quiz 2 Flashcards
Explain a randomized control trial
When you take a group of volunteers, and randomly assign them to two groups.
One groups gets an intervention and the other gets a non-intervention or placebo. These two groups are compared to see if there’s a difference between them overtime.
What is a confounding variable?
An alternative explanation for a relationship that has been observed
Why are RCTs good at dealing with confounding variables?
In an RCT, the confounding variable will equally affect both of the randomly split group in the RCT
What is a population?
persons, objects, or events that meet a specific set of criterias
What is a target population?
the larger population to which results of a study will be generalized to
What is an accessible population
the actual population of subjects available to be chosen for a study. (a subset of the target population)
What is a sample?
a subgroup of the accessible population which allows the results to be generalized to the population
What is a sampling bias?
the extent that sample SYSTEMATICALLY misrepresents populations. Can be conscious or unconscious
What is a sampling error?
the extent that a sample RANDOMLY misrepresents population
What are the types of sample techniques?
Probability samples and non-probability samples
What is a probability sample?
Random selection (or other known probability). Sample is considered, representative of population(although it may not be). Can estimate sampling error
What is a non-probability sample?
Non-random samples. Generalization is difficult.
Can’t estimate sampling error (should not use inferential statistics)
Types of probability sampling
Simple random sampling, systematic sampling, stratified random sampling, and cluster sampling
What is simple random sampling?
Random sampling. Everyone has an equal chance of getting in the study
What is systematic sampling?
systematically choosing samples. Ex: selecting every 10th person from a group. So you can estimate the probability of a person being in the sample
What is stratified random sampling?
specifying a number from each category in the sample.
Ex: 20 students from each grade level
What is a cluster sampling?
multi-layer/stage
Types of non-probability sampling
convenience sampling, quota sampling, purposive sampling
What is convenience sampling?
when subjects are chosen on basis of availability. This is the most common form
Types of convenience sampling
- Consecutive sampling: recruit everyone to meet inclusion/exclusion criteria
- Volunteers(flyers/ads): what attributes lead subjects to volunteer & potential bias of self-selection
What is quota sampling?
Like stratified, where you select a specific amount from each category, but this is not random
What is purposive sampling?
where subjects are hand-picked by specific criteria. Ex: case series
What is an extraneous variable?
any factor not related to the purpose of the study, which may affect dependent variables.
- can become confounding variable if uncontrolled
3 essential components of experimental research
- include a control or comparison group.
- independent variable manipulated by the experimenter
- subjects are randomly assigned to groups, as opposed to naturally occurring or provider choice
What is random assignment?
each subject has an equal chance of being assigned to any group.
- helps control for extraneous variables or prognostic indicators
Random sampling is used to ____.
Random assignment is used to _____.
Generalize to a population.
Equalize two groups (controls and experimental)
What is concealing allocation?
Separating people should do the “random assignment” & data collection.
Sealed envelopes and locked cabinets
What is blinding?
To control conscious & unconscious bias in anyone involved in a study.
What is a single blind?
A case where just the measurer is blinded
What is double blind?
A case where both the measurer and patient are blinded
What type of blinding is most imp in PT research?
Double blinding
At what percentage does missing data become concerning? Especially if related to a treatment
20%
What is a per-protocol analysis?
When analysis is only done on subjects who completed the study & were compliant with the methodology. Regardless of how many started
What is intention to treat?
When analysis is done in the group that subjects were assigned to, regardless of whether or not subjects finish or switch groups. Highly recommended
What are imputations?
Estimating missing data. making up data
What are design strategies for controlling inter-subject differences (confounders)
- Homogenous. Ex: only males between age 12-19
- Build confounder into the study as an IV. Analyze it as an attribute
- Matching: matching individuals from different groups, to avoid differences
- Using subjects as their own control. (within-subject design)
- Analysis of covariance (ANCOVA). Statistically adjust DV, based on differences in covariate
Types of experimental design validity
External validity, construct validity, internal validity, statistical conclusion validity
Which of the experimental design validities are the most imp?
External and internal validity
What is external validity?
Whether or not the results of the experiment can be generalized to other persons, settings, or times. randon sampling
What is construct validity?
To what theoretical constructs can results be generalized
What is internal validity?
is there evidence of causal relationship between IV & DV? random assignment
What is statistical conclusion validity?
is there a relationship between the IV & DV
What are the threats to external validity(generalizability)?
- Selection (sampling): sample does not represent target pop.
- Setting
- History: conclusions made 30 yrs ago might not be valid today
What is a consort statement?
A checklist of items to include when reporting a randomized trial
What is as treated analysis?
when patients data are examined in groups that subjects actually conform to, regardless of groups that they were randomized into
What are the types of experimental design?
True experimental design & Quasi-experimental design
Characteristic of true experimental design
IV is manipulated
RCT= gold standard
At least 2 comparison groups
Subjects are randomly assigned
Characteristic of a quasi-experimental design
May lack randomization & comparison groups
What is a single factor design?
One that has only 1 IV
Types of single factor design
Pretest-posttest control groups & posttest only control groups
Characteristics of a pretest-posttest control group design
- Traditional structure of the RCT
- Used to compare 2 or more groups formed by random assignment
- One group gets an intervention and the other is the control
- Both are measured before and after treatment
- Differences between the groups can be attributed to the treatment
- Cause and effect
- only single factor if time is not a 2nd IV*
Characteristics of posttest only control group design
identical to the pretest-posttest, except that pretest is not administered to any group
- Advantages: less laborious & less costly to perform, may be most appropriate when a pretest is impractical or has the potential to skew the study results
- Disadvantages: you don’t know if the groups were equivalent @ baseline
What is a multi-factor design?
two or more IVs.
- Interaction effects: do scores across levels of factors A depend on the levels of factor B
What is a randomized block design?
Same as multi-factorial except 1 IV is not randomized.
Blocked ensures that the same number of subjects in each level
Experimental designs for independent groups (between subjects) are
Single factor designs & multi-factor designs
Experimental designs for repeated measures (within-subjects) are
Single factor (one-way), crossover design, multi-factor repeated measure design
What is a single factor repeated measure design?
Subjects are exposed to all levels of the IV
- There is no control subjects act as their own control
What is an order effects?
The outcome may be dependent on the order of measurement. downside to repeated factors
Solution: randomize order, counterbalance(alternate for two levels), latin square(alternate for 3 or more level)
What is a cross-over design?
Similar to the single factor pretest-posttest design. Only works under stable conditions
_- Used when there are 2 levels of the IV & the patient’s conditions is stable over time
- Subjects are randomized to 2 groups
- 1st group gets an intervention, the other a placebo, then they switch.
Problem with this is that the 1st group might still be exerting effects of the intervention by the time the switch happens. combated by the washout period
What is the washout period?
A sufficient period of time given to eliminate the prolonged effects of the treatment
What is a multi-factor repeated measure design?
All IVs are repeated or within subject.
Mixed design: at least 1 IV is between subject & the other is within subject
The effects of treatment can be referred to as treatment ____ or ____
Effectiveness or efficacy
What is efficacy?
The benefit of a treatment delivered in an under highly controlled & ideal environment
What is effectiveness?
the benefit of a treatment delivered in a pragmatic manner under real world conditions
What are the 3 main steps of appraising therapy studies?
Are the results valid?
What are the results?
Will the results help care for my patient?