Biostatistics Lecture Notes

Comprehensive vocabulary flashcards covering fundamental biostatistical terms, study designs, summary metrics, sampling techniques, probability concepts, and normal distributions.

Biostats

The application of statistical principles to medical, public health, and biological applications.

Quantitative Data

Numbers and measurements.

Histogram

A graph that shows how numerical data is spread out, used when showing the distribution of quantitative data.

Qualitative Data

Data that describes a category or quality (the same as categorical).

Categorical Variables

Variables that tell you what group someone belongs to, rather than measuring how much of something someone has.

Bar Chart

A chart featuring separate categories used to see how many people fall into each category.

Ordinal Variables

Categorical data organized in a numerical order (e.g., cancer stages I, II, III, IV).

Observational Studies

Studies in which researchers observe exposures and outcomes without assigning the exposure.

Experimental Studies

Studies where investigators assign a treatment or intervention.

Cohort Study

A study that begins with people with an outcome and without, tracking both past exposures; the odds ratio is used frequently.

Odds Ratio

A measure that helps researchers evaluate how strongly an exposure is associated with an outcome; values less than $1$ indicate lower odds in the exposed group, values greater than $1$ indicate higher odds, and an odds ratio equal to $1$ means both groups have equal odds.

Case-Control Study

A study that begins with the exposed group and follows outcomes over time.

Risk

The proportion of people who develop an outcome during a specific time period, calculated as: $\frac{\text{\# of people who develop the risk}}{\text{total number of people who are at risk at the start}} = \text{Risk (\%)}$.

Relative Risk

A calculation that compares the risk of the exposed group to the risk in an unexposed group: $\frac{\text{Risk in exposed (\%)}}{\text{Risk in unexposed (\%)}}$, showing how many times greater the risk of exposure is in the exposed group.

Intention to Treat

Analyzing participants according to their original randomized groups, even if they stop treatment, switch treatments, or do not fully follow instructions; it helps preserve randomization and reduce bias.

Loss to Follow-up

A situation where researchers do not know what happened to some participants, which makes the study results not accurate.

Prevalence

How many people have the disease at a certain point in time, calculated as: $\frac{\text{Existing cases}}{\text{population}}$.

Incidence

How many new cases develop during a time period; a low prevalence with high incidence can occur when disease duration is short (e.g., rapid recovery or death).

Cumulative Incidence (Risk)

The proportion of people at risk at the beginning who develop the disease during a specific period of time, calculated as: $\frac{\text{New cases}}{\text{people at risk at the start}}$.

Risk Difference

The difference between two risk groups (exposed and unexposed), calculated as: $\text{Risk(exposed)} - \text{risk(unexposed)}$.

Population Attributable Fraction

Describes the proportion of all cases in the population attributable to the exposure, calculated as: $\frac{\text{Cases attributed to exposure}}{\text{all cases}}$.

Person-Time

The sum of the time each participant contributes while at risk; useful when follow-up times differ.

Mean

The average of a set of numbers.

Median

The middle data value of a set of numbers; if there is an even number of numbers, it is found by taking the average of the two middle values.

Mode

The most frequent value in a dataset.

Outlier

A value that is unusually far away from the others; it can pull the mean upward, making it higher than it should be, while the median is much less affected.

Skew

Describes data when it is unevenly distributed, characterized by a longer tail on one side.

Right-Skewed

A distribution where most values are low, but a few values stretch the distribution to the right.

Left-Skewed

A distribution where most values are high, but a few values stretch the distribution to the left.

Symmetric Data

Data where values are evenly distributed around the center, and the mean and median are often similar.

Standard Deviation

Measures how much values typically vary around the mean; small values indicate data points close to the mean, while large values indicate data points are more spread out.

Interquartile Range (IQR)

Measures the spread of the middle $50\%$ of the data, calculated as: $\text{IQR} = Q_3 - Q_1$.

Simple Random Sampling

A sampling method where every person has a chance of being randomly selected.

Systematic Sampling

A sampling method where every $k\text{th}$ person on a list is selected after choosing a starting point.

Stratified Sampling

A sampling method where the population is divided into subgroups (strata) and people are sampled from each subgroup.

Convenience Sample

A sampling method where researchers select people who are the easiest to reach.

Sampling Variability

The difference between estimates in samples that occurs because each sample contains different people.

Sampling Error

Tells how much a sample estimate would vary from sample to sample if samples were repeatedly taken from the same population using the same method.

Standard Error

Reflects that different samples produce slightly different estimates, which might fall above or below the true value.

Systematic Bias

A flaw where a sampling method repeatedly favors certain people or leaves out important groups, pushing the results in a particular direction.

Margin of Error

Describes the estimated amount of uncertainty around a sample estimate at a specified confidence level.

Probability

Tells how likely something is to happen, taking a value between $0$ and $1$ ($0$ is impossible, $0.5$ represents a $50\%$ chance, and $1$ is certain).

Conditional Probability

Finding the probability of something happening within a specific group, denoted as $P(A|B)$ ('the probability of A, given B').

Sensitivity

Correctly identifying people with a disease, defined as: $P(\text{positive test} \mid \text{disease present})$.

Specificity

Correctly identifying people without the disease, defined as: $P(\text{negative test} \mid \text{disease absent})$.

Independent Events

Two events where one event happening does not change the probability of the other event; calculated as $P(\text{both happen}) = P(A) \times P(B)$ and $P(\text{all work}) = p^n$.

Normal Distribution

A bell-shaped curve where most values are near the average and fewer values are far away; $68\%$ of data lies within $1\text{ SD}$, $95\%$ lies within $2\text{ SDs}$, and $99.7\%$ lies within $3\text{ SDs}$ of the mean.

Z Scores

Tells how many standard deviations a value is from the mean, calculated as: $\frac{\text{Observed value} - \text{mean}}{\text{SD}}$.

Z Tables

Tables that give the proportion of values below a given Z score; the proportion above is calculated as $1 - \text{table value}$.

Percentiles

Tells the percentage of values that are at or below a particular value.