MODULE 2 - STUDY DESIGNS AND SAMPLING Flashcards
4 goals that all sampling designs must achieve
- all sampling units are selectable
- selection is unbiased
- selection is independent
- all samples are possible
all sampling units are selectable explanation
This goal is about the sampling unit. Every sampling unit in the statistical population must have some non-zero probability of being included in your sample.
selection is unbiased explanation
The probability of selecting a particular sampling unit cannot depend on any attribute of that sampling unit and—on average—the sampling units must have the same attributes as the statistical population.
selection is independent explanation
Selection of a particular sampling unit must not increase or decrease the probability that any other sampling units is selected.
all samples are possible explanation
This goal is about the sample composition. All samples that could be created from the statistical population are possible.
With the aim of increasing the number of responses, a phone pollster decided to ask 2 people in each randomly selected house to answer a questionnaire. What type of error will this cause?
non-independence
What goal does the following statement refer to? “every combination of units must be possible in your sample”
equal chances for all samples
what is an observational study
a study using observations from a statistical population where the investigator has no control over the explanatory variables
primary goal of an observational study
to characterize something about an existing statistical population
problem with observational studies
provide a tool for discovering associations, but they cannot make statements about whether a factor causes the response you are interested in
what is an explanatory variable?
a variable that an investigator believes may explain the response variable
what is a confounding variable?
unobserved variables that affect a response variable
5 designs for observation studies
- simple random survey
- stratified survey
- cluster survey
- case-control survey
- cohort survey
what is a simple random survey?
sampling units are selected at random from the statistical population
what is a stratified survey?
sampling units are selected from within predefined groups
what is a cluster survey?
groups of observation units are selected at random
what is a case-control survey?
sampling units are selected based on the response variable
what is a cohort survey?
sampling units are selected and then followed through time
what is a retrospective study?
where the outcome is already known, which comes with an increased risk of spurious relationships if you are selecting groups based on the outcome
case-control studies are a good example of a retrospective study
what is a prospective study?
where the outcome is not yet known
cohort studies are a good example of a prospective study
what is a cross-sectional study?
that study a response variable at only a single snapshot in time
what is a longitudinal study?
that study a response variable at multiple points in time
what is an experimental study?
a study using involving a statistical population where the investigator can control the explanatory variables
primary goal of an experimental study
to study the effect of one (or more) manipulated variables on one (or more) response variables
what is a factor?
any manipulated variable in an experimental study
what is a level?
the manipulated value within each factor
2 key things that distinguish experimental studies from observational studies
- the explanatory variable is manipulated by the researcher
- sampling units are randomly assigned to each level in a factor
what is replication?
the idea that a treatment will be repeated a number of times to see how repeatable a measured outcome is
the number of times a treatment is repeated on independent, representative and randomly selected units
what is pseudoreplication?
an error in the design of an experimental study where the observation units are analyzed rather than the sampling units
5 design elements of experimental studies
- control treatment
- blocking
- blinded
- placebo
- sham treatment
what is a control treatment?
is common in experimental studies and is intended as a reference treatment to compare against the treatment levels that alter the explanatory variable
It contains everything that the treatment levels do, except the treatment itself
what is blocking?
is analogous to stratified sampling, but for experimental studies
It is used to control for variation among the sampling units that is not of interest to the researcher
what is blinded?
commonly used in experimental studies involving people and refers to a design where the sampling unit (usually a person, but could be a group of people) does not know what treatment they are being exposed to
what is a placebo?
a method often used in medical trials for the control treatment that helps accomplish a blinded design
It is substance, or treatment, that has no effect on the response variable
what is a sham treatment?
similar to placebo in that it is a method used in control treatments
the purpose of a sham treatment
is slightly different in that it aims to account for the effect of delivery of a treatment that is not of interest to the researcher
what is a sham treatment?
similar to placebo in that it is a method used in control treatments
the purpose of a sham treatment
is slightly different in that it aims to account for the effect of delivery of a treatment that is not of interest to the researcher
Imagine a study that evaluates the effectiveness of different over-the-counter pain relievers in alleviating the symptoms of arthritis: acetaminophen, ibuprofen and acetylsalicylic acid. Two hundred patients are randomly assigned to receive one of these three pain relievers, or to receive a placebo (control). How many factors and levels are evident in this study?
1 factor with 4 levels
Patients who are blinded to the experimental treatment is a crucial part of a randomized clinical trial. Why?
Reduces the possibility of placebo effects
Reduces biases in measurements stemming from the anticipation of a treatment effect
A researcher studied the effect of the prescription drug raloxifene on fracture risk in postmenopausal women. They found that women who took raloxifene over a five year period reduced their risk of clinical vertebrate fracture compared to women who did not take the drug. What are the factors and levels in this experiment?
1 factor (drug) and 2 levels (raloxifene, no raloxifene)