Research & Research Stats Review Flashcards
The Research Question
P.I.C.O stands for = ?
- P = Population, patient, problem
- I = Intervention or exposure
- C = comparison
- O = Outcome
Components of Theories
Essential building blocks of theories = ?
Concepts = essential building blocks of theories
Components of Theories
concepts that are intangible = ?
Constructs = concepts that are intangible
Components of Theories
concepts are integrated into a generalized theory = ?
Propositions = concepts are integrated into a generalized theory
Components of Theories
A simplification of a theory; a structural representation of interrelated concepts = ?
Models = a simplification of a theory; a structural representation of interrelated concepts.
What kind of reasoning
Broad to specific = ?
Deductive reasoning:
- Top-down
- Broad to specific
What kind of reasoning
Specific to broad = ?
Inductive:
- Bottom-up
- Specific to broad
3 Pillars of EBP = ?
- Best available research
- Patient values and preferences
- Clinical expertise
Levels of evidence
- 1a = ?
- 1b = ?
- 2a = ?
- 2b = ?
- 3a = ?
- 3b = ?
- 4 = ?
- 5 = ?
- Level 1a: SR/MA of RCTs
- Level 1b: RCTs
- Level 2a: SR of Cohort Studies
- Level 2b: Cohort studies
- Level 3a: SR of case control studies
- Level 3b: Case-control study
- Level 4: case study or case series, cross sectional study
- Level 5: Clinical experience, expert opinion, mechanism-based reasoning
5 A’s of EBP = ?
- Ask
- Acquire
- Appraise
- Apply
- Assess
Four parts of informed consent = ?
- Disclosure
- Comprehension
- Voluntariness
- Competence
3 Ethical principles = ?
- Respect for persons =
- Beneficence =
- Justice =
Groups who are considered vulnerable are = ?
- Prisoners
- Pregnant women
- Children
Assent: the form of consent with children, but still need a yes from parent/legal guardian
Types of harm = ?
- Physical: injury, side effects, or no improvement
- Economic: cost to participate
- Social
Examples of ethical issues in research = ?
- Tuskegee Syphilis study: no equitable distribution
- Nuremberg trials: lack of consent and no benefits
- Henrietta Lacks cells: lacked consent
- Willowbrook hospital hepatitis study: lacked consent
Code of ethics
First formal guidelines, voluntary consent to participate = ?
- Nuremberg code: first formal guidelines, voluntary consent to participate
Code of ethics
Independent review of protocols = ?
- Declaration of Helsinki: independent review of protocols
Code of ethics
Clearly stated research design, informed consent = ?
- National research act: clearly stated research design, informed consent
Code of ethics
The common rule:
Belmont report: the common rule
- Respect for persons
- Beneficence (Do no harm)
- Justice
What level of measurement ?
Nominal: Named Variables
What level of measurement ?
Ordinal:
- Named
- Ordered Variables
What level of measurement ?
Interval:
- Named
- Ordered
- Propotionate interval between variables
What level of measurement ?
Ratio:
- Named
- Ordered
- Propotionate interval between variables
- Can accomodate absolute zero
Reproducibility and consistency of a measurement = ?
Reliability:
- Reproducibility and consistency
- Consistency of a measurement
Accuracy and correctness / Alignment of the measurement with a targeted construct = ?
Validity:
- Accuracy and correctness
- Alignment of the measurement with a targeted construct
Reliability is a prerequisite for ?
Reliability is a prerequisite for validity
Types of evidence for validity
- Face: appears to test what is intended to; judged by users of a test after the test is developed
- Content: do the measurements represent the concept
- More commonly used in questionnaire development
3 requirements:
- Must represent full scope of construct being studied
- The number of items are proportionate to the importance of that component
- No irrelevant items
- Criterion-related: can the outcomes of the test/instrument establish a gold standard, Highest and most objective form, and measured by correlation coefficients.
- Concurrent: scores from a new measure correlate with scores from a well-known measure
- Predictive: outcome of target test can be used to predict a future criterion score or outcome
- Construct: reflects the ability of an instrument to measure the theoretical dimensions of a construct
- Cannot observe but can measure!
- Convergence: correlation with groups that go hand in hand
- Divergence: low correlation on different tests
Types of measurement error = ?
2
- Systematic error: always overestimating or underestimating
- Random error: due to chance; unpredictable
Sources of measurement error = ?
3
- Rater: error in perception or reporting of measurement
- Instrument
- Variability of characteristic being observed
Quantifying reliability
Relative reliability vs. Absolute reliability = ?
- Relative reliability: ratio of total variability between scores to variability within scores
- Absolute reliability: how much of a measured value is likely due to error; expressed in same units as measurement
ICC scores
- ICC > 0.9 = ?
- ICC > 0.8 = ?
- ICC > 0.75 = ?
- ICC < 0.75 = ?
- ICC > 0.9 = best for clinical measures
- ICC > 0.8 = acceptable
- ICC > 0.75 = good
- ICC < 0.75 = poor to moderate
Types of reliabilities = ?
6
- Test - retest:
- An instrument/ tool is capable of measuring a variable consistently
- Conditions that noticeably change over time are not good for this
- Intra-rater
- “with-in raters”: degree that examiner agrees with themselves
- Rater bias: when one rater takes 2 measurements; influenced by memory of 1st score
- Inter-rater
- “between” raters: 2+ raters who measure the same group of people
- Best assessed in a single trial
- Internal consistency:
- Often used to construct and evaluate scales/questionnaires
- Estimates how well the items that reflect the same construct yield similar results
- Internal consistency – split half:
- Take all questions, divide (odd-even), and compare halves
- Alternate form:
- To see if 2 versions of the same instrument are equivalent
MDC vs. MCID = ?
Responsiveness:
- Minimal detectable change (MDC)
- Value at which observed change is NOT measurement error
- Below this score = measurement error
- Minimally clinically important difference (MCID)
- Ability of an instrument to detect minimally important change
Components of Pedro = ?
- Eligibility criteria were specific
- Random assignment
- Allocation concealed
- Were the groups similar at baseline
- Subject blinding
- Therapists
- Raters
- Key outcome measure in >/= 85% of subjects
- Intention to treat
- Was there between-group statistical compairson for at least one key outcome
- Is there a point measure and measure of variability for at least one outcome
- Probability of type 1 error = ?
- Probability of type 2 error = ?
- Alpha: Probability of type 1 error (set before the study)
- Beta: Probability of type 2 error
?
Type I Error (False positive)
?
Correct outcome (True positive)
?
Correct outcome (True negative)
?
Type II Error (False negative)
calculated probability of type 1error = ?
p-value: calculated probability of type 1error (calculated after the study)
calculated value for t-test = ?
t-statistic: calculated value for t-test
calculated value for ANOVA = ?
F-statistic: calculated value for ANOVA
equal variances for RM ANOVA = ?
Mauchly’s W: equal variances for RM ANOVA
- r: correlation coefficient
- r2: the percent variance in one variable explained by the other
effect size for t-test = ?
Cohen’s d: effect size for t-test
effect size for ANOVA = ?
Eta squared: effect size for ANOVA
reliability for continuous data (unitless) = ?
ICC: reliability for continuous data (unitless)
reliability for categorical data (unitless) = ?
Kappa: reliability for categorical data (unitless)
measure of internal consistency = ?
Cronbach’s alpha: measure of internal consistency
Three things we need to know about confidence intervals = ?
- Point estimate: a single value that represents the best estimate of the population value
- Confidence interval: a range of values that we are confident contains the confidence interval
- 95% Confidence interval: If we were to repeat sampling many times, 95% of the time the confidence interval would contain the true population mean.
Multiple comparison tests
- Independent groups ANOVA = ?
- Repeated MeasuresANOVA = ?
Multiple comparison tests - Anova
- Independent groups: Fisher’s, Tukey’s HSD, Bonferroni
- Repeated Measures: Fisher’s, Sidak’s, Bonferroni
Equal Variances - Levene’s Test
What should we know about p-values = ?
Equal Variances - Levene’s Test
- p < 0.05 variances are not equal
- P > 0.05 variances are equal
This is done before the interpretation of the results of the t-test or ANOVA
Utilized for a one-way RM ANOVA and is interpreted the same as Levene’s = ?
Mauchly’s W is utilized for a one-way RM ANOVA and is interpreted the same as Levene’s.
How can we increase power = ?
- increase alpha
- increase effect size
- decrease variance
- increase sample size
Strength of correlations
- 0-0.25 = ?
- 0.25-0.5 = ?
- 0.5-0.75 = ?
- 0.75-1.0 = ?
Strength of correlations
- 0-0.25 = None to low
- 0.25-0.5 = Low to medium
- 0.5-0.75 = Medium to strong
- 0.75-1= Strong
Correlations - Effect Size
- 0-0.1= ?
- 0.1-0.3 = ?
- 0.3-0.5 = ?
Effect size
- 0-0.1= Low
- 0.1-0.3 = Medium
- 0.3-0.5 = Large
Relative Risk
RR of 1.5 means that those with exposure are = ?
Relative Risk
- Used for cohort studies
- Classify participants by exposure
- RR of 1.5 means that those with exposure are 1.5 times more likely to develop the outcome
Odds ratio
OR of 1.5 means that the odds of the outcome are = ?
Odds ratio
- Used for case control studies
- Classify participants by outcome
- OR of 1.5 means that the odds of the outcome are 1.2 times higher in those that have been exposed
Number needed to treat vs. Number needed to harm
- Number needed to treat = The number of patients needed to treat to avoid one poor outcome
- Number needed to harm = The number of patients needed to treat to cause one poor outcome
Diagnostic accuracy
- Sensitivity = ?
- Specificity = ?
- Positive Likelihood Ratio = ?
- Sensitivity: The probability of detecting a true positive
- Specificity: The probability of detecting a true negative
Positive Likelihood Ratio vs. Negative Likelihood Ratio
- Positive Likelihood Ratio: the likelihood that a person who tests positive has the condition
- 1-2 = Small
- 2-5 = Small, sometimes important
- 5-10 = Moderate
- 10+ = Large and often conclusive shift
- Negative Likelihood Ratio: The likelihood that a person who tests negative has the condition
- 0.5-1= Small
- 0.2-0.5 = Small, sometimes important
- 0.1-0.2 = Moderate shift
- < 0.1= Large and often conclusive shift
?
Mann-Whitney U
?
IG ANOVA
?
Krusal-Wallis ANOVA
?
Paired T-Test
?
Wilcoxon Signed-Rank
?
RM ANOVA
?
Friedman’s ANOVA
?
Unpaired T-Test
?
Pearon’s r
?
Spearman’s rho
?
Exposure - Relative Risk
?
Outcome - Odds Ratio
?
Chi Square
