CDIS 3413 1st EditionExam # 2 Study Guide- Simple Random Sampling is a process where everyone has an equal chance of being chosen for the study, and participants are chosen by a chance process.o Benefits: Ensures that results can be generalized to a large group.- Stratified Sampling is an approach that divides the target population into strata, or subgroups that don’t overlap.o Benefit: Makes the results more generalizable since more than one type of group is being tested.- Historical research is a non-experimental method that collects data from existing records and analyzes it.o Sheds light on long-term outcomes of treatmentso Retrospective design whose purpose is to explain unusual phenomena or construct new theories- Ethnographic methods used in qualitative research include:o Participant observation- you can observe as an insider (from within the scene) or as an outsider (from outside of the action, such as in an observation room).o Data is gathered from In-depth interviews Focus groups Document analysis Triangulation Video/audio analysiso There are no limitations to what can be observed.o There is no set ending point of observation.- Experimental research tries to find cause-and-effect relationships between two variables. - Non-experimental research does not have a control group or multiple measurements, and they do not determine cause and effect.o Provide descriptions of behaviors and associations between behaviorso The most common types include: Case studies Ethnographic designs Historical designs Correlational designs Developmental designs Surveys- Benefits of group research designs:o There are many types, one of which can answer almost any research question.o The results will be more generalizable if the sample is a representative one.o Some types can establish cause-effect relationships.- Types of group research designso Single group designs: one group of participants that act as their own comparisons No outside comparison group Pretest/posttest design Problems:- Lack of external comparison- Vulnerable to maturation effects and test sensitization effectso Two-group designs: comparison at two levels of the independent variable Usually done to confirm the presence or absence of a treatment effect Helpful in reducing the influence of confounding variables Types:- Parallel research designs are where participants are assigned to one oftwo groupso One group receives the treatment while the second serves as the control group.o The design is common in speech, language, and hearing research.- In crossover designs the participants alternate between the treatmentgroup and the control group, and each participant acts as their own control.o More economical because it requires less participantso Not recommended if you might lose participant- Experimental- Quasiexperimental- Pretest/post test- Posttest onlyo Factorial designs or multivariate designs include all combinations of the levels of two or more independent variables.  The effects of two or more independent variables can be observed within a single day. Model real life events in which multiple variables may effect behaviors when happening at the same time- Statistical Power: how well your statistics relate to the real world, or how generalizable it is.- Single case designs are less popular, and they focus on the individual, but are not very generalizable.o ABA designs: initial baseline, intervention phase, second baseline (without treatment)o ABAB just adds an extra treatment phase.o ABAC design: compare the effectiveness of different types of treatments Baseline, treatment 1, second baseline, treatment 2 (different from the first treatment)- Multiple treatment interference can be a problemo Simultaneous treatments: two treatment within one person at the same time- Percentage of non-overlapping data is the percentage of data points in treatment phase thatexceed the largest value in the baseline phaseo >90 is large, 90 to 70 is moderate, >70 is small- Correlation is the degree of relation between two variables.o Negative: as one variable increases, the other decreases o Positive: when on increases the other increases and when one decreases the other decreases.- Correlational Research Designs include several groups of participants. Each group represents a point in time along the developmental continuum.o Reduces the threat of participant loss and reduces the cost.- Longitudinal Research Designs observe the same subjects over an extended period of time. o Increased loss of participants- Qualitative Research explores a topic or idea through observation.o For answering questions about how something took placeo Generalizability is not a concern o It is more flexibleo Good for developing theory- Quantitative Research quantifies answers and generalizes findings to a broader populaceo For answering specific questions- Deductive analysiso Uses prior theory to reason from general to specific- Inductive analysiso Reasons from specific to general to generate theories- Credibility/dependability is the extent to which the conclusions of the study accurately represent the participants’ views. The steps for determining this are:o Peer debriefingo Member checkso Triangulationo Saturationo Reflexivityo Audit trail- Steps for performing qualitative research:o Observe events or ask open-ended questionso Record what is observed, said, or doneo Interpret the datao Return to further observe and ask more questionso Repeat steps 2-4o Develop theories to explain the datao Formulate conclusions and generate