Unit 12 Experimental Design Flashcards
What’s a population?
a population is the entire group of individuals whom you want to describe.
What’s a sample?
a sample is a smaller group of individuals you select FROM the population
What’s the sampling frame?
the sampling frame is a list of individuals from the population of interest from which the sample is drawn.
What happens if
- the sampling frame is NOT equal to the population of interest
- and is different from the population in some way that may affect the response variable?
Then the sample will be BIASED in some way.
When is a sample biased?
- the sampling frame is NOT equal to the population of interest
- and the sampling frame is different from the population in some way that may affect the response variable
What’s sampling variability?
sampling variability is the natural tendency of probability samples to vary.
Is sampling variability an error?
NO! sampling variability is not an error.
If sampling variability isn’t an error, then what is it?
sampling variability is the result of properly applied random sampling techniques.
What’s the sample size?
the sample size is the number of individuals you select from your sampling frame.
does the size of the population dictate the size of the sample?
NO!
What dictates the size of the sample?
ten percent rule dictates the size of the sample
What’s the ten percent (10%) rule?
the sample size should not be bigger than 10% of the population size.
What’s a census?
a census is the sample that’s the size of the entire population.
What’s a parameter?
a parameter is a number that describes an aspect of the population.
Do we know the parameter of a population?
No, we usually do not know the true value of a parameter, so we have to estimate it.
What’s a statistic?
a statistic is a number than describes an aspect of the SAMPLE DATA.
What do we use statistics for?
we use statistics to ESTIMATE values in the POPULATION.
What’s a sampling design?
a sampling design is the method we use to select the sample for a study.
What do all appropriate statistical sampling designs include?
all appropriate statistical sampling designs include the idea that chance, rather than choice, is used to select the sample.
What’re poor methods of selecting a sample? (2)
poor methods of selecting a sample:
- voluntary response sampling
- convenience sampling
What’s voluntary response sampling?
A voluntary response sample is a sample made up of volunteers.
What’s convenience sampling?
Convenience sampling is when you select subjects that’s close and convenient for you to sample
What’s the best defense against systematic errors in way that the data represents the population?
RANDOMIZATION is the best defense against systematic errors in way that the data represents the population.
What’re the appropriate selection methods?
appropriate selection methods:
- simple random sampling
- systematic random sampling
- stratified random sampling
- cluster sampling
When can you use cluster sampling?
you can use cluster sampling only when the clusters are all HETEROGENEOUS.
What’s heterogeneous?
HETEROGENEOUS means DIVERSE
What’s systematic random sampling?
systematic random sampling is when
- you assign all items a value and order them
- you select samples based on a system of intervals in a numbered population.
example of systematic random sampling?
systematic random sampling:
- Lucas can give a survey to every fourth customer that comes in to the movie theater.
- The fact that Lucas is giving the survey to every fourth customer is what makes the sampling systematic because there is an interval system.
What’s stratified random sampling?
stratified random sampling is when
- you divide a population into smaller groups known as strata based on shared characteristics
- then you choose some individuals from each strata
What’s stratified random sampling in summary?
some from all
- choosing some from all strata/clusters
What’s cluster sampling?
cluster sampling is when
- a population is broken up into clusters that represent the population in terms of diversity.
- you number each cluster
- you choose entire clusters based on randomly picking numbers
What’s cluster sampling in summary?
all from some
- choosing all individuals in a cluster from some clusters
What happens in a simple random sample?
in a simple random sample, every member of the population has an equal chance of being included in a sample of size “n”
True or False?
The simple random sampling technique eliminates bias
True, if you’re referring to SELECTION bias
What’s sample design or sampling technique?
sample design or sampling technique is the method you use to select participants and gather data.
What sampling techniques are prone to selection bias?
sampling techniques prone to selection bias:
- voluntary response
- convenience
What probability sampling techniques get rid of selection bias?
probability sampling techniques that get rid of selection bias:
- simple random sample
- stratified
- systematic
- cluster
What is the only method that eliminates all bias from a sample?
a CENSUS is the only method that eliminates all bias from a sample.
What’s bias?
bias is any systematic failure of a sample to represent its population of interest
What’re the types of response bias?
types of response bias:
- question wording
- interviewer bias
- order of choices
- honesty
What’s non-response bias?
non-response bias is when respondents differ in meaningful ways from nonrespondents.
What’s under-coverage bias?
Undercoverage occurs when some members of the population are inadequately represented in the sample.
What’s voluntary response bias?
voluntary response bias occurs when sample members are self-selected volunteers, as in voluntary samples
How to describe how you would select an SRS?
- describe the picking process: include the phrase “without replacement” and “repetition.
- include a conclusion: what do the things you pick represent?
- ex: these are the 8 people selected to go to the US Capitol.
What’s an observational study?
an observational study is when researchers
1. OBSERVE individuals 2. record variables of interest.
What can you not determine from an observational study?
you can’t determine a CAUSE AND EFFECT relationship from an observational study
What’s an experiment?
an experiment is when you
- deliberately apply treatments
- in addition to observing
- and recording variables of
What can you perform to determine a cause and effect relationship?
you can perform an EXPERIMENT to determine a cause and effect relationship.
How to use an experiment to determine cause and effect?
you use an experiment to determine cause and effect between the explanatory and response variables.
What’s an explanatory variable?
an explanatory variable is the INDEPENDENT variable
What’s a response variable?
a response variable is the DEPENDENT variable
What’s the treatment?
the treatment are the CHOICES for the independent variable
Identify the population of interest, explanatory variable, the treatment, and the response variable.
Researchers believe that cranberry juice may be as effective in reducing a person’s blood pressure as a conventional drug (Drug A), which is much more expensive and has more unwanted side effects.
- population of interest: people with high blood pressure
- explanatory variable: what is consumed
- treatment: cranberry juice, Drug A
- response variable: blood pressure
What’re the steps to designing an experiment?
- “select subjects” (from the population of interest)
- decide which treatment to give
- state how you’ll randomly assign it
- ex. coin flip
- specific the treatment
- amount, frequency of treatment, etc.
- detail how you’re collecting the response variable
- how you’re going to measure it
- “compare variables.”
In designing an experiment, what do you use blocks for?
use blocks to reduce variability and bias.
What’s binding?
blinding is when the subject does not know which treatment (if any) that they are receiving
What’s double-blind?
double-blind is when neither the experimenter nor the subjects know what treatments are being given.
What’s the placebo effect?
the placebo effect is when the subject isn’t given a treatment but responds bc they think the treatment is effective.
What’s blocking?
blocking is when you sort a POPULATION into GROUPS based on criteria in EXPERIMENTAL DESIGN
What’s stratifying?
stratifying is when you sort a POPULATION into GROUPS based on criteria in SAMPLING.
What’s the difference between blocking and stratifying?
- they both involve sorting a POPULATION into GROUPS based on criteria
- but, blocking is for experimental design, and stratifying is for sampling
What’re the benefits of observational studies?
benefits of observational studies:
- less expensive
- takes less time
- the situation is more “real life”
What’re the cons of observational studies?
cons of observational studies:
- lurking variables
- confounding variables
What’re lurking variables?
lurking variables are variables that impact the response variable but are not measured as part of the study of interest
What’s the relationship between lurking and confounding variables?
lurking variables are a subset of confounding variables
What’re confounding variables?
confounding variables are two variables that BOTH have an impact but the investigator can’t separately identify their impact on the response variable.
How do you minimize the effects of confounding and lurking variables?
minimize the effects of confounding and lurking variables by doing observational studies on subjects that are as HOMOGENOUS as possible.
What’s the difference between observational studies and experiments?
difference between observational studies and experiments is how much CONTROL you have over the variables.
How is an observational study different from an experiment?
observational study:
- you simply collect data based on what is seen and heard
- you infer based on the data collected
- you DO NOT interfere with the subjects or variables in any way (you don’t add a treatment)
Why can you not determine cause and effect from an observational study?
because there may be LURKING variables.
What’re the key elements of experimental design? (important)
key elements of experimental design:
- control
- randomization
- replication (can it be replicated?)
What’s the basic design method? (important)
basic design method:
- you randomly select individuals
- you randomly assign treatments
- via coin flip, etc.
- you measure the treatments
- you administer the treatments
- time elapses and treatment finishes
- you measure the treatments again
- you analyze data
- you make conclusions.
What’s blocking?
Blocking is when you sort a experimental units or subjects that are known before the experiment to be similar in some way that is expected to affect the response to the treatments by group.
What’s a block?
a block is A group of experimental units or subjects that are known before the experiment to be similar in some way that is expected to affect the response to the treatments.
What’s an example of blocking?
males and females may respond to a diet supplement differently, so you block the experiment by gender.
What’s matched pairs design a part of?
matched pairs design is a FORM OF BLOCKING
What’s a form of blocking?
matched pairs design is a form of blocking
What’s matched pairs design?
matched pairs design is when 1. you give two people or objects who are similar or identical DIFFERENT TREATMENTS
2. then you make them go through SIMILAR CONDITIONS
Example of matched pairs design?
matched pairs design:
- one shoe receives one shining treatment
- another shoe receives another shining treatment
- you see how well they hold up with daily wear in a week.
What’s a census?
census is when you collect data from ALL of the members of a population in the study.
What’s quota sampling?
quota sampling is when interviewers interview a fixed quota or number of members of the population.
What’re the chances involved in simple random sampling?
chances involved in simple random sampling:
- you have an equal chance of choosing every sample of size “n.”
- every member of the population has an equal chance of being included in the sample.
What’s simple random sampling?
simple random sampling
- involves choosing individuals at random from the population WITHOUT replacement
- you choose a sample of size “n” from the population
What’s bias?
bias is a systematic error that favors a particular segment of the population or encourages certain results
What type of sampling technique eliminates selection bias?
the SIMPLE RANDOM sample eliminates selection bias
Why does the simple random sample eliminate selection bias?
since all subsets of a given size have the same chance of being selected from the population.
What’s a strata?
strata are different groups of your population
What’s the multi-stage random sample?
multi-stage random sample is when you repeatedly select simple random samples within previously selected simple random samples
What’s an example of the multi-stage random sample?
- a wholesaler selects a simple random sample of cases of fruit from a shipment.
- he then selects a simple random sample of fruit within one case
What’s a sampling frame?
a sampling frame is the list of possible subjects who could be selected in a sample.
What happens if the sampling frame isn’t equal to the population?
the sample will be biased in the same way the sampling frame is biased.
What’s a random sampling error?
a random sampling error is an error that occurs because of chance variation.
What’s a sampling method error?
a sampling method error is an error that occurs because of the sampling method you chose.
Example of a sampling method error?
you choose a convenience sample
What’s a non-sampling method error?
a non-sampling method error is an error that members in a sample cause
What an example of a non-sampling method error?
a non-sampling method error is when dishonest responses will create error in the sample.
What’s a control?
a control includes the units that do not receive the independent variable or treatment
When are results statistically significant?
if the differences between the groups are greater than what would be expected through chance variation alone.
When can you investigate a cause and effect relationship?
when an experiment meets the following criteria:
- control
- randomization
- replication
What’re historical controls?
historical controls are when you compare subjects who receive the treatment to historical records of subjects who didn’t receive the treatment
why do researchers use historical controls?
it’s cheaper and less time consuming
When do you use block design?
block design when one specific characteristic of the experimental units will likely affect the results of the experiment
what’s one example of when you would use block design?
you would use block design if when testing blood pressure medication, you anticipate gender affecting the results, you create separate blocks for males and females
What’s the purpose of matched pairs design?
To control for lurking variables