Quiz 4 (Ch. 5) Flashcards
What is an odds ratio?
Examining a correlation b/t 2 variables by comparing how often the bx of interest occurs in the presence and absence of a particular variable
Odds ratio b/t smoking and lung cancer = 25
People who smoke are 25x more likely to get lung cancer
Odds ratio b/t texting on a cell phone and crashing = 6
A person is 6x as likely to have a crash while texting than when they are not
Selection bias
Bias introduced by the selection of individuals, groups, or data for analysis. Proper randomization is not achieved, thus the sample is not representative of the population. Correlations b/t a variable you’re studying and a more powerful variable can lead to selection bias
5 advantages of replication based designs
- It’s not necessary to randomly assign participants to a tx and a control group
- Doesn’t require a large number of participants
- Doesn’t need an untreated control group
- Allows the researcher to determine whether the tx worked for each individual
- Allows the researcher to control sources of variability and make changes to the tx if it doesn’t work as expected
3 advantages of AB design
- Timeline during baseline can document that the bx is remaining stable and is not improving
- The tx can be associated w/a level change, transition phase, or both
- The effect of the tx is monitored over time on a day-to-day or week-to-week basis
3 weaknesses of AB design
- Possible that the change, if gradual, may have occurred as a result of other variable that were there all along
- Some confounded variable that was correlated w/the intro of the tx was responsible for the changes
- Regression toward the mean could be a factor
2 reasons why AB design is important
- Design is a component of all BA research designs and each requires making a discrimination about the differences b/t baseline and tx conditions
- This design is used by BA practitioners to guide their work on a QD basis
5 other factors to consider when evaluating an AB design
- How likely is it that the bx would have change that much on its own?
- Can any other variables be ID that could have been responsible for the change?
- The longer the baseline, the greater the confidence. Addresses regression toward the mean
- If data are collected on a large group of individuals over a large geographic area and remains stable until the tx is introduced, it adds support to the results
- A large # of independent AB designs provide overwhelming evidence that a confounding variable correlated w/the intro of the tx did not produce the effect
2 disadvantages of the ABA design
- Tx effects may persist even if they produced the change
- You leave the person in the baseline rather than the tx condition
Advantage of BAB over ABA
Person ends the study in the tx condition
What is the ABAB/reversal design? Do you always need to start w/baseline?
Alternate b/t 2 conditions, typically baseline and tx
Bx permitted to stabilize before each condition change
You don’t have to start w/a baseline. Can start w/tx or w/both tx
Looking at 2 or more conditions
What would you recommend to someone who thinks the treatment will lead to irreversibility who is choosing a reversal design? In other words how can they increase the chance of the design working?
Intro the tx for a brief period and then return to baseline. The procedure minimizes the chance that new contingencies develop that can trap or maintain the bx
What are three options if you have conducted ABAB phases and there appears to be an effect but there is a lot of overlap between the data?
- Track down and control the sources of variability
- Introduce additional replications
- Intensify the tx
Options 1 and 3 typically work best
What does the C stand for in an ABCB design? Give an example?
Control condition. Could be non-contingent reinforcement or a reversal of the reinforcement contingency
What is the most likely cause of partial reversibility?
A new reinforcer has taken effect, or you may not be able to easily remove all aspects of the tx b/c some elements may be under the clients control
