Chapter 4: Designing Studies

What is a population?

The entire group of individuals we want information about.

What is a census?

Collecting data from every individual in the population.

What is a sample?

A subset of individuals from the population that we collect data from.

What is a convenience sample?

Choosing individuals who are easiest to reach.

Why is convenience sampling biased?

It often produces unrepresentative data.

What is a voluntary response sample?

People choose themselves by responding to a general invitation.

Why are voluntary response samples biased?

People with strong opinions are more likely to respond.

What is bias?

When a study consistently overestimates or underestimates the value being measured.

What is random sampling?

Using chance to determine which members of a population are selected.

What is a simple random sample (SRS)?

Every possible group of n individuals has an equal chance of being selected.

How do you choose an SRS using slips of paper?

Write names on identical slips, mix them in a hat, and draw randomly.

How do you choose an SRS using technology?

Label everyone from 1 to N and use a random number generator to select n different numbers.

What calculator command generates random integers?

randInt(x, y).

How do you choose an SRS using Table D?

Label everyone with equal-digit numbers, read groups of digits, and skip duplicates or unused labels.

What is a stratified random sample?

Divide the population into similar groups (strata), randomly sample each group, and combine the samples.

What are strata?

Groups of similar individuals within a population.

What is the advantage of stratified sampling?

It provides more precise estimates because all important groups are represented.

What is a cluster sample?

Divide the population into groups, randomly select clusters, and include everyone in those clusters.

What are clusters?

Groups of individuals located near each other.

What is the difference between strata and clusters?

Strata are similar within but different between. Clusters are different within but similar between.

What is undercoverage?

When some members of the population cannot be selected for a sample.

What is nonresponse?

When someone selected cannot be contacted or refuses to participate.

How can question wording cause bias?

Leading or confusing questions can influence people's answers.

Do larger random samples provide better information?

Yes, larger random samples generally provide more reliable information.

What is an observational study?

Observing individuals and measuring variables without influencing their responses.

What is an experiment?

Deliberately imposing a treatment on individuals to measure their responses.

What is the difference between an observational study and an experiment?

An experiment imposes treatments; an observational study does not.

What is confounding?

When the effects of two or more variables on a response cannot be distinguished.

What is a treatment?

A specific condition applied to individuals in an experiment.

What are experimental units?

The smallest collection of individuals to which treatments are applied.

What are subjects?

Experimental units that are human beings.

What is random assignment?

Using chance to assign experimental units to treatments.

What is the purpose of comparison?

To compare two or more treatments.

What is the purpose of random assignment?

To create roughly equivalent groups by balancing other variables.

What is the purpose of control?

To keep other variables that might affect the response the same for all groups.

What is the purpose of replication?

To use enough experimental units to distinguish treatment effects from chance differences.

What is a completely randomized design?

An experiment where experimental units are assigned to treatments completely by chance.

What is the placebo effect?

When someone experiences a response because they believe they are receiving a treatment, even if they receive an inactive treatment.

What is blinding?

Keeping subjects or researchers unaware of which treatment subjects receive.

What is a double-blind experiment?

Neither the subjects nor the people interacting with them and measuring responses know which treatment was received.

What does statistically significant mean?

An observed effect so large that it would rarely occur by chance.

What is a block?

A group of experimental units that are similar in a way that may affect the response.

What is a randomized block design?

Randomly assigning treatments separately within each block.

What is a matched pairs design?

A blocking design used to compare exactly two treatments.

How does a matched pairs design work?

One subject receives both treatments in random order, or two similar subjects are paired and treatments are randomly assigned within each pair.

What is the purpose of blocking?

To account for differences among experimental units and reduce unwanted variation.

What is causation?

A direct cause-and-effect relationship between two variables.

What is common response?

When two variables are associated because both are affected by a lurking variable.

What is confounding between variables?

When the effects of explanatory variables or lurking variables cannot be distinguished.

What is a lurking variable?

A variable not included in the study whose effects on the response cannot be distinguished from those of an explanatory variable.

What is a confounding variable?

A variable whose effect on the response cannot be distinguished from another explanatory variable's effect.

What are the three possible explanations for an association between two variables?

• Causation

• Common response

• Confounding

What is the best way to establish causation?

A carefully designed experiment, although experiments are not always practical, ethical, or possible.

Convenience sample

Easy-to-reach people

Voluntary response

People choose themselves

SRS

Equal chance for every possible sample

Stratified

Similar within, different between

Cluster

Different within, similar between

Observational study

No treatment imposed

Experiment

Treatment imposed

Confounding

Effects cannot be separated

Common response

Lurking variable affects both

Matched pairs

Compares two treatments

Double-blind

Neither subjects nor researchers know

Statistical significance

Unlikely to happen by chance

What are the four principles of a well-designed experiment?

Comparison, Random Assignment, Control, and Replication.

What is comparison?

Using a design that compares two or more treatments.

What is the purpose of comparison?

To determine whether different treatments produce different responses.

What is random assignment?

Using chance to assign experimental units to treatments.

What is the purpose of random assignment?

To create roughly equivalent groups by balancing the effects of other variables.

What is control?

Keeping other variables that might affect the response the same for all groups.

What is the purpose of control?

To prevent outside variables from influencing the results.

What is replication?

Using enough experimental units in each treatment group.

What is the purpose of replication?

To distinguish actual treatment effects from chance differences between groups.

Why are all four principles important?

They help make experiments fair, reliable, and produce trustworthy results.

What is the difference between a population and a sample?

A population is the entire group being studied; a sample is a smaller subset of that group.

Why is random sampling important?

It helps reduce bias and makes the sample more representative of the population.

What makes a sample representative?

It accurately reflects the characteristics of the population.

What is the disadvantage of an SRS?

It may not include enough individuals from smaller or important subgroups.

What is the disadvantage of stratified sampling?

You must know how to divide the population into appropriate strata.

What is the disadvantage of cluster sampling?

Selected clusters may not represent the entire population well.

Can a large sample still be biased?

Yes. A large sample does not eliminate bias caused by poor sampling methods.

What is the difference between sampling error and bias?

Sampling error is natural variation from selecting a sample; bias consistently pushes results in one direction.

What is an explanatory variable?

A variable that may explain or cause changes in the response variable.

What is a response variable?

The variable measured to determine the effect of the treatment.

What is the difference between explanatory and response variables?

The explanatory variable may influence the response variable.

Why can't observational studies usually establish causation?

Because researchers do not control treatments or randomly assign individuals.

Can experiments always be used to establish causation?

No. Some experiments are unethical, impractical, or impossible.

Why do experiments use placebos?

To determine whether results are caused by the actual treatment rather than expectations.

What is a single-blind experiment?

An experiment in which the subjects do not know which treatment they receive.

What is the difference between single-blind and double-blind?

Single-blind keeps subjects unaware; double-blind keeps both subjects and researchers interacting with them unaware.

Why is random assignment different from random sampling?

Random sampling selects individuals from a population; random assignment places experimental units into treatment groups.

When should researchers use blocking?

When experimental units have known differences that may affect the response.

Why is blocking helpful?

It reduces variation caused by differences among experimental units.

What happens in a randomized block design?

Individuals are divided into blocks, then randomly assigned to treatments within each block.

When is a matched pairs design appropriate?

When comparing exactly two treatments.

What are the two ways to conduct a matched pairs design?

One subject receives both treatments in random order, or two similar subjects are paired and randomly assigned different treatments.

What are the three explanations for an association between two variables?

Causation, common response, and confounding.

How does causation differ from common response?

Causation means one variable directly affects another; common response means a lurking variable affects both.

What is the difference between a lurking variable and a confounding variable?

A lurking variable is not included in the study design. A confounding variable has effects that cannot be separated from another explanatory variable.