Looks like no one added any tags here yet for you.
Observational Study
no treatment, can’t show causation
treatments
what is done/not done to experimental units
Experimental units
who/what treatment is imposed on
Factors
explanatory variables
levels
value of factors
control group
used to provide baseline data for comparisonb
blinding
when subjects (single-blind) and/or experimenter (double) who interact are unaware of what treatment is given
placebo effect
when a fake treatment (placebo) works.
key principles of experiments
Comparison ( 2+ treatments)
Random Assignment
Control ( keep all other variables beside treatments constant)
Replication ( use enough expSt units to distinguish difference )
Statistically Significant
if the likelihood of it occurring by chance alone is less than 5%. (larger sample sizes decrease the variability of estimates and make it easier to determine statistical significance
How to Describe an Experiment
State subjects if not yet listed
Justify blocks/pairs similar if needed - choose one
Randomly assign to treatments stating how many go to each
Repeat if needed for other blocks
State what you are comparing in context
Random sampling
allows us to generalize the results to the population from which we sampled
Random assignment
allows us to say a treatment causes changes in the response variables
population
the pool of individuals from which a statistical sample is drawn for a study
sample
the specific group that you will collect data from
sample
the specific group that you will collect data from
parameter of interest
a statistical value that gives you info about the research sample or population being studied
Simple random sample
choose a group from the population so that every individual and every group of individuals is equally likely to be chosen.
Pro: removes all hint of bias
Con: difficulty gaining access to a list of a larger population, time, costs, and that bias can still occur under certain circumstances
Cluster sample
splits population into groups that are different. Randomly selects groups + samples everyone in the group. Each cluster is a mini representation of the entire population.
Pro: requires fewer resources and is more feasible
Con: causes biased samples and high sampling error
Stratified sample
split the population into groups that are the same. Chose an SRS from each strata
Pros: It ensures each subgroup within the population receives proper representation within the sample
Con: it is unusable when researchers cannot confidently classify every member of the populationSys
Systematic Sample
chose a random starting point. Use equal intervals to move to the next individual until you have as many as you need
Pro: eliminates the phenomenon of clustered selection and a low probability of contaminating data
Con: over/underrepresentation of particular patterns and a greater risk of data manipulation
Convenience sample
Selects individuals from the population who are easy to reach.
Pro: cheap, efficient, and simple to implement
Con: produces unrepresentative data
Voluntary response
People choose whether or not they want to be involved
Pro: an inexpensive way to conduct a study as data is very easy to gather
Con: the researcher has no control over the makeup of the sample. Untrustworthy
Observational Study
observes individuals + measures variables of interest; doesn’t attempt to influence the responses
Pro: provide critical descriptive data and information on long-term efficacy and safety that clinical trials cannot provide, at generally much less expense
Con: lower standard of evidence than experimental studies, are more prone to bias and confounding, and cannot be used to demonstrate causality
sampling bias
the design of a study shows bias if it’s likely to under/overestimate the value you want to know. It’s going to systematically fail due to a faulty design, resulting in the over/underestimations
sampling variability
the idea that different random samples of the same size from the population produce different estimate. Reduce sampling variability by sample size, what we call error.
confounding
occurs when 2 variables are associate in such a way that their effects on a response variable cannot be distinguished from each other
experiment
deliberately imposes treatments on individuals to measure their response; show causation
completely randomized design
an experiment done without blocking
randomized block design
separate subjects into blocks and then randomly assign treatments within each blockb
block
group of experimental units that are similar in some way that could be a cause of confounding. used to remove the effects of confounding
matched pairs design
a common experimental design for comparing 2 treatments that uses blocks of size 2
scope of inference
random sample (inference about population) + random assignment (inference about cause/effect)
undercoverage
when some members of a population cannot or are less likely to be included in a sample
nonresponse
when an individual is part of a survey but chooses not to respond or cannot be reached. this is different than voluntary response where people put themselves in the sample
response bias
pattern of inaccurate results ( wording of a questions, interviewer, lying, etc.)
explanatory variable
helps explain/predict response (on the x-axis)
response variable
outcome being measured ( on the y-axis )