Unit 3: Sampling and Experimentation Flashcards
Population
The entire group we want information about
Sample
A part of the population that we use to gather information about the entire population
Census
Collects data from every individual in the population
Observational Study
observes individuals and measures variables of interest but doesn’t influence the responses
Experiment
purposefully imposes treatment on individuals to observe their responses
What kind of study is a sample survey?
Observational Study
Convenience Sampling
Definition: Using subjects that are readily available
Advantages: Easy and less expensive to collect
Disadvantages: Not representative of the population
(e.g.) standing outside the library and asking the first 10 students if they like school
Voluntary Response Sample
Obtained by allowing subjects to decide whether or not to respond
Advantages: easy to collect
Disadvantages: People who choose to respond probably have really strong opinions
Simple Random Sample
Individuals chosen from a population in such a way that each person has an equal chance of getting selected as everyone else
Advantages: Easy to accomplish by using a random number generator
Disadvantages: None
Stratified Random Sample
Divide population into groups of similar individuals (strata) and then take a simple random sample of each individual group
Advantages: May produce more exact information
Disadvantages: No appropriate if the strata you are using aren’t clearly defined.
Systematic Random Sampling
randomly select a starting point and then choose every nth person
Advantages: Every member has an equal probability of being selected
Disadvantages: Not every sample size of n has an equal chance of being selected
Without Replacement
When an item selected from a population can only be chosen onece
With Replacement
When an Item from a population can be chosen more than once.
Biased
systematically not representative of the entire population
Voluntary Response Bias
just using volunteers will probably not be representative of the entire population
Undercoverage
Certain groups within the population are left out of the process of choosing a sample
Non response
occurs when an individual for a sample can’t be contacted or won’t respond
Response Bias
biased caused by the behavior of the respondent or the interviewer
Untruthful Answers
People giving untruthful answers to sensitive questions, social norms, and interviewer bias
Ignorance
people will say dumb things when responding to questions they don’t know the answer to to make themselves look better.
Lack of Memory
People may not remember the answers to a question
Timing
The time that the survey is taken may have an impact on the answer.
Phrasing
Differences in a way a question is phrased could influence the answers
Sampling error
difference between the sampling result and the result of the entire population. error results from chance variation. It can be minimized by increasing the sample size
Non-Sampling Error
Occurs when sampling data are incorrectly, collected, recorded, or analyzed
Benefits of Experiements
- we are able to study the information that we want while controlling other variables
-allow us to study the combined effects of several factors
Factors
Explanatory variable in the experiment
Level
various groups that factors take
What are the four critical elements of experimental design
- Comparison
- Randomization
- Control
- Replication
Comparison
Make sure that groups you are comparing don’t differ greatly. prevent this using randomization
Randomization
Most important element of the experiment. Can do it in either the process of selection or process of distributing treatments
Control
Treated like the baseline for the experiment. Experimental Units in the control group will be in the same conditions as the other groups, however they will not receive treatment
Replication
Using enough experimental units so that effects caused by the treatment can be distinguished from effects caused by the sampling error.
Blinding
when the subject doesn’t know whether or not they are receiving the placebo
Double Blinding
When the subject nor experimenter know whether or not they are receiving the placebo
Completely Randomized Design
experimental units are assigned to treatment completely by chance.
-Treatment + Control groups will be completely equal in size
Lurking
When there’s an outside variable that controls each of the two variables and makes it appear as if they have a correlation
Confounding
When there is another is explanatory variable that you did not control that may be affecting the outcome, but you wouldn’t know.
Matched Pair Design
designs in which the same individual or two matched individuals are assigned both a control and a placebo. Should be conducted as double blind. If an individual takes part, the order in which they get the placebo and the real treatment should be random
Block Designs
Randomized design is taken on multiple broken down groups of the experimental units
What does random selection of individuals do? (inference)
Allows us to make inferences about the population
What does random assignment of treatments do? (inference)
Allows us to make inference about the cause and effect of two variables