statistical Significance
The claim that a result from data generated by testing or experimentation is not likely to occur randomly or by chance, but is instead likely to be attributable to a specific cause
True negative
A test result that accurately indicates a condition is absent
True positive
A test result that accurately indicates that a condition is present
Validity
Extent that the instrument measures what it is designed to measure
Nominal data
Two categories “yes or no” “boy or girl”
Ordinal data
Has order, but no rank
- strongly agree, agree, disagree
Interval
Has rank order: 1-4, 5-8, etc
Ratio
Has rank, order, and countable.
Ex. Weight, temperature, age
Parametric vs. nonparametric tests
- Parametric tests = used when data is normally distributed
— tests group means - Non-parametric = distribution-free test, don’t assume that your data follows a specific distribution.
— used with smaller sample size
— when you want to be more conservative with your analysis
— tests group medians
Tests of differences
- designed to measure differences in sample group means, medians, or proportions
Tests of association
Used to discover if there is a relationship b/t two or more variables.
Sensitivity Test
Proportion of pts who have the disease, who register a positive finding
- rule in
Specificity
Proportion of pts who do not have the disease, who register a negative test finding
- rule out
Positive predictive value
Probability that subjects with a (+) screening test, truly have the disease
Negative predictive value
Probability that subjects with a (-) screening test, truly don’t have the disease
Positive Likelihood Ratio
Reflects the probability of a pt with the disease + a positive test, divided by: probability of a pt without the disease+ a positive test.
- used to rule in a condition
Negative Likelihood Ratio
Probability of a person who has the disease + negative test. Divided by: probability of a person who does not have the disease + negative test.
- commonly used to rule out a condition
Accuracy
Calculated by examining the proportion of true positive and true negative in all evaluated cases
Accuracy = TP+TN/TP+TN+FP+FN
Journal Impact Factor
Number of citations made in the current year to articles in the previous two years/total number of citable articles from the previous two years.
