Research & Stats Flashcards
decide if study is ethical
Institutional Review Board (IRB)
prediction of relation
hypothesis
can take on different values
variable
outcome of interest, measured to see if it changes in a study i.e. the plant
dependent variable
variable manipulated in the study i.e. the water
independent variable
named attribute, order doesn’t matter, ex: martial status, blood type
nominal variable
named attribute, order does matter
ex: stages of cancer
ordinal variable
can take on any value in a range
ex: temp, weight, speed
continuous variable (vs. discrete)
only 2 possible outcomes
ex: disease or no disease
dichotomous variable
hypothesis, methods used
methods
Am I Literally Making Rice Despite Carb Intolerance?
Research Report Components
-can draw associations, but not casual relationships
-no hypothesis tested
descriptive research
-descriptive research types
-Quit Sharing Credit Cards
-Qualitative, surveys, correlation, case report
-tests hypothesis
-can draw causal relationships
analytical research
formative evaluation
pilot study
data as words (narrative vs. numbers)
ex: focus group
qualitative research
-single subject or group w same condition
-case report/study
-series (many together)
-study examines if 2 variables are related
-no manipulation of variables
-ex: relationship between stress and illness?
correlation study
X-Axis (horizon)
Independent variable
Y axis (vert)
dependent variable
-variables increase/decrease together
-move in same direction
-ex: shoe size increases with foot length
positive correlation
correlation strengths
-0.4-0.7 moderate
-below weak, above strong
-random group assignments
-no selection bias
-ex: double blind study
experimental study
-gold standard for causal relationships
-uses random assignment and control group receiving placebo
RCTs
participants stay in same group
parallel RCT
-2 period
-washout phase to reduce carryover effect
-participants switch groups, receive opposite txt
-participants act as their own control
cross over RCT
-Queezy there’s no random assign or control group
-ex: time series
Quasi-experimental
-risk factor exposure
-retrospective (recalling past behaviours i.e. smoking)
-ex: disease (cases), no disease (control)
case control study
-groups of individuals who share commonality (behaviour or disease)
-healthy participants exposed to a risk factor
-followed over a period of time or look back
-other name “incidence
cohort study
studies current disease prevalence
cross-sectional
PRISMA, CONSORT, QUOROM, MOOSE
Types of meta analysis
-can you apply the study results to general population?
external validity
-REpeatability of results
Reliability
-how likely a test is to correctly identify ppl WITH a disease
-ex: initial screening
sensitivity
-how likely a test is to correctly identify ppl WITHOUT a disease
-specific diagnosis testing
specificity
-participants selected by convenience or judgment;not random
non-probability sampling
-summarize, describe & synth collected data
-ex: mean, mode, range
descriptive stats
-analysis of data, prediction beyond study
-ex: hypothesis test
inferential stats
most frequently occurring value in data set
mode
how spread out are the data values relative to average
standard deviation
mean of 2 groups
T-test
mean of 3 groups
ANOVA
outcome of study isn’t luck, real finding
statistical significance
P<0.5
significant
P<0.1
very significant
P>0.5
not significant
smaller the P value
higher statistical significance