Stats Vocab - Unit 3

Sample Survey

An inquiry that asks questions of individuals drawn from a population (such as a poll)

Census

A sample that consists of the entire population

Population

The entire group about whom a statistician hopes to learn

Sample

A portion of a population examined to learn about that population

Population Parameter

A numerical attribute for an entire population (rarely know this)

Sample Statistic

A numerical attribute calculated for sampled data and used to infer the associated population parameter

Sample Size

The number of individuals, n, in a sample

Randomization

Each individual is given a fair, random chance of selection

Representative

The sample statistic(s) computed from a sample accurately reflect(s) the corresponding population parameter(s)... you took a "good sample"

Simple Random Sampling (SRS)

Each combination of n individuals in a population has an equal chance of selection... often done by assigning a number to each individual and generating n random numbers

Stratified Random Sampling

Sampling method where the population is divided into several homogeneous subsets (called strata) and THEN random samples are proportionally drawn from each stratum

Cluster Sampling

Sampling method where entire groups or "clusters" or subjects are selected at random; the clusters must be heterogeneous and individually representative of the population

Systematic Sampling

Sampling method in which subjects are listed as one subject is selected, followed by each of the next nth subjects

Sampling Variability

The natural tendency of randomly drawn samples to differ from one another

Sampling Bias

Systematic failure of a sampling method to properly represent its population

Voluntary Response Bias

Individuals must volunteer or take some prior action to be included in a sample

Convenience Sampling Bias

Subjects chosen for a sample are simply those conveniently available due to some factor

Undercoverage Bias

Some portion of the population is not sampled at all or has a reduced chance of being included in the sample

Nonresponse Bias

A substantial fraction of those properly sampled fail to respond

Response Bias

Anything in a survey design that influences the responses, such as wording survey questions in a way that suggests a favored response or perceived pressure in the survey

Observational Study

Study in which the researchers do not assign treatments, they simply observe them

Retrospective Study

Observational study in which subjects are selected and then their previous conditions or behaviors are observed

Prospective Study

Observational study in which subjects are selected and then followed to observe future outcomes

Experiment

An experimenter controls factor levels to create treatments, randomly assigns subjects to those treatments, replicates the experiment with all of those subjects, and then compares the responses of the subjects

Completely Randomized Design

Every subject and combination of subjects has an equal chance of receiving each treatment

Factor

An explanatory variable whose levels are controlled by an experimenter

Response Variable

The variable whose resulting values are compared across different treatments

Level

The specific value that the experimenter chooses for a factor

Treatment

The controlled circumstances applied to experimental groups, made of different levels of a single factor or combinations of two or more factor levels

Control Group

A baseline treatment level that provides a basis for comparison; often either a well-understood level or a null treatment

Subject/Experimental Unit

Individuals or objects on whom an experiment is performed

Randomization

An experiment must assign subjects to treatment groups at random

Replication

Each treatment must be repeated for multiple subjects

Statistically Significant

When an observed difference is too large to believe that is was likely to have occurred naturally

Single-Blind

When either the subjects themselves or the administrators/evaluators in an experiment are unaware of which subjects were assigned to each treatment group

Double-Blind

When both subjects themselves and the administrators/evaluators in an experiment are unaware of which subjects were assigned to each treatment group

Placebo

A treatment known to have no effect, administered to the control group so that all groups experience the same condition (such as a sugar pill)

Placebo Effect

A perceived reaction to an ineffective treatment... mist compare to a placebo to be sure that the observed effect of a treatment is not simply due to the placebo effect

Blocking

Grouping subjects together by uncontrollable attributes, thus allowing an experimenter to isolate the variability attributable to differences in the blocks and see the differences caused by the treatments more clearly

Randomized Block Design

Experimental design in which randomization occurs only within blocks; treatment groups are not mixed for different blocks

Confounding Variable

When the levels of one factor are associated with the levels of another factor and this their effects cannot be separated

Matched Pairs Design

Experimental design where the same individual for multiple treatment trials or individuals with similarities not under study are paired together when undergoing the experiment, hopefully reducing the variation in the data from other factors