Introduction to Data and Sampling Flashcards

Vocabulary terms and definitions from introductory statistics covering data types, study designs, and sampling methods.

Summary statistic

A single number that condenses a lot of data, such as a proportion.

Treatment group

The group in a study that receives the intervention being tested.

Control group

The comparison group that does not receive the intervention; it serves as the reference point.

Data matrix

Data organized as $\text{rows (cases)} \times \text{columns (variables)}$.

Case / observational unit

One row in a data matrix representing one entity measured, such as a patient or a county.

Variable

One column in a data matrix representing a characteristic measured on every case.

Numerical variable

Values where arithmetic is meaningful.

Continuous numerical variable

Numerical values over a range, such as height or rate.

Discrete numerical variable

Numerical values that are counts or jumps, such as population or \text{# siblings}.

Categorical variable

Values that represent categories.

Level

A possible value or category of a categorical variable.

Nominal categorical variable

Categorical values with no natural order, such as state names.

Ordinal categorical variable

Categorical values with a natural order, such as "below hs" reaching to "bachelors".

Scatterplot

A graph of two numerical variables, with one dot per case.

Associated (dependent) variables

Variables that show a discernible pattern together.

Positive association

A relationship where both variables move in the same direction.

Negative association

A relationship where one variable rises as the other falls.

Independent variables

Variables with no evident relationship; variables are either associated or independent, never both.

Explanatory variable

The variable suspected of affecting the other.

Response variable

The variable suspected of being affected by the explanatory variable.

Observational study

Data collected without interfering; it shows association only.

Experiment

A study where treatments are actively assigned to subjects.

Randomized experiment

An experiment where subjects are assigned to groups at random, which licenses causal claims.

Population (target population)

The full set of cases the question is about.

Sample

A subset of the population that is actually measured.

Anecdotal evidence

Haphazard data from a few striking cases that is usually unrepresentative.

Bias

Systematic skew that makes a sample unrepresentative.

Simple random sample

A sample where every case has an equal chance of being selected and selections are unconnected, like a raffle.

Non-response bias

Skew that occurs when sampled people do not respond.

Convenience sample

A potentially unrepresentative sample where only easily-reached cases are included.

Observational data

Data with no treatment applied or withheld.

Confounding variable (lurking variable)

A variable correlated with both explanatory and response variables; the reason observational data cannot prove causation.

Prospective study

A study that follows cases forward as events unfold.

Retrospective study

A study that looks backward through records after events have occurred.

Stratified sampling

Splitting the population into similar groups (strata), then random-sampling within each stratum.

Cluster sampling

Splitting the population into clusters, picking a few whole clusters, and taking all cases within them.

Multistage sampling

A process similar to cluster sampling, but random-sampling is performed within each chosen cluster.

Sample statistic

A number computed from the sample used as an estimate.

Population parameter

The true value for the whole population, which is usually unknown.