Research Methods I Pt. 3 New Materials Flashcards
Quasi-Experimental Research
Research which manipulates the independent variable but does not randomly assign participants to conditions or orders thereof.
Nonequivalent Groups Design
Between-subjects design where participants are not randomly assigned conditions, the groups in question have an important difference between them. Has a number of extraneous variables, steps can be taken to minimize these extraneous variables
Pretest-Posttest Design
Dependent variable measured before treatment and then after treatment, within-subjects but order is not counterbalanced. If posttest is better than pretest then logical to conclude treatment may be responsible but not the definitive answer.
History
One reason why posttest scores are better and sees other events occurring between tests causing an improvement.
Maturation
The other reason why posttest scores improve, sees participants changing in inevitable ways due to learning and growth.
Regression to the mean
Statistical fact that individuals scoring extremely on variable will score less extremely on next occasion and could point to improvement or lack of in pretest-posttest design.
Spontaneous Remission
Tendency for medical and psychological disorders to improve over time with no treatment.
Interrupted Time Series design
Variant of pretest-posttest design. Time series see measurements taken at fixed intervals of time, interrupted has the time series interrupted by the treatment.
Combination Designs
Type of design combining both nonequivalent and pretest-posttest design, if changes were result of maturation or history then scores should be similar in amount of change.
Qualitative Research
Begin with less clear question or hypothesis, collect large amounts of unfiltered data from small number of people and describe data with nonstatistical techniques, less concerned with general conclusions, more interested in experience of research participants.
Purpose of qualitative research
To generate novel and interesting research questions, detailed descriptions of behaviours in situations, and communicating experiences of group.
Data Collection
Can involve interviews, naturalistic observation, archival data, artwork.
Most common is interviews can be unstructured asking general questions to have participant to talk of interest. Can be structured with a strict script, or semi-structured with some consistent questions and can ask more detailed questions.
Focus Groups
Small groups of people participating in interviews focused on topic or issue, interaction is focus to bring information that one-on-one interview could not, group dynamics are at play and so it helps to be aware of them.
Participant Observation
Researchers participate in group/situation being studied, collect unstructured data, notes, documents, photos, and rationale has important information only able to be interpreted by active member.
Grounded Theory
Identify ideas repeated throughout data, organize into smaller number of broad themes, then write a theoretical narrative or interpretation of data in term of the themes identified. Interpretation focus on experience of participants and supported by quotes.
Debate between Quantitative and Qualitative
Cons of quantitative: lack objectivity, difficult in reliability and validity, no generalization other than those studied.
Quantitative cons: Overlook richness of human behaviour and experience, focus on simpler question with easily quantifiable variables.
Mixed-Methods Research
Combines quantitative and qualitative, take a general observation is a specific group which is then turned into a testable hypothesis by the quantitative group.
Triangulation
Use both quantitative and qualitative methods at same to and compare results, if converge on same general conclusion, reinforce and enrich each other.
Single-Subject Research
Quantitative research studying behaviour of small number of participants, can be 2-10, contrasted with group research with large numbers of participants and examine behaviour in group means, Standard deviation, etc.
What do Single-subject research studies contrast?
Qualitative research, focus on objective behaviour with manipulation and control, structured data and quantitative analysis.
Case Studies, no determining whether specific events are causally related, can be unusual in some way and not representative of people in a general sense.
Assumptions in Single-Subject Research
1) Focus intensely on behaviour of individual participants, individual could have side effect which would be hidden in group study, side effect may be of interest.
2) Manipulate independent variables measure dependent variable, control extraneous variables to find causal relationships.
3) Study strong and consistent effects with biological or social importance, applied researchers look at significant treatments on behaviours which can generalized to the real-world, a social validity.
Experimental Analysis of Behaviour
B.F. Skinner saw it as a subfield of psychology which relies on single-subject research,
Applied behavioural analysis
Roel in contemporary research on developmental disabilities, education, organizational behaviour, and health, and other areas.
Who else uses Single-Subject Research?
Not only behavioural perspective but theoretical perspectives like cognition, psychodynamic, or humanistic to study therapeutic changes with individual clients and document improvement.
Features of Single-Subject Designs
Dependent variable is measured repeatedly over time at intervals
2) Study is divided into phases and test participant in one condition per phase.
3) Change from one condition to next depends on participant’s behaviour, will wait for consistent behaviour before changing- steady stage strategy were dependent variable reaches steady state to make changes easier to detect.
Reversal Designs
AKA ABA design has phase A, baseline established, then phase B begins with treatment introduced, wait for steady state, then remove treatment until steady state once more, phase A. Can reintroduce treatment (ABAB), then go back to baseline (ABABA)
Why Reversal Design?
Possible for something else changing at time of treatment being introduced which caused change. If dependent variable changes with introduction and removal of treatment, then treatment is more likely the cause.
Multiple-treatment reversal design
Has baseline established, then introduce first treatment, then a new treatment which is different than first a ABCACB design.
Alternating Treatments Design
Two or more treatments are alternated quickly on regular interval/schedule, quick and effective in comparing treatments when they are fast acting.
Problem with reversal design
If treatment works, then unethical to remove it.
Dependent variable may not return to baseline when treatment is removed.
Multiple-baseline design
Has baseline established for several participants, then treatment introduced for each one at a different time. If dependent variable changes when treatment is introduced at different times then unlikely to be a coincidence.
Other version sees multiple baselines for same participant with different dependent variables with treatments at different times for each dependent variable.
Third version has multiple baselines in different settings for one participant, if change is found after treatment in each setting then confidence in treatment grows.
Visual Inspectino
Data-analysis in single-subject research, plotting individual data points, observe and make judgments about whether independent variable had effect on dependent variable.
Level
Changes in level of dependent variable from condition to condition sees it being higher or lower than other condition pointing to an effect.
Trend
Gradual decreases/increases in dependent variable across observations, if change with change in conditions then points to an effect, telling when trend changes directions as well.
Latency
Time taken for dependent variable to change after conditions have changed, points to treatment being responsible once more.
Results of Single-Subject Research
Statistical procedures can be used with means and standard deviations.
Alternatively can compute percentage of nonoverlapping data or PND for participants, the most extreme response compared to most extreme response in relevant control condition. The greater the PND the stronger the treatment effect.
Debate between group-/single-research
Single-research con: visual inspection not sensitive to detect weak effects, unreliable with different conclusions, and results cannot be clearly or efficiently summarized and compared across studies.
Steady state strategy was argued to minimize most of the problems, reduce noise or control extraneous variables.
Group Research Cons: Focus on means is misleading, strong positive effect on half people exposed to it and strong negative effect in other half would be cancelled out so treatment has no effects.
Rebuttal: Emphasize importance of examining distributions of individual scores, detected by histogram of data, within-subjects study allows observation at individual level and specify effects.
External Validity in single vs. group research
Single Cons: Difficult to know whether results in few participants can be generalized to population.
Rebuttal: Strong and consistent effects are likely generalized to others in population, even try to replicate with slightly different group.
Group Cons: Studying large group does not solve issue of generalizing to other individuals. Small, medium and large effects in a group would come out as medium effect but when it comes to the individual we can not be sure what kind of effect will actually occur. Also requires careful consideration of similar peoples and situations studied to population of participants and situations.
Complementary Methods
Single Research is good for effectiveness on individuals with a focus on strong, consistent, biological or socially important effects, and when behaviour is interesting.
Group research is ideal for effective treatments at group level, allows to detect weak effects, can help refine treatment and produce larger meaningful effect.
Both methods present different research traditions, analysis of behaviour look to conceptualize research questions in amenable ways to single-subject approach. Others conceptualize to group approach.
