SWK 420 Chapter Notes - Chapter 5: Crystal Computing, Time Series, Random Assignment
9 views9 pages
For unlimited access to Textbook Notes, a Class+ subscription is required.
SWK 420 Chapter 5
• Good studies have:
o Random assignments
o Control groups
• Threats to Internal Validity
o History: Natural disasters (hurricanes)
o Maturation: Developmental maturity.
o Statistical regression to the mean: Giving the anxiety questionnaire right before
o Instrumentation or Testing: Questionnaire too long or tested too many times.
o Mortality: People missing the intervention
• Quasi-Experimental Group Design
• Random group is missing.
Choosing the Right Design
• Design issues related to finances include:
o The use of staff time for collecting data
o Postage, telephone, or travel expenses
o Purchase of copyrighted instruments or scales
o Computer processing of the data
o The use of consultants
o Release time to write the report
• In terms of evidence-based practice, group research designs with randomized
assignment to the treatment condition (experiments) or with time-series designs with
a good comparison provide the most credible evidence.
• Time-series designs without a comparison group provide a middle range of
• Proactive research: Being about to plan months in advance and can build a process for
future recruitment of eligible clients
• Retroactive research: Current clients may be compared to clients who previously
completed a program or who have already dropped out in the past.
• The classic experimental design is the idea, the gold standard or model to which all other
research designs are compared.
SWK 420 Chapter 5
• What is an experiment? –It is a controlled study where clients or subjects are randomly
assigned to a group (sometimes called a condition) where they will receive a new (or
different) intervention from those designated to be in the control group.
• The notion of random assignment is crucial—it prevents a number of extraneous
variables or possibly cofounding variables and biases from interfering with the
researcher’s ability to make a conclusion about the strength or weakness of the
The Classic Experimental Research Design
• These designs have two main features:
o Clients (or subjects) who participate in the experiment are randomly assigned.
o Besides the experimental group, some subjects must be assigned to a control (or
comparison) group that does not get the new intervention.
• Shorthand notation for the classic pretest-posttest control group design is written:
R O1 X O2
R O1 O2
• The R in this notation scheme stands for the random assignment of clients to either the
experimental or the control group.
• The X represents the intervention or stimulus.
• The first observation (O1) is called the pretest, and the second observation (O2) is called
• This design provides information not only about changes in the group that receives the
intervention, but also comparable information from the group that does not get the
• Random assignment must occur prior to the intervention.
• In double-blind procedure, neither the researcher nor the research subject knows
whether participants are in the control or the experimental group.
The Posttest-Only Control Group Design
• This design is handy for those situations where a pretest might affect the posttest results
or when it is not possible to conduct a pretest.
• This design is also useful in situations where anonymity is paramount—so it would not
be possible to compare an individual’s pretest and posttest scores.
• Shorthand notation for the Posttest-Only Control Group Design:
R X O1
SWK 420 Chapter 5
The Solomon Four-Group Design
• Shorthand notation for the Solomon Four-Group Design
R O1 X O2
R O1 O2
R X O2
• This design provides two opportunities for the treatment effect to be demonstrated.
• The design’s strong point is that the investigator can maximally control for alternative
explanations and thus increase the confidence that can be placed in the findings.
Internal and External Validity
• Internal Validity: whether the intervention was truly responsible for the observed
differences in the dependent variable
• The internal validity of a study can be threatened by extraneous variable.
• Extraneous variables: those not considered or purposely incorporated into the experiment.
• External validity is concerned with whether a study can generalize well to other
populations, locations, and client groupings.
• Measurement validity is concerned with how well an assessment instrument measures the
concept it was designed to capture.
• Conversational uses of “validity” or “valid” when applied to a research project generally
mean that a study was real, authentic, meaningful, relevant, or simply just a “good”
Major Threats to Internal Validity
• Maturation: The aging of participants or different rates of physical growth or
development within comparison groups are examples of influences that should recognize
and attempt to control as much as possible.
• History: This refers to specific events that occur between the pretest and posttest that
were not part of the researcher’s design and that could influence the results. (Personal
events in a client’s life are not considered to be threats to the internal validity of a study.)
• Testing: Repeated testing provides practice and in itself can improve test scores. The
threat of testing is sometimes known as the practice effect.
• Instrumentation: This term refers to changes in:
o The use of the measuring instrument
o The way the instrument is scored
o Procedures used during the study
o The way the dependent carriable is counted or measured.