Week 5 Notes: Association, Relative Risk, Confidence Intervals, and NHST

Vocabulary flashcards covering key concepts from Week 5: association testing, chi-square, relative risk, confidence intervals, NHST, study designs, and related statistical concepts.

Null hypothesis significance testing (NHST)

A framework for testing whether observed results could be due to random chance by comparing a null hypothesis to an alternative, using a predefined alpha and p-values.

Association

A relationship between two nominal or ordinal variables; not the same as correlation (which applies to continuous data).

Pearson's chi-square test

A nonparametric test used to assess whether two nominal/ordinal variables are related in a contingency table; does not require normal distribution.

Nominal variable

Categorical data with no inherent order (e.g., handedness: left/right; aphasia: yes/no).

Independent variable

The variable presumed to influence the other in an association (e.g., handedness).

Dependent variable

The outcome that may be influenced by the independent variable (e.g., aphasia).

2x2 contingency table

A table with two binary variables (two rows and two columns) used to summarize observed frequencies for chi-square tests.

Observed frequencies

The actual counts observed in the data for each cell of the contingency table.

Expected frequencies

The counts that would be expected in each cell if the null hypothesis were true.

Significance level (alpha)

The threshold for declaring statistical significance, commonly set at 0.05.

Random sample

A sample where every member of the population has an equal chance of selection.

Independence (chi-square assumption)

Observations must be independent with no duplicates or linked responses.

Expected frequencies ≥ 5

Chi-square validity requirement that each cell's expected count is at least five.

Relative Risk (RR)

The ratio of risk of an outcome in exposed vs non-exposed groups; RR > 1 indicates higher risk with exposure, RR < 1 indicates lower risk.

Confidence Interval (95% CI) for RR

A range around the RR within which the true population value lies with 95% confidence; if it includes 1, the result may not be statistically significant.

Practical significance

The real-world importance or meaningfulness of an effect, not determined by p-values.

Prospective cohort

A cohort design where participants are followed forward in time to observe outcomes.

Retrospective cohort

A cohort design that uses existing records to examine outcomes after the fact.

Clustered bar chart

A bar chart that groups data by a second variable to visually compare proportions across categories.

P-value

The probability, under the null hypothesis, of obtaining data as extreme or more extreme than observed; used to assess statistical significance.

Exposure

In cohort studies, the factor being studied for its effect on the outcome (e.g., left-handedness); can be beneficial or harmful.

Odds Ratio (OR)

The odds of an outcome in the exposed group divided by the odds in the non-exposed group; commonly used in case-control studies and related to relative risk.