Midterm Flashcards
Constructivism
Knowledge generated by individual perception, construction of reality
Methodology
Systematic process used to gather information
Epistemology
Beliefs about how knowledge is generated
Fidelity
Consistency with which a treatment is delivered
Peer reviewed articles
Published in established journals, accepted under double-blind peer review process (meaning authors and reviewers are unknown to each other)
Types of literature review
Linear versus integrated
Goals of literature review
What is known and what still needs to be learned
Linear literature review
Organized by study (a named and then summarized)
Integrated literature review
Research organized by topic, each topic summarized noting confirming and disconfirming evidence from individual studies
Experimental research design
Makes comparisons between groups. True experimental means groups are randomly assigned and there is a control group
Pre-experimental research design
No random assignment and no control group, not generalizable. Single group: pre-test, intervention, post-test
Quasi-experimental research design
Control group but no random assignment. Used to describe differences between preexisting groups, where assignment can’t be random. (independent variable cannot be manipulated)
Correlational research design
Most dominant form. Explores how phenomena relate to each other, no need for control group. Usually involves regression procedures, canonical correlation, structural equation modeling and other advanced methods
IRB
Institutional review boards, mandated by the Common Rule (sub part A) from 45 code of federal regulations of the U.S. Dept. Of health and human services.
Minimal risk
Harm or discomfort can’t be greater than those encountered in daily life
Informed consent
Required for human subject research. Individual must be informed of the duration, procedures and point of the research; the risks and benefits; the manner of maintaining confidentiality or anonymity; any processes that involve more than minimal risk and procedures in case of injury; contact info for researcher and IRB; statement that participation is voluntary and participant can withdraw at any time
Assent
A child’s affirmative agreement to participate when legal consent is not possible
Conflicts of interest
Objectivity: researcher feelings, beliefs or attitudes incompatible with responsible and unbiased research
Benefit: researchers receive some type of welfare, service or profit for engaging in research
Not a dealbreaker but just be addressed
Vulnerable populations
Children, prisoners, pregnant women, mentally disabled persons, economically or educationally disadvantaged persons.
IRBs concerned with coercion and assent (and for children), and representation (by the institution in question, on the IRB). Requires additional provisions to protect, so document protocol, risks and benefits
Unanticipated problems
IRB requires written procedures for dealing with them and reporting promptly to IRB. OHRP distinguishes these from adverse effects.
Adverse events
Physical or psychological harm. Need to be reported if unanticipated, only if it happens as a result of participation. Might result in revised protocols, suspension, termination of study.
ACA code of ethics, section G
Expands on the common rule. Subsections for:
Research responsibilities, rights of participants, managing and maintaining boundaries, reporting results, publications and presentations
IRB protocol
Research project should be worthwhile.
Provide the following info:
Description of research
Participants
Risks and benefits
Informed consent
Confidentiality or anonymity and data security
Methods and measures
Generalizability
Can findings be attributed to larger population
Target population
The population of interest in a study
Accessible population
People (representative of the target population) that the researcher can access
Operational definition
Clear description of what is to be measured (usually based on some construct that can’t be directly observed)
Simple random sampling
All individuals in an accessible sample have equal chance of being selected (usually chosen by random number generator)
Stratified random sampling
Equal chance of participation but places emphasis on desired characteristics (also you might choose 50% Christian / 50% non-Christian)
Cluster random sampling
Random sampling from groups rather than individuals (a common in education research)
Systematic random sampling
Combination of cluster and simple random sampling, with potential participants placed into groups, and then random number from each group (like 13th person in each group) is invited
Nonprobabability sampling
Aka convenience sampling. Recruits and selects members who are accessible to researcher. Most common type in counseling research. Limited generalizability.
Representative sampling
Most counseling research participants are volunteers, so sampling is not random. But we can show representativeness of sample by including demographics, incl. frequencies and percentages for sex, ethnicity and mean and standard deviation for age
Random assignment
Process by which all extraneous variables that could confound study are dispersed among the various groups in study. Only possible when the groups being compared can be assigned.
Extraneous variables
Anything that can get in the way and confound data: random dispersement theoretically distributes all this unknowable stuff equally among groups
Narrative data
The data in qualitative research, usually coded into themes
Theoretical sensitivity
The personal, professional and research experiences that facilitate researcher providing meaning and understanding to qualitative data
Trustworthiness
Rigor and credibility of a qualitative study. Some features are: prolonged engagement, persistent engagement, triangulation, peer debriefing, member checks, audit trail
Purposeful sampling
Researcher intentionally selects participants in qualitative research
Construct
Phenomenon of interest that cannot be measured directly
Nominal
Discrete, categorical variables (sex, ethnicity, religion)
Ordinal
Discrete variables representing an ordered sequence with unspecified magnitude between values (high, middle, low SES; or Likert-style ratings)
Interval
Ordered sequence of variables with values that are equidistant. No true zero (no absence of phenomenon) exists. Rarely used in counseling. Usually quasi-interval variables instead, like (subjective) levels of pain on a scale.
Ratio
Like interval scale variables but include a true zero. Usually not constructs but phenomena that can be observed directly, like age or number of years in school.
Converting variables between scales
Can only be done from higher to lower, e.g. ratio to nominal:
Ratio: 9 credits enrolled
Ordinal: 6-9 credits
nominal: full time enrollment
Validity
Degree to which evidence and theory support interpretations of test scores
Evidence of validity (of a measure)
based on test content (a aligned with existing theory)
based on response process (how people perform)
based on internal structure (psychometric properties - .40 or higher)
based on consequences of testing
based on relationship to other variables
Mean
Average score for a set of observations
Mode
Most frequent score in a set of observations
Statistical error
Inaccuracies
Variance
A measure of error, represents the squared value of the standard deviation.
Range
Another measure of variability, representing the distance between highest and lowest scores
Normal curve
Mean, median and mode are equal; area under curve can be approximated with 68-95-99 rule. 68% have score within 1 standard deviation from mean.
Measure of relationship
Extent to which researcher can explain relationship between two variables, referred to as correlation. Pearson’s r is the most common correlation coefficient
Assignment
Process of placing individuals into experimental or control groups
Instrumentation
The validity of the measures of a particular study.
Types of error
Type 1: rejects null hypothesis when it should be retained (people believe the boy who cried wolf first time)
Type 2: retains null hypothesis when it should be rejected (people don’t believe when a wolf is really there the second time)
Statistical versus practical significance
Statistical significance is whether the event is pure chance or not (effect/error), practical significance is whether phenomenon is meaningful or not; practical significance is the extent to which group differences/relationships exist (size of effect)
Correlational analysis
No manipulation of independent variable, no comparison of treatment or effect
Pearson’s R
Product-moment correlation coefficient
Expression of linear relationship between two variables
Indicates whether correlation is statistically significant
Regression
Examines a linear relationship between variables to establish a prediction model. (values of predictor variable used to predict values of a criterion variable)
Simple regression
Tests the linear relationship between two variables (one predictor, one criterion).
Y = a + bX
Multiple regression
One criterion variable but two or more predictor variables. Examines linear relationship between, e.g., quality of relationships and friendship, and interpersonal relations.
Y=a+b1X1+b2X2…
Moderation
An interaction between predictor and criterion variables. E.g., the relationship between teacher rating and teaching style is moderated by teacher’s attractiveness (correlation is stronger when teacher is attractive)
Mediation
Another variable is used to explain relationship between predictor variable and criterion variable: relationship weakened when third variable is controlled for. E.g.,
Chi square
Test for association used to find relationship between two categorical variables. Test statistical significance of a relationship.
Between group designs
Usually either to identify how a treatment differs between groups, or to describe how groups vary across a domain o/r construct (e.g., gender differences when processing issues of forgiveness)
Univariant analysis
Between-groups analysis with one or more independent variables and one dependent variable
Multi variant analysis
Between-groups analysis with one or more independent variables and two or more dependent variables
IV
Independent variable in between-groups analysis. A discrete variable, categorical in nature (the group(s) being studied)
DV
Dependent variable in between-groups analysis: typically a continuous variable, usually some kind of measure (of outcome)
Experimental validity
Impacts generalizability and refers to influences inside and outside the study
Internal experimental validity
Whether any change in the DV was a result of change in or manipulation of the IV
Maturation
Threat to internal experimental validity; refers to change that occurs in participants over time
History
Threat to internal experiment validity: refers to events outside study that can affect participants
Testing
Threat to internal experimental validity: occurs when a measure is administered two or more times
Instrumentation
Threat to internal experimental validity: refers to accuracy, consistency and utility of administered measure
Statistical regression
Threat to internal experimental validity: occurs when scores are on low or high end of a measure administered more than once (baseline effect and ceiling effect)
Selection bias
Threat to internal experimental validity: occurs when two or more groups in study are unequal across the phenomenon of interest. Or the groups are so different that difference in DV can’t be attributed to group differences
Mortality
Threat to internal experimental validity: refers to participants not completing study
Random assignment
Randomly designating individuals to groups. Addresses threats to internal experimental validity
Statistical control
Addresses threats to internal experimental validity by controlling for inequality between groups at outset or matching traits between groups
Covariates are variables that could impact a study
External experimental validity
Extent to which intervention or treatment may be extended to a variety of settings or environments
Must consider artificiality of setting
Interference of prior treatment
Threat to external experimental validity. Results can be corrupted if participants are predisposed to a past history of treatment/intervention
Experimental design
Has 3 key aspects: random assignment, a manipulated independent variable, and a measure of effect or change. Often in counseling the control group will be another comparison group.
Post-test only control group
Simplest experimental design. Treatment group gets intervention, control group gets none, both receive some kind of outcome measure
Pretest-posttest control group
Appropriate for quasi-experimental design. To establish group equivalence, both control and treatment groups are given a pre-test (then post-test following intervention/non-intervention). One drawback is the testing effect.
Solomon four-group design
To control for testing effect, 2 treatment groups get a pretest and two don’t, everything else is the same
Statistical test
Determines whether a statistically significant difference exists between groups and what it’s magnitude is
Z test, t test, F test
Z test
Rarely used because it requires a sample greater than 30. Determines whether statistically significant difference exists between a sample and a known population mean. Assumes a normally distributed population in which population mean and variance are known
t Test
One sample t test: tests whether a statistically significant difference exists between a sample and a hypothesized mean
Independent samples t test: evaluate significant difference between 2 means
F test
Tests used to evaluate mean differences between two or more groups in univariate and multivariate designs. Extension of t Test.
Survey and longitudinal research
Studies to understand changes over time in specific populations, or differences in respondents who share similar qualities. A within-groups research design.
Panel studies
Classic within-subjects design. Same respondents over two points in time.
Subject to internal threats to validity, like testing effect and mortality. Might include qualitative component.
Cohort studies
Study a specific population with a common characteristic over time. Participants are different at each administration, which reduces testing effect.
Trend studies
Like cohort studies (specific population over time), but constant common characteristic is not necessary. Large, representative sample is important.
Meta analysis
Statistical method of quantitatively summarizing the results from many primary studies to get a more precise estimate of intervention effects (with greater statistical power, greater diversity in population samples, more generalizable).
Either looking at mean effect across eligible studies or variables that influence the mean effect.
Phenomenology
Understanding how humans make sense of lived experience. Major type of qualitative study.
Reliability
Accuracy and consistency of scores on a measure
Test-retest reliability
Scores should be consistent and correlate (greater than .70) between two administrations of same test with same sample
Parallel forms reliability
Researcher compares two or more alternative versions of a measure to see if scores are consistent/correlate
Split-half reliability
Half of a measure is compared to the other half to determine if scores are consistent/correlate
Cronbach’s alpha
Aka coefficient alpha. An estimate of reliability. Creates a coefficient from all possible split-half analyses.
Program evaluation
Systematic collection of information about a program or aspects of a program in order to make decisions about it. “What do I need to know in order to make the decisions I need to make?”
