Chapter 4 – How to Get a Good Sample
Sample survey is a subgroup of a large population that is questioned on a set of topics. The results are
used like they represent a larger population.
An experiment measures the effects of manipulation on the environment in some way. A randomized
experiment is when the manipulation is assigned to participants on a random basis.
The explanatory variable is the result of the feature being manipulated. The outcome (response)
variable is the outcome.
Randomized experiments allow us to determine cause and effect. The randomizing is to make the
groups approximately equal to each other. Differences in the outcome that are large enough can rule
out natural variability.
Observational study is when the manipulation occurs naturally. Reliance on natural manipulation
disables us from finding cause and effect because it is hard to have a control group.
Case-control studies attempt to include a control group. An advantage to observational studies is that
participants just do what they do naturally.
Meta-analysis is a quantitative collection of studies done on a similar topic. It combines information
from different researchers to find patterns or effects that weren’t available from the individual studies.
Case study is an in-depth examination of one or a small group of individuals. Don’t assume you can
extend findings of a case study to a person or situation other than the one studied. They are usually
used in rare or unrepresentative cases.
Unit – single individual or object being measured.
Population – entire collection of units.
Sample – collection of units we measure.
Sampling Frame – list of units the sample was chosen from.
Sample Survey – measurements are taken on a subset of units from the population.
Census – entire population is measured.
Sample surveys are used to estimate the proportion of people who have a certain trait or opinion.
Newspapers/Magazines often conduct sample surveys on topics of interest. These surveys can be super
accurate. The number that measures accuracy is the margin of error. Margin of error is usually 1/sqrt(n).
Sqrt = square root. N = # of people. The advantages of sample surveys are that they are accurate, can be
used when a census isn’t possible (sample of blood vs. all your blood), and speed.
The ability of a small sample to represent a large population doe