independent variable
- the cause
- experimenter has control over it
example: treatment for depression
IV = type of treatment
DV = level of depression
dependent variable
- the effect
- the thing that changes in response to the independent variable
example: treatment for depression
IV = type of treatment
DV = level of depression
manipulated vs. non-manipulated IV’s
manipulated = 3 different kinds of treatment for depression
non-manipulated = men vs. women (pre-existing, cannot be changed)
true experiment
(types of research)
- at least one independent variable is manipulated and subjects are randomly assigned
example: comparing two types of treatment for anxiety
quasi-experimental
(types of research)
- at least one independent variable is manipulated, but there is non-random assignment of subjects
example: comparing two types of treatment for anxiety, and the groups are already pre-existing (ward A in a hospital vs. ward B)
observational/passive/non-experimental
(types of research)
- no interventions or manipulation of variables
example: studying the extent of cigarette smoking in males and females
between-group designs
(group designs)
- compares two independent groups
within subjects designs
(group designs)
- the groups being compared are correlated or related
counterbalancing
- each group starts with a different treatment or tasks in order to avoid carryover effects
example: Latin square
AB design
(single subject designs)
- baseline condition followed by a treatment condition
associated problem: history
ABAB design
(single subject designs)
- baseline condition, followed by treatment condition, then baseline condition again, then treatment condition again
associated problem: failure of return to baseline, ethical concerns around removing treatment that is helpful
example of a multiple baseline design across subjects
(single subject designs)
looking at the effect of medication on hyperactivity for 3 different children
example of a multiple baseline design across situations
(single subject designs)
applying an intervention to one problem behaviour (example: biting) in three different settings (home, school, park)
example of a multiple baseline design across behaviours
(single subject designs)
applying one intervention for three different problems in one subject (biting, head-banging, and rocking)
simultaneous treatment design
(single subject designs)
- two or more interventions are implemented during the treatment phase in a balanced format (example: alternating during the day and balancing across days)
simple random sampling
- every member of the targeted population has an equal chance of being randomly selected
example: a study focused on registered voters in Alberta, subjects would be randomly selected from a pool of all registered voters
stratified random sampling
- first the targeted population is divided into strata (example: ethnicity, age groups, income levels)
- then a random sample of equal size from each stratum is selected
proportional sampling
- individuals are randomly selected in proportion to their representation in the general population
systematic sampling
- involves selecting every (n)th element
example: if 100 out of 1000 people are needed, every 10th person would be selected
cluster sampling
- identifying naturally occurring groups of subjects and then randomly selecting certain clusters
example: randomly selecting 10 schools in a school district and studying all of the 5th graders from those schools
threats to internal validity
- history (ex - trauma)
- maturation (ex - aging)
- testing / test practice (use the Solomon Four-group design)
- instrumentation (ex- equipment breakdown)
- statistical regression (ex - extreme scores naturally become less extreme)
- selection bias (ex - all volunteers)
- attrition (ex - drop out)
- diffusion
threats to construct validity
- attention and contact with clients (this alone can have an effect)
- experimenter cues/clues
- demand characteristics (ex - informing subjects about side effects)
- john henry effect (ex - competition between groups)
threats to external validity
- sample characteristics (differences between sample & population are too extreme)
- stimulus characteristics
- contextual characteristics (ex - people react in different ways when they know they are being observed)
threats to statistical conclusion validity
- low power (contributors are small sample size / inadequate interventions)
- unreliable measures
- variability in procedures
- subject heterogeneity
