research - quiz #2 Flashcards
ANATOMY OF AN ARTICLE
Abstract
Introduction
Methods
Results
Discussion/Conclusion
References
abstract
summary/ highlight of paper
introduction
most references are here
methods
“how was the study carried out?”
results
“What Happened?” qualitative or quantitative
objective data only - demographics & outcomes analysis - visual presentation of data
discussion/ conclusion
“Why does it matter?” authors interpretation of the data
tying results back to research question
references
“how does this relate to past research?”
SELECTION BIAS
-which articles you select to include in your review
-which subjects are selected to participate in study
PUBLICATION BIAS
studies with positive results are more likely to get published
HAWTHORNE EFFECT
acting differently because you know you are being watched
RECALL BIAS
issue with retrospective studies: people may not recall things accurately
-relying on participant memory
STUDY HETEROGENEITY
studies in a lit review are too different to be compared
NULL HYPOTHESIS
the idea that your hypothesis will not work
= opposite → strengthen credibility
inferential statistics: disprove the null hypothesis
MEAN, MEDIAN, MODE
measures central tendency
MEAN
average
add them all & divide by number of items
-affected by extremes
MEDIAN
the middle number in a set
-less affected by extremes
MODE
the number that appears the most
-more useful for categorical data
RANGE
distance between the highest and lowest values
DESCRIPTIVE STUDY DESIGN
-describes clinical experiences, thoughts & observations
-can not demonstrate cause & effect
-lack control/comparison group
-can be used to form hypothesis, leading to further investigation
TYPES of descriptive study design
Case Study
Case Series
Correlational Study
Qualitative Study
EXPERIMENTAL EXPLORATORY STUDY DESIGN
(cross-sectional, case-control, cohort)
-evaluates efficiency
-researcher has active role in intervention (controls some variables)
-relationship between exposure & outcome
-ethical considerations as researcher is controlling environments
OBSERVATIONAL/ ANALYTICAL EXPLORATORY STUDY DESIGN (cross-sectional, case-control, cohort)
-seeks to establish cause, factors & predictors
-investigator observes what is happening
-relationship between exposure & outcome
-less ethical consideration as exposures & outcomes happen naturally
LITERATURE REVIEW STUDY DESIGN
-summarize existing evidence on a topic
*question is posed:
-guidelines are set for which types of research will be included in review
-research is analyzed to see how strong it is
-important; if result repeated, more likely to be true!
= ‘reproducibility’
TYPES of lit review study design
Narrative Review
Meta Analysis
Systematic Review
P-SCORE/ VALUE NUMBER (#VALUES & MEANING)
-tells odds of getting particular result
-based on number of participants in study
-if p score small enough, conclude that outcome is not random
-less than 5% is key - usually
-statistical significance = p<0.05
-rules out CHANCE as possible explanation for outcome
FALSE POSITIVE
when the null hypothesis is falsely rejected (alpha)
Usual levels: no higher than .05 for alpha
FALSE NEGATIVE
when the null hypothesis is falsely accepted (beta)
usual levels: no lower than 80% for beta
HIGH LEVEL OF EVIDENCE
1 = systemic reviews
2 = critically appraised topics
3 = individual articles
4 = randomized controlled trials
5 = cohort studies
6 = case control studies
7 = background information
EXTERNAL VALIDITY
can the findings be generalized to a larger group than the one in the study
INTERNAL VALIDITY
extent to which observed results represent the truth in population we are studying
STATISTICAL VALIDITY
is the statistical analysis properly chosen and used
THREATS TO RESEARCH VALIDITY - external
lack of Randomization, lack of control
THREATS TO RESEARCH VALIDITY - internal
confounds, bias, history, maturation, testing, instrumentation, regression, selection, mortality
THREATS TO RESEARCH VALIDITY - statistical
sample size (power), lack of standardization
TYPES OF VALIDITY
-Face: makes sense according to a lay person
-Content: makes sense according to an expert
-Predictive: properly measures an outcome
-Concurrent: results are same as other similar outcomes
-Discriminant: scores are high & low as expected
-Construct: measures what it is supposed to
NOMINAL VARIABLE
= category
-can be dichotomous – ie, yes or no
-or can have more categories – ie, Blood type (A, B, AB, O)
CONFOUNDING VARIABLE
an unmeasured third variable that influences both the supposed cause and the supposed effect
-is there another variable that could explain this outcome?
Lack of control group
“Chicken or the egg” problem – temporal relationship - which came first?
CORRELATION
mutual relationship or connection between two or more things
*correlation does NOT mean causation
RELIABILITY
how consistently a method measures something
TYPES of reliability
-Inter-rater: do different raters get the same results
-Intra-rater: does the same rater consistently get the same result
-Test-Retest: is the same result repeated, time and time again
ATTRITION
the loss of study units from a sample
= a reduction or decrease in numbers, size, or strength
COMPLIANCE
willingly do what they are asked to do
patient compliance
participant Compliance & Attrition
-adherence should be achievable
-intention to treat analysis
-includes ‘dropouts’ in statistical analysis
HETEROGENEITY
signifies diversity/ variability
