1/37
Explanatory Variables(Factor)
Predicts, explains or influences the response variables(the outcome)
Response Variable
The measured outcome
Confounding Variable
A variable that is associated with both the explanatory and response variables. It creates doubt about a true causal connection between the explanatory and response variables.
Observational study
A study in which variables are observed without imposing an conditions or treatments
Observational studies
Because conditions are not imposed, they cannot eliminate confounding and causal conclusions are not possible
Prospective
Analyzing data that are gathered now and into the future
example: tracking covid 19 mortality for the next year
Retrospective
Analyzing data that were already collected in the past
example: analyzing covid 19 mortality from the last two years
Direct Observation
Analyzing data measured through observation or records of observations
example: collecting data on vaccinations from hospital records
Survey
Analyzing data gathered from responses to questions
example: asking people if they’ve been vaccinated
Experiment
A study in which conditions, or treatments, are imposed on experiemental units
Experimental units
Subjects, participants or units to which treatments are assigned
Treatments
Different categories or levels of an explanatory variables that are imposed on experimental units
Census
Measuring all items or individuals in a population, problem: infeasible
Sample
Measuring a subset of the population, problem: inaccurate
Simple Random Sample(SRS)
A sampling method in which every possible group of n individuals in the population has an equal chance of selection.
Generalizability
When individuals are randomly selected from target population, results can be _______ to the whole population
Well-designed experiment
Random assignment, controlling for confounding variables and replication
Why do well-designed experiments with random assignment to treatments allow for causal conclusions?
Potential confounding variables are balanced between groups and causal conclusiosn are possible.
Placebo
An inactive treatment. It provides comparison of treatment to placebo effect
Placebo effect
When subjects’ belief of receiving an active treatment leads to a measured response, even though the treatment is actually inactive.
Single-blind experiment
Either subjects or the researchers are unaware of who receives active treatment or placebo. This reduces bias
Double-blind experiment
Analysis people don’t know which group is which, participants don’t know if they are getting pacebo and anyone who interacts with them cannot know. This reduces bias
Population
The entire group that gets generalized
Cluster sample
Divide the population into groups and randomly select clusters. Then collect data from everyone in those clusters.
Stratified Sample
Divide the population into non-overlapping homogeneous groups called strata then do a simple random sample within each strata
Bias
Having an opinion that can alter data
Voluntary response bias
When a sample is comprised entirely of volunteers or people who choose to participate, the sample will typically not be representative of the population
Nonresponse bias
Individuals chosen for the sample for whom data cannot be obtained (or who refuse to respond) may differ from those for whom data can be obtained.
Undercoverage bias
When part of the population has a reduced chance of being included in the sample, the sample will typically not be representative of the population.
Response bias
Problems in the data gathering process, such as questions that are confusing or leading, or when people lie because they don’t want to tell the pollster the truth.
Convenience sample
Choosing people for data based on convinience
example: choosing the first ten people to come to class for the sample
Random Assignment
Each participant is randomly assigned. This Reduces effect of confounding variables
Control group
Provides baseline for comparison
Completely randomized design
An experiemental design in which all experiemental units are allocated to treatments completely at random, with no prior grouping, blocking or pairing
Matched pairs
Experimental units are grouped into pairs of size two based on similar characteristics or where a single subject receives both treatments. Reduces variability created by differences in the subjects
Blocked design
Separating on something inherent in subjects first then running the experiment. Reduces variability created by differences in the subjects
Why blind researchers
Prevents confounding
Why blind subjects?
Prevents placebo effect