Design strategies Flashcards
Group designs
Several subjs studied
Groups formed by investigator
Between group designs
Separate groups of subjs formed and ea group receives only one condition
Single-case experiments
Characterized by an investigation of a given individual, few individuals, or one group over time
One or a few subjs studied
Dependent measures administered over time
Manner in which the IV is implemented is examined in relation to the data pattern for the subj(s) over time
Conditions of experimentation
Laboratory versus applied
Analogue versus clinical
Efficacious versus effectiveness
Laboratory versus applied
Highly controlled settings that depart from conditions of everyday life
Versus
Real-world settings - what can be accomplished in everyday studies
Analogue versus clinical
The extent to which research studies a phenomenon that is intended to resemble something that occurs in everyday life
Efficacy versus Effectiveness
Efficacy = research that is directed to more controlled conditions of a lab; evaluates tx in controlled conditions. Can rule out threats to validity Effectiveness = intervention research in applied settings under conditions in which tx is usually administered. Greater vulnerability to threats to all types of validity but external
Random selection versus random assignment
Selection occurs before assignment
Selection = equal probability that subjs w/in a population can be selected
Assignment = allocating chosen subjs to groups randomly - ensures grp equivalence
Nuisance variables
those characteristics in which one is not interested but that, in principle, could infl the results
Random assignment ensures these variables will be distributed unsystematically
Matching
Refers to grouping subjects together on the basis of their similarity on a particular characteristic or set of characteristics
Groups will not differ on that characteristic prior to tx
Ways to accomplish matching
- Identical pretreatment scores
- Rank all of the subjs
- Categorical variable such as age, gender
Mismatching
Subjs matched first then randomly assigned to groups
Used in an attempt to equalize groups when random assignment is not possible
Regression to the mean is a concern when subjs selected bc of extreme scores
Pretest-posttest control group designs
Essential feature = subjs tested before and after
Adv: controls for threats to validity
Disadv: pretest sensitization
R O1 X O2
R O3 X O4
Posttest only
consists of a minimum of two groups and essentially is the same as the previous design except no pretest
No effect of pretest
R X O1
R O2
Less popular bc no pretest (need to know baseline functioning usually)
Solomon four-group design
Purpose = evaluate effect of pretest Evaluate w two-way ANOVA Group 1 = Pretest, intervention, post Group 2 = Pretest, NO intervention, post Group 3 = No pre, intervention, post Group 4 = No pre, no inter, post
R O1 X O2
R O3 O4
R X O5
R O6
Factorial designs
Allows for more than one IV to be examined
Need larger sample size
2x2 design
Within each variable, 2 or more conditions administered
Eg two variables (tp experience and type of tx) would consist of two conditions (experience versus lack; tx A versus tx B)
Reason = combined effects of two or more variables may be of interest
Quasi experimental designs
Same as experimental designs but NOT randomly assigned
eg
nonR X O1
nonR X O2
Multiple tx designs
Ea participant gets more than one tx
Different tx presented to ea subj – WITHIN subjects
Tx presented in different orders - to COUNTERBALANCE effects
Cross over designs
Partway thru the experiment, subjs crossover to another experimental condition
Design used w 2 different tis
R O1 XA O2 XB O3
R O4 XB O5 XA O6
Multiple tx counterbalanced design
Select a set of sequences in advance to and to assign subjs randomly to these sequences as they arrive to the experiment
Latin Square design
Each group has a different sequence that includes each of the 4 txs
Considerations when using the designs
Order and sequence effects
Restrictions w various independent and dependent measures
Ceiling and floor effects
Tx strategies
Tx package Dismantling Constructive Parametric Comparative-tx Tx-moderator Tx-mediator
Tx package
Does tx produce therapeutic change
Requires:
Tx versus no-tx or waiting list control
Dismantling
What components are necessary, sufficient, and facilitative of therapeutic change?
Requires:
2 or more tx grps that vary in the components of tx provided
Constructive
What components or other txs can be added to enhance therapeutic change?
Requires:
2 or more tx groups that vary in components
Parametric
What changes can be made in the specific tx to increase its effectiveness?
Requires:
2 or more tx groups that differ in one or more facets of the tx
Comparative tx
Which tx is more or most effective for a particular problem?
Requires:
2 or more different txs for a given clinical problem
Tx-moderator
What patient, family, or tp characteristics does tx depend on to be effective?
Requires:
Tx as applied separately to different types of cases, tps, etc.
Tx-mediator
What processes occur in tx that affect w/in-session performance and that may contribute to tx outcome?
Why use control groups?
Controls for threats to testing, hx, maturation, selection, etc.
No tx versus wait-list control - to show it wasn’t an effect of time; Ethics, recruitment, attrition
Nonspecific tx control grp - attn placebo
Instrument develop and assessment - construct validity
