Statistics Unit 11-12 (Sampling and Selection Bias)

Population Parameter

Summary taken from entire population

Sample Statistic

Summary taken from the data

Sampling Frame

a list of individuals from whom the sample is drawn

Representative Sample

a sample that accurately reflects the characteristics of the population as a whole. (Taken from SAMPLING FRAME)

Sample Size

How large the wanted SAMPLE is.

Sampling Variability

samples drawn at random generally differ from one another

Simple Random Sampling (SRS)

A sample in which every possible sample of desired size is equally likely to be chosen. (Randomness to the max)

What is an example of Simple Random Sampling

SAMPLING FRAME of 1000, wanting a sample size of 100. Assign a # to each of the #'s in those thousand and choose 100 randomly, ignoring repeaters.

Stratified Random Sampling

The population is divided into groups called STRATA. Then SIMPLE RANDOM SAMPLING is performed in each STRATA.

What is Stratified Random Sampling good for?

Reduces sampling variability that results solely from differences in the STRATA. (Individual part of a slice of cake)

Cluster Sampling

When sample is split into 2 or more CLUSTERS that represent the full population. Then select one cluster at a time and perform a CENSUS within each one. (whole slice of cake)

Census

the official count of a population

Systematic Sampling

Devising a system to choose our sample. “a system to your randomness” (Ex: choosing every 5th person)

Undercoverage Bias

occurs when some groups in the population are left out of the process of choosing the sample

Measurement Bias

Comes when using an improperly calibrate scale, causing us to collect inaccurate measurements

Response Bias

Poorly worded questions that could lead to different results

What are ambiguously worded questions?

Related to RESPONSE BIAS. When question has multiple right answers based on different topics.

Social Desirability Bias

Another types of RESPONSE BIAS. Happens when people say what they believe is the appropriate answer based on society and not what the person actually believes in.

nonresponse bias

When a major portion of individuals from the population just decides not to answer or partake in the survey.

Voluntary Response Bias

Bias happens happen when large group of individuals are INVITED to respond, then those who respond are counted.

Convenience Sampling

selection of individuals who are easiest to reach and survey. Can cause bias since only asking those convenient enough to answer.

Convenient sampling is related to what Bias??

Undercoverage bias

What does confound mean?

Confound

What is a lurking variable?

A factor that was not included in the experiment but is driving the ones that are.

Example of lurking variable

Although age is not mentioned in the question, if you were to graph shoe size and reading level, they would both go up together due to age.

Component

Outcome of a TRIAL. (Pair of an individual trial)

Example of Component

Voting for best foods with 100 voters, with only about 50% voting pizza and other 50% voting cheeseburger. Every voter is a COMPONENT. You then assign number from 0-99, with 0-49 being pizza, and 50-99 cheeseburger. Within this trial of 100 votes, every two digit number within that trial is a COMPONENT.

What is a trial?

Every time we repeat our simulation to get our components.

Example of trial

if 100 people are voting on which food is better, 50% chicken nuggs and 50% fish, 100 votes is your TRIAL. You would then examine the components to see the results of your trial.

What is the RESPONSE VARIABLE?

The answer/overall result from each trial

Retrospective Study

Observe Historical Data

Prospective study

Identify subjects in advance and collect data as events unfold

Observations show what relationships?

Strong association, NOT CAUSATION RELATIONSHIP

Factor (Explanatory variable)

Variable we use to manipulate

Response Variable

The variable that we measure

Subjects (participants)

Things being tested in experiment

Treatment

Combinations of specific LEVELS from all factors that an experimental unit receives

Most of the time, every factor has how many levels?

2 levels. Example: water temp is factor, levels are hot and cold.

Formula for treatments

#of levels in one factor x #of levels in other factor.

what are the 4 principles of experimental design?

CONTROL, RANDOMIZE, REPLICATION, and BLOCK

Control

Controlling all sources of variation (extraneous factors) other than factors we are testing. Make all conditions as similar as possible .

Randomize

Allows us to equalize the effects of UNKNOWN or UNCONTROLLABLE sources of variation. (Spreads sources of extraneous factors out)

Replication

Making sure each experiment is tested on multiple number of subjects, not just two. (Also have a controlled group to compare experiment)

Block

When we know attributes subjects have, causing us to split the similar individuals into blocks, than performing the experiment within the blocks.

Example of BLOCK

You want to evaluate cleanliness of Tshirts based on laundry detergent. You have 30 shirts, 18 are cotton, 12 are linen. Divide the shirts into different blocks, one for linen shirts, one for cotton shirts. Now conduct experiment like normal.

Matching

Used in RETROSPECTIVE and PROSPECTIVE STUDIES (observational)

How are matching and blocking similar

Same thing, except BLOCKING is for EXPERIMENTS and MATCHING is for OBSERVATIONS (BEMO)

Control group

Group used to compare results of other groups with treatment. (No treatment done at all in this group)

Blinding

Making our subject unaware of which treatment they are receiving.

Placebo

Common method of blinding. Make groups think they are receiving treatment, when in reality they are receiving nothing. (Acting as a control group)

Double blind

When everyone in groups are completely blinded. (No idea of any outcomes or possible RESULTS)

For real this time, what is a Confounding Variable?

Something that is associated with a factor, which affects the outcome (RESPONSE VARIABLE)