research design Flashcards
what is the inductive method
experiment first and explain later
what is the deductive method
explain first and verify later (hypothesis testing)
what is a null hypothesis
two variables are not related
what is predictive or criterion validity
how the test predicts future performance on a related task
what is concurrent validity
how the test correlates with established tests of known validity
what is construct validity
measures are consistent with theoretical constructs (expectation that as children grow older, their language skills improve
what is content validity
determines if the test samples he full range of skills that it purports to measure
what is a correlational coefficient
number that indicates the relationship between 2 or more independent measures
- expressed through Pearson Product Moment r
- r value of 0.0 means there s no relationship
- highest possible is 1.0 and lowest is -1.0
- the closer r is to 1.0, the greater the reliability
what is test-retest reliability
consistency of measures when the same test is administered to the same people twice
what is alternate-form reliability or parallel form
consistency of measures when 2 parallel forms of a test are administered
what is split half reliability
measure of internal consistency of a test. responses to items on the first half are correlated with those on second
what is a single subject design
- used in treatment evaluations
- usually involve 6 to 10 participants
- establish cause-effect based on individual performance as opposed to group average
- help establish cause based on differential performance under different conditions
- there are no pre and post tests. SSD measures are measured continuously and repeatedly to establish reliability
- do not always have just a single participant but the data are analyzed separately
- 3 designs include ABA, ABAB and multiple-baseline design
what is an independent variable
produces changes in a dependent variable
all treatments are independent variables
what is a dependent variable
affected by manipulation of the independent variable
all disorders are dependent variables
what are confounding variables
variables that should be ruled out in experiments
what are nonexperimental groups
consist of only one group
case study
what is matching
- alternative to random selection
- researcher identifies 2 participants who are similar on relevant variables and assigns one to the experimental group and the other to the control group
what are pretest-posttest control group designs
- experiemental/control group
- each participant undergoes a pretest and a posttest
- pretests are existing behaviors before treatment
- posttests are measures after treatment
- helps demonstrate the effects of treatment while ruling out the influence of confounding variables by showing that only the experimental group changed and the control group remained close to pretest measures
what is a multigroup pretest-posttest design
the groups are compared to each other and a control group is optional but the experimenter will only know whih treatment is better without the control and will not know if the treatment is better than no treatment
what is a limitation of group experimental designs
results may be extended to groups but not individual clients. Single subject is better if a practitioner wants apply treatment results to individuals
what is the ABA design
- basic experimental SSD design
- during the first A condition, the skills to be treatedare baserated
- during the second B condition, treatment (independent variable) is applied
- final A, the treatment is withdrawn and skills are measured
- shows that base ratedskills were stable, increased with treatment, and decreased when treatment was withdrawn
- the problem with this design is that it ends with no benefit to the participant
what is the ABAB design
- first A, skills are baserated
- B, new treatments are offered
- A: conditions are removed
- B: treatment reinstated. Final phase lasts until behavior is established
what are multiple-baseline designs
- a SSD that demonstrates the effects of the treatment by showing that untreated skills did not change and only treated skills did
- 3 variations: multiple baseline across subjects, multiple baseine across settings, and multiple baseline across behaviors
what is sensitivity vs specificity
sensitivity is the ability of a test to correctly identify those with a disease (true positive)
specificity is the ability of a test to correctly identify those without the disease (true negative)
what is the multiple baseline across subjects design
- shows that only treated participants changed
- researcher selects a target behavior to be taught to 3 or more
- the behavior is baserated
- one participant is treated while the others get a repeated baserate
- second is treated with others baserated etc.
- alternated baserated and treated until all participants have been treated
what is the multiple baseline across settings design
- involves a single behavior sequentially taught in different settings to show that behavior changed only in treated setting
- behavior baserated in 3 or more settings
- behavior taught in one setting
- repeats base rates in untreated setting
- teaches the behavior in another setting
- alternates until all settings have been used
what is a multiple baseline across behaviors design
- several behaviors are sequentially taught to show that only treated behaviors change
- three or more targets are selected and baserated
- first behavior taught to a criterion (80% accuracy)
- other behaviors baserated
- alternate until all behaviors are taught
what is the difference between experimental and descriptive research
experimental research explains the effects by findng the cause
descriptive research describes the phenomena
what are the different types of descriptive research
- ex post facto
- survey
- comparative
- developmental
- correlational
- ethnographic
what are the two variables in some types of descriptive research
- classification: analogous to independent variable
2. criterion: analogous to the dependent variable
what is ex post facto research (retrospective)
- retrospective searches for the cause of events that have already happened
- prospective may follow a group for years to see the effect of some factor (smokers 10 years later, cancer)
- help study rare, individual cases of disease
- can only suggest cause
- help investigate phenomena that are not easily manipulated by experiment (unethical)
what is survey research
- random population that is representative is selected
- most common types are questionnaires and interviews
- wide range of data can be obtained
- problems are that they can’t be used for cause-effect and may be biased because those who return surveys may not be comparable to those who don’t
what is comparative research
- measures similarities and differences of groups of people with defined characteristics (attention and memor of people with head injury)
- problem with this is that the similarities and differences might be due to other variables (education level etc)
what is developmental or normative research
- measures changes in individuals over time as they mature
- independent variable is aging which implies that age is the cause of change
- used to create developmental norms
- three methods: longitudinal, cross-sectional and semi-longitudinal
what is longitudinal research
developmental research where the same participants are studied over time
what is cross-sectional research
developmental research where researchers select subjects from different age levels and study them simultaneously
what is semi-longitudinal research
developmental research where the total age range to be studied is divided into several overlapping age groups. Participants chose are at the lower end of the age group and followed for that span
what is correlational research
- tries to find relationships between variables
- a positive correlation is when high values of one variable predict high values of another
- a negative correlation is when high values of one predict low values of the other
what is ethnographic research
observation and description of naturally occurring phenomena
- researchers immerse themselves in situations being studied to gain deeper understanding
- researchers conduct detailed observations, making notes and video
- qualatative
what is internal validity
degree to which data reflects a true cause-effect relationship
no confounding variables
what are threats to internal validity
- instrumentation
- history
- statistical regression
- maturation
- attrition
- testing
- subject selection
- biases
- interaction of factors
what role does instrumentation have in internal validity
- problems with measuring devices
- mechanical instruments may not be properly calibrated
- standardized tests administered to inappropriate samples, or on ESL
- differences in observers
what role does history have in internal validity
participant’s life events that occur during the course of an experiment that may affect the results (child receives speech therapy during the time)
what role does statistical regression have in internal validity
refers to a behavior that goes from an extreme high or low point to an average level
many clients seek treatment when the problem is worst
natural improvement may happen
what role does maturation have in internal validity
biological changes within participants can have an effet on dependent variable
ex: language stimulation program conducted on kindergartners over a year run risk of maturation affecting results
what role does attrition have on internal validity
participants lost over the time affect the final results
what role does testing have on internal validity
changes that occur in a dependent variable because it has been measured more than once
ex: questionnaires participants fill out before and after treatment are reative. Client may show significant attitude change on posttreatment questionnaire, even though their attitude may not have changed
what role does subject selection bias have on internal validity
subjective factors that influence the selection of who participates in a study
to control for this either randomly select or match
what is external validity
generalizability of research findings
single subject designs cannot be generalized
external validity can be threatened by the Hawthorne effect, multiple-treatment interference, and reactive effects of pretesting
what is the Hawthorne effect
study results are affected by the participants knowledge that they are taking part in an experiment
what is multiple-treatment interference
positive or negative effect of one treatment over the other when 2 or more treatments are evaluated in a single study and all participants receive all treatments
what is pretest, posttest sensitization to treatment
pre and posttest might sensitize the participants so that the treatment effect is enhanced. The participants become more receptive to treatment
what is a nominal scale
items are classified into named groupings that do not have a numerical relationship.
Never sometimes always
or labeled under diagnostic categories
what is an ordinal scale
numerical scale that can be arranged according to rank. The numbers do not have a mathematical meaning
1= strongly agree 2 = agree etc
what is an interval scale
numerical scale that can be arranged according to rank. the interval between the numbers are equal
1….2….3….4….5
what is a ratio scale
same properties as an interval scale but values related to an absolute zero point, where zero suggests an absence of what is being measured
0…1…2…3…4…5
what is a type 1 error
when the null hypothesis is rejected but true (false positive)
•Called a “type one error”
•When the researcher rejects the null and concludes that a study supports the research hypothesis when the null is in fact correct.
what is a type 2 error
fail to reject the null hypothesis when not true (false negative)
•Called a “type two error”
•When the researcher fails to reject the null hypothesis but it should have been rejected