EXSC 4241/5241 Clinical Research & Design - Unit 3 Flashcards

Flashcards for EXSC 4241/5241 Clinical Research & Design Unit 3, focusing on statistical power, p-values, confidence intervals, and integrating study results into clinical practice.

Null Hypothesis (Ho)

The hypothesis tested by the statistical test, often characterized as no difference between groups in a treatment study. It is either rejected or not rejected based on statistical significance.

Alpha (α) / Significance Level / P Value Criterion / Type I Error

The probability of wrongly rejecting the null hypothesis (false positive). Traditionally researchers set alpha at 5% which is the criterion to determine statistical significance.

Beta (β) / Type II Error

The probability of wrongly accepting the null hypothesis (false negative). Traditionally, researchers set beta at 20%.

Statistical Power

The probability that an experiment will find a treatment effect when a treatment effect really exists in the population. Power = (1 - β); traditionally set at 80%.

Effect Size / Treatment Effect / Minimal Clinically Important Difference (MCID)

The amount of change in the treated group considered important by clinicians and patients, indicating the clinical meaningfulness of the finding. Includes measures like mean difference, standardized mean difference, Cohen’s d, odds ratio, relative risk, and correlations.

P Value

The probability of obtaining data or more extreme data given that the null hypothesis is true. If the p-value is equal to or less than the alpha level, the null hypothesis is rejected, and the result is statistically significant.

Confidence Interval (CI)

A range of values (lower and upper limits) within which the true population parameter is estimated to lie, with a specified degree of confidence (e.g., 95% or 99%). The width of the interval conveys statistical and clinical information.

Minimal Clinically Important Difference (MCID)

The smallest change in an outcome measure that is considered important by patients and clinicians, distinguishing real improvement from measurement error or trivial change. Should exceed measurement error to be meaningful.

Type I Error

The error of rejecting a true null hypothesis (false positive). The probability of making this error is denoted by alpha (α).

Type II Error

The error of failing to reject a false null hypothesis (false negative). The probability of making this error is denoted by beta (β).

Descriptive Statistics

Statistics used to summarize and describe the characteristics of a data set, including measures of central tendency (mean, median, mode) and measures of dispersion (range, variance, standard deviation, interquartile range).

Inferential Statistics

Statistics used to make inferences about a population based on a sample of data, including parametric and nonparametric tests.

Parametric Statistics

Statistical tests that assume the data are sampled from a particular distribution (e.g., normal distribution) and require specific assumptions to be met.

Nonparametric Statistics

Statistical tests that do not require specific assumptions about the distribution of the data and are used when parametric assumptions are not met.

Variability

The extent to which data points in a statistical distribution or data set diverge from the average value.

One-Tailed Test

A statistical test in which the critical area of a distribution is one-sided so that it tests whether the parameter is greater than or less than a certain value, but not both.

Two-Tailed Test

A statistical test in which the critical area of a distribution is two-sided and tests whether the parameter is either greater than or less than a certain range of values.

Kurtosis

A measure of whether the data is heavy-tailed or light-tailed relative to a normal distribution. High kurtosis means more of the variance is the result of infrequent extreme deviations, as opposed to frequent modestly-sized deviations.

Skewness

A measure of the asymmetry of the probability distribution of a real-valued random variable about its mean. The skewness value can be positive, zero, negative, or undefined.

ANOVA (Analysis of Variance)

An inferential statistical test designed for use with interval/ratio (score) data that is used in studies with 1 independent variable that has 3 or more levels or with 2 or more independent variables to examine the interaction between the independent variables.

Familywise Error Rate

The probability of making at least one Type I error (false positive) when performing multiple statistical tests simultaneously. It increases with the number of tests performed.

Post Hoc Tests

Statistical tests performed after a significant ANOVA result to determine which specific groups differ significantly from each other. Examples include Bonferroni adjustment, Scheffé test, and Tukey’s HSD.

Bonferroni Adjustment

A multiple comparison procedure that controls the familywise error rate by dividing the alpha level by the number of tests performed, resulting in a more stringent significance level for each test.

Odds Ratio (OR)

A measure of association between an exposure and an outcome, representing the odds that an outcome will occur given a particular exposure, compared to the odds of the outcome occurring in the absence of that exposure. The only sensible measure to report in case control studies.

Relative Risk (RR)

The ratio of the probability of an event occurring in an exposed group to the probability of it occurring in an unexposed group. Used in cohort or experimental studies and cannot be calculated from a case-control study. RR = CER/EER

Absolute Risk Reduction (ARR)

The difference in the event rate between the control group and the experimental group, representing the reduction in risk due to the intervention. Should be reported with RRR. ARR = |CER - EER|

Relative Risk Reduction (RRR)

The proportion of baseline risk that is removed by the treatment. RRR = 1-RR

Number Needed to Treat (NNT)

The number of patients who need to be treated to achieve one additional good outcome or prevent one additional bad outcome. Useful, but does not indicate who will benefit and who will not. NNT = 1 / ARR

Number Needed to Harm (NNH)

The number of patients needed to be treated to achieve one additional harm..NNH = the inverse of the absolute risk increase (ARI), expressed as a percentage (100/ARI)

Chi-Square (χ2) Test of Independence

A statistical test used to determine if there is a significant association between two categorical variables. Analyzes frequency data.

McNemar Test

A statistical test used on paired nominal data. It is applied to 2 × 2 contingency tables, in which each subject is represented twice, to determine if the row and column marginal frequencies are equal (that is, whether there is a 'treatment effect').

Fisher’s Exact Test

Statistical test used to determine if there are nonrandom associations between two categorical variables. Use when sample size is small and Chi Squared is innapropriate.

Pearson Correlation (Pearson r)

A parametric calculation to measure the linear dependence between two variables . Values range from -1 to +1. Used with interval / ratio with bivariate normality.

Spearman Correlation (Spearman rho)

A nonparametric measure of the monotonic relationship between two datasets. Used with ordinal variables or interval / ratio with normality cannot be met.

Coefficient of Determination (r^2)

The square of the correlation coefficient (r), representing the proportion of variance in one variable that is explained or accounted for by the other variable (shared variance), Can be used to judge clinical meaningfulness of correlations.

Multiple Regression

Develop a model to predict or explain the value of one variable using the values of predictor variables

Logistic Regression

Develop a model to predict or explain group membership, the variable predicted or explained is binary; the outcome being predicted (Yes vs. No) is binary / categorical at the nominal level

Standardized Coefficients

Unitless measure enabling comparison of the magnitude of coefficients in different units. Original metrics have been re-scaled.