True or False: Although most confounds occur at the stage of operationalising constructs, there are some confounds that can still occur in studies designed well, and are not problems of operationalisation.
True - Person variable confounds as a consequence of differential attrition and confounds due to floor/ceiling effects
What can we do if we detect a confound in our own study?
Conduct a new study to test the alternative explanation, or a statistical correlation
Define sequencing confounds
Sequencing confounds are confounds introduced by the specific IV level sequence. Previous blocked conditions or single trials can facilitate performance on current test, hinder performance on current test, or promote people continuing in a behaviour
What are 3 possible remedies for sequencing confounds
Counterbalancing the order of conditions, using a fixed sequence of conditions that works against the predictions, and/or using a between-participants design with random assignment or matched pairs
What is counterbalancing the order of conditions?
Systematically varying the order of conditions across participants or groups
What is a Latin square, and what makes a good square?
A solution to counterbalancing order of conditions. A good Latin square has (1) each participant needs to undergo all conditions, (2) each condition should be represented in each position, and (3) pairs are counter-balanced
How are counterbalanced designs analysed?
You can analyse the effects of order by entering ORDER as a between-subjects variable in ANOVA or regression. If the results show no interaction of Iv with order then we can be relatively sure that effect is not due to the order in which conditions were presented. If the results show an interaction with order we need to analyse the effects of order seperately.
When are fixed sequences used?
Used when carryover effects cannot be balanced out or equated.
Can a fixed sequence be a counter-confound?
Yes
What are the problems with fixed sequences?
Fixed sequence can have opposite effects. These effects can override intended effects. They can be a counter-confound when the results are opposite to predictions, or a confound when researchers overlooked other effects of the sequence
When are matched-pairs used?
When it is not possible to counterbalance conditions, it is often best to switch to a between-subjects design, where the conditions contain different groups of participants
What are testing confounds?
Confounds introduced by the test or DV measurements that influence subsequent tests
What is the difference between testing and sequencing confounds?
Testing confounds are caused by DV measurement, whereas sequencing confounds are caused by IV level sequence
What is an observer criterion shift?
It is a shift in measurement over time that results in apparent change in the DV, if the change is in line with predictions. It is caused by human observers (learning leads to tighter or more lenient criterion for assessing behaviour) or non-human measures (equipment failure)
How can criterion shifts be prevented?
By using video recordings of the behaviour instead of on-line behaviour observations. If not possible, use well-trained, practiced observers operating with a clear cut criteria
What is a procedural confound?
An extraneous variable that is introduced in the procedures used by the researcher to test a research question. They arise from decisions about the procedures used in the study - the step by step actions taken to conduct the study
