There are four common Random Sampling Methods: Random sampling* *Random sampling involves a chance process to select

*Random sampling involves a chance process to select individuals.

• Simple Random Sample (SRS): the simplest possible method of obtaining a sample. Every individual has an Simple Random Stratified Random
equal opportunity of being selected AND also every possible sample has the same chance of being selected. Three
common methods to collect a SRS is the lottery method (hat method), table of random digits, and using technology
(random number generator).
• Stratified Random Sampling: divide the population into subgroups (strata), then conduct a simple random sample from Every possible combination Groups are different,
individuals are similar.
of samples can occur.
each subgroup to attain your sample. Cluster Systematic
• Cluster Sampling: create clusters from the population, randomly select at least one cluster, and includes each member of
the selected cluster in the sample.
• Systematic Random Sampling: selects a sample from an ordered arrangement of the population by randomly selecting
one of the first k individuals and choosing every kth individual thereafter. Groups are similar, 2nd person was selected
individuals are different. and then every 3rd after
Observational studies
Observational Study: observes response to a variable. They do not influence the participants. Observational studies generally do not determine causation (only correlation).
• Retrospective: individuals are sampled and information is collected about their past.
• Prospective: individuals are followed over time and data about them is collected as their characteristics or circumstances change.

Experiments
Experiments observe responses to a variable, administer a treatment to observe the response, and attempt to determine causation (when random assignment takes place)!
The three common Experimental Designs are bolded and can be found below:
• Completely Randomized Design: Experimental units are assigned to the treatments completely at random.
• Block Design: Experimental units are assigned to treatments separately within each block. Block designs are similar to Stratified Sample. However, they are not the same.
 Matched Pairs Design: Block design that uses only blocks of size two and used to compare two treatments. In some designs, two very similar experimental units are
paired and the two treatments are randomly assigned within each pair. In others, each experimental unit receives both treatments in a random order.
Terms that should be known in Experimental Design: AP Tip for Experimental Design: Don’t be a minimalist!! When explaining your experimental design give
LOTS of information! For example: if you have 100 subjects and need to assign them to different treatment
• Experimental Units • Treatments groups, how are you going to do that? How are you randomly allocating the experimental units to the
• Factors/Explanatory Variables • Causation groups? Tell the AP Graders! For example: I will put all the names on slips of paper in a hat, shuffle well,
• Response Variables • Blinding and select the first 50 slips for the treatment group (in context) and then the next 50 for the placebo group
(in context). Then, be sure to include the response variable that will be collected.
• Random Allocation • Levels
Random Selection: allows us to generalize our findings to the population from which individuals were selected.
• Placebo Effect • Replication
• Control Group • Comparison Random Allocation: allows us to establish a cause-and-effect (causation) relationship in the experiment.
• Confounding Variables • Control

Sample vs. Population Bias (poor sampling)

A population in a statistical study is the entire group of individuals we want information Bias is systematically favoring a particular outcome (more favored than other outcomes).
about. Typically referred to as the target population. Different types of Bias when sampling:
• When the entire population is surveyed, this is called conducting a census. Convenience Sample: when a researcher chooses individuals for a sample who are
easy to access. This sampling method does not represent the actual population. (for
A sample is a subset of individuals in the population from which we collect data.
example, sampling the 1st 100 people in a store).
• When selecting a sample, in order to be representative of the population, Voluntary Response: occurs when the sample members are allowed to self-reflect or
they should be randomly selected. volunteer for participation in a study by receiving a general invitation. (for example,
posting a survey link on a website and asking people to respond).
Common mistakes Undercoverage: occurs when some members of the population are represented
inadequately in a sample. (for example, leaving out all 9th graders in a survey or
• Many students lose credit on the AP statistics exam for failing to adequately experiment when the target population was the entire school).
describe how they assign the treatments to experimental units in an Nonresponse: occurs when the individual selected for the sample can’t be contacted
or refuses to participate. (for example, telephone surveys leave out the homeless).
experiment. Most importantly, the methods the students use must be
Response Bias: a systematic incorrect response in survey. Generally, something in the
random. Also, the method must be described in sufficient detail so that the
survey design influences a particular response of an individual or group. (for example,
two knowledgeable users of statistics can follow the student's description and a teacher asking their students if they like them as a teacher will likely result in bias).
carry out the method in the exact same way.
Some students misuse the term voluntary response to explain why certain individuals
• Confusing stratified random sampling with randomized block design is do not respond to a sample survey. Not responding to a survey would be nonresponse.
another way that students mix up the language of sample surveys and
Bias is not just bad luck in one sample. It’s the result of a bad study design that will
experiments. These two ideas are alike in some ways, because they both
consistently miss the truth about the population in some way.
involve forming groups of similar subjects before random selection or random
assignment. Both methods also help to account for the variability created by * If you are asked to describe how the design of a sample survey leads to bias, you’re
other variables. However, stratified random sampling is done only when tak- expected to do two things: (1) describe how the members of the sample might
respond differently from the rest of the population, and (2) explain how this differ-
ing a sample from a population. Likewise, blocking happens only when assign-
ence would lead to an underestimate or overestimate. Do NOT identify the bias by
ing units to treatments and in an experiment.
name unless specifically asked to name the type of bias that is present.
• If you're asked to describe how the design of a sample survey leads to bias, you're expected to do two things: first, describe how the members of
the sample might respond differently from the rest of the population, and second explain how this difference would lead to an underestimate or
overestimate.

• Be careful not to misuse the term voluntary response to explain why certain individuals don't respond in a sample survey. When individuals chosen
for the sample can't be contacted or refused to participate this is nonresponse.

• If you are asked to identify a possible confounding variable in a given setting, you are expected to explain how the variable you chose (1) is
associated with the explanatory variable and is (2) associated with the response variable.

• Many students lose credit on the AP statistics exam for failing to adequately describe how they assign the treatments to experimental units in an
experiment. Most importantly, the methods the students use must be random. Also, the method must be described in sufficient detail so that the
two knowledgeable users of statistics can follow the student's description and carry out the method in the exact same way.

• In statistics, replication means “use enough subjects.” In other fields, the term replication has a different meaning. There must be at least two ex-
perimental units receiving each treatment to achieve replication.

• Many students confuse block and treatment groups. Blocks are not formed at random. Instead, they are formed by grouping experimental units that
are similar in some way that is expected to systematically affect their response to the treatments. This means that each block should be very
different from other blocks. However, a treatment group is formed at random, with the goal that the treatment groups be as similar as possible to
each other, except for the treatment itself.

• Confusing stratified random sampling with randomized block design is another way that students mix up the language of sample surveys and experi-
ments. These two ideas are alike in some ways, because they both involve forming groups of similar subjects before random selection or random
assignment. Both methods also helped to account for the variability created by other variables. However, stratified random sampling is done only
when taking a sample from a population. Likewise, blocking happens only when assigning units to treatments and in an experiment.

• Students often confuse the two types of inferences we can make: inferences about a population and inferences about cause-and-effect. Be sure to
understand how the role of randomization in a study helps to decide which type of inference, if any, is appropriate.

• Confounding is an important concept. You should be able to identify potentially confounding variables and when given a description of the study.

• Diagrams are a useful tool for organizing your thinking about an experiment, but a diagram cannot stand alone as a description of an experimental
design

• You need to be able to use the terms control groups, placebo, single-blind, and double-blind correctly and in context when describing an experi-
ment.

• When designing a double-blind experiment, it is important to describe the experimental unit who receives the treatment as not being aware of
which treatment they received and the individual evaluating the response variable not being aware either. Whether or not the individual who is
administering the treatment is aware or not is irrelevant. The goal is to prevent bias from the experimental unit and the individual who is collecting
the response variable.