quiz 3 Flashcards
What is the difference between a quasi-experimental design and an experimental design?
They are the same, but a quasi experimental design lacks either random assignment, a comparison or both
What are the types of a quasi-experimental design?
One group design & Multi-group without randomization
What are the types of a one group design?
- One group pretest-posttest & one way repeated measure design
- Time series design
What is the difference between a one group pretest-posttest & one way repeated measure design?
The only difference is there are a couple of follow ups in the one way repeated measures
What is the One group pretest-posttest & one way repeated measure design vulnerable to?
Internal validity
When is the one group pretest-posttest & one way repeated measure design defendable?
Defendable when previous research has documented behavior of control group (known natural history)
Defendable when follow up timeframe is short
What is a time series design
It has 1 intervention, a pretest-posttest, time is the IV, and you take a series of pretest-posttest which acts as a pseudocontrol condition.
Why does a time series design have better internal validity?
Better internal validity because series of test can be compared with each other.
Types of a multiple group without randomization design?
- Nonequivalent pretest-posttest control group & historical controls
- Nonequivalent posttest only design
Characteristics of nonequivalent pretest-posttest control group & historical controls
- Includes a control and an intervention group.
- Includes a pretest, so there is some control over internal validity
- Test equivalency of groups is based on the prescores
- May be warranted ethical concerns preclude a randomized control group
Characteristics of nonequivalent posttest only design
Has a control and intervention group
No pretest, so no knowledge of where subjects started
NO way to test equivalency of groups based on pre-scores
Not really used
Single subject design (N=1 designs) are similar to
Time series design
Single subject design allows for “cause & effect” inferences due to
Rigorous planning, including reliable & valid outcome measures.
Consistent with EBP focus on individual patients
Has a control phase.
What is a control phase?
When you can see what happens without intervention.
In a single subject design, independent variables are called _____ and dependent variables are called _____
Treatment(levels=phases)
Patient response/target behavior
How does n=1 differ from case studies/series?
It differs due to repeat measurements & design phases
What are the design phases in an n=1?
Baseline & intervention
What does the baseline phase look for?
Stability and trend
What does stability do?
ideally want to observe little variability/change. If baseline variability is large, may need to obtain more measurements
What does trend indicate
indicates a change is occurring without intervention. Evaluate whether rate of change is stable or unstable.
Unlike group experimental research, n=1 timeline studies can be
flexible
Characteristics of length phase?
- Important to extend phase until stability is reached
- Minimum 3-4 data points for each phase
- best to have relatively equal phase and interval lenghts
- can use summary assessments
Characteristics of a target behavior (DV)
- Patient- focused & even subjective measures
- Needs to be quantitative & should provide proof of reliability
- Can be “time to something” “frequency of something” “standardized outcome measure”
What is single subject A B design?
Baseline phase (A) Treatment/intervention phase (B)
Advantages and disadvantages of single subject A B design
Advantages: can observe changes between A & B
Disadvantages: Unsure if B was due to some confounding or maturation effect, learned effect
What is single subject A B A design?
When treatment is applied, then removed.
Baseline (A)
Treatment/intervention: (B)
Advantages and disadvantages of single subject A B A design
Advantages: can determine if A truly caused B to occur (expose any probable confounding variable). Since target behavior should return to baseline
Disadvantages: behavior must be reversible. There are ethical issues with withdrawal treatment
Data analysis is generally based on either ____ or _____
Visual trends or variability assessment
How are n=1 studies done?
- Blind patients to avoid bias
- Randomize periods of time that patient receives intervention/control
- Measure outcome of variables. Measurer should be blinded to which option they are measuring/patient is receiving
- Determine where patient showed the best result & you’ll know which was best for that specific patient
What are the disadvantages of n=1 studies?
- Not all therapeutic options lend themselves well to randomization for all individual patients.
- Ensuring that there’s a clinically relevant outcome measure to look at.
- Can take a long time to complete
- Not really publishable & the types of people involved: patient compliance & patience
What are advantages of n=1 studies?
Allows us to be definite in which treatment/intervention works for the individual patient. the most patient centered
What is an observational studies?
Collecting data as they naturally exist. No manipulation of variables
What is an exploratory research?
Systematic investigation of relationships among 2 or more variables
What is a longitudinal research?
anytime there’s a time component to a study. Usually weeks or months
What is a cross-sectional research?
One time point
What is prospective research? in regards to longitudinal
Start a study & it goes forward in time. More control, more valid Harder
What is retrospective research? in regards to longitudinal
Start a study and look back in time. * less control, less valid* easier
What is correlation & predictive studies?
Based on “covariation” in data, meaning if two variable. Vary together.
Does not imply causation.
- Prediction think regression (association between variables given a time component)
What is a case-control & cohort studies?
Intended to study risk factors.
Association (cause) between disease & exposure.
What is a case-control study?
Subjects are selected based on whether or not they have disorder.
Controls should be analogous group of people without the condition but is in every other = to the cases.
Examine if exposure is different between cases & control.
works especially well for rare conditions
Retrospective
Cheaper and shorter
What is a cohort study?
Subjects are selected based on exposure or not.
Find analogous group that is equal.
Usually prospective, but can be either prospective/retrospective.
Examine if different incidence of disease.
Doesn’t really work well for rare conditions
More expensive and longer
Types of cohort studies?
Prospective cohort study & retrospective cohort study
What is prospective cohort studies?
Gather cohort/ identify, determine exposure status, follow-up to determine outcome.
What is retrospective cohort study?
Determine outcomes, determine exposure status, identify cohort
5 criterias to provide evidence of cause & effect
- Time sequence. (if cause happens before effect)
- Strength of association (strong association between something)
- Biologic credibility (does it make sense biologically)
- Consistency with other studies
- Dose response relationship (more exposure, more likely to have disease)
What is secondary analysis study?
Uses existing database to examine relationship between variable.
Secondary analysis of projects with other main focus can combine comparable data sets
Characteristics of a secondary analysis study
- less control than primary analyses/studies
- Can be pre-planned
Types of descriptive research
Natural history, normative studies, qualitative research, surveys, & case studies
What is natural history?
Things that just happen without explanation
What are normative studies?
Document things that are normal, in order to decide things that are not normal
What is qualitative research?
goal is to describe & create theories
Case studies are usually ____
retrospective, less standardized & controlled( no repeat baseline measurements & assessments, time frames are naturally occurring)
Does not meet the IRB definition of research, less internal validity, & often less external validity
What is the IRB definition of research?
a systematic investigation designed to develop or contribute generalizable knowledge
Purpose of a case study
- Understanding unusual patient condition
- Examples of innovative or creative therapies
- Generating & testing theory(how theories or systems can be applied)
- Providing future research directives
What does qualitative research do?
Answers questions (who, what, where, why, when, and how). Tries to understand the nature and qualities of a phenomenom
What does quantitative research do?
Answers the question of how much.
Considers the idea of magnitude
How much of an occurrence and how much of a relationship. Association between a risk factor and an outcome
In qualitative research, you may do
A simple observation
In depth interviews
Focus groups
Studies
In quantitative research, you may do
Non-interventional or interventional quantitative reseach
Non interventional quantitative research is
Observational, watching to see what happens and counting things up.
Types of non-interventional quantitative research
Case-control study, Cohort study, Ecological study, & survey
Interventional quantitative research study is
adding an intervention to see the outcome
Types of interventional quantitative research study
RCTs, & double blind RCTs
Descriptive studies ____ compare relationships
cannot
Types of studies that can be used to give info about prognosis in order of preference
Cohort study, RCT, & case control study
____ is the best for answering prognosis questions
Cohort study
What is a prognostic cohort study?
One that follows patients on different stages of a disease/condition
What is the problem with prognostic RCTs?
All prognostic factors are equalized by randomization (except for the intervention)
What are the 3 steps in critically appraising a prognostic study?
- Are the results valid?
- What are the results?
- Will they help me care for my patients?
4 explanations of study findings
Truth, Chance, Bias, & confounding
4 questions to ask when reading a prognostic study
- Was there a representative & well defined sample of patients assembled at a common point in the course of disease?
- Is follow up sufficiently long and complete?
- Were objective outcome criteria applied in a blind fashion?
- Was there an adjustment for important prognostic factors?
What is a referral bias?
where patients are assembled in a study from an academic testing medical canter, where people refer very unusual severe cases.