Midterm 1 Flashcards
What makes a fair comparison?
Randomized participants
Observations
Group of people/what surveying
Variables
What’s being measure
Categorical
small number of values
Binary
2 possible values (yes or no)
Nominal
order of categories is irrelevant
Ordinal
order of categories is meaningful
Discrete
values equal to integers
Continuous
values on a continuum
Histogram
graphical display of the distribution of quantitative data
In a stem and leaf plot, which digits are the stem and which are the leaf?
Stem is 1st digits and leaf is last digit
Central tendency
the middle of the data
Variation
how “spread out” the data is
Mean
average of all values in dataset
Median
middle value
Mode
most common value
mean > median
right (positively) skewed
mean < median
left (negatively) skewed
mean = median
symmetrical
Range
smallest and largest values in data
Interquartile Range (IQR)
25th and 75th percentile (1st and 3rd quartiles)
Varience
average of (X-X bar)^2
Standard deviation
square root of varience
What is the box of a box and whisker plot?
25th to 75th percentile
Scatter plot
Illustrates the relationship between 2 quantitative measures
Correlation
a numerical descriptive of the strength of a linear association between 2 variables
r > 0
higher values of one variable correspond to higher values of the other
r < 0
higher values of one variable correspond to lower values of the other
r = +/-1
perfect linear association
r = 0
no linear association
Cumulative incidence
proportion of individuals newly acquiring the disease over a specific period of time
Risk factor
variable that may increase or decrease the chance (risk) of outcome
Relative risk
a measure of the strength of an association
RR < 1
treatment is associated with lower risk of outcome
RR > 1
treatment is associated with higher risk of outcome
RR = 0
no association of treatment with outcome
How do you calculate relative risk?
risk (treated) / risk (untreated)
Risk difference
how much of the disease can be attributed to a risk factor
How do we calculate risk difference?
risk (treated) - risk (untreated)
RD < 0
treatment associated with lower risk of outcome
RD > 0
treatment associated with higher risk of outcome
RD = 0
no association of treatment with outcome
What is the wording when reporting risk difference?
percentage point increase/decrease
What is the wording when reporting relative risk?
Percent increase/decrease
What constitutes stronger association when comparing risk difference?
A value farther away from 0
Incidence Rate (IR)
number of people who develop a condition (incident events) per person in the population over time
What are the units for incidence rate?
per person-years
Censored
when participants drop out or have not had the event by the end of follow-up
Cumulative Incidence
similar to incidence rate but for a specific length of time. can’t calculate if there are drop outs
Incidence Rate Ratio (IRR)
IR (treated) / IR (untreated)
IRR < 1
treatment associated with lower rate of outcome (protection)
IRR > 1
treatment associated with higher rate of outcome (harmful)
IRR = 1
No association between treatment and outcome
Incidence Rate Difference (IRD)
IR (treated) - IR (untreated)
IRD < 0
treatment associated with lower rate of outcome (protection)
IRD > 0
treatment associated with higher rate of outcome (harmful)
IRD = 0
No association between treatment and outcome
Prevalence
the fraction of individuals with the disease at one specific point in time
PR < 1
exposure associated with lower prevalence
PR > 1
exposure associated with higher prevalence
PR = 1
no association between exposure and outcome
Prevalence Ratio
PR
Prevalence Difference
PD
PD < 0
exposure associated with lower prevalence
PD > 0
exposure associated with higher prevalence
PD = 0
no association between exposure and outcome
Population
collection of all individuals (units) that you wish to make inferences about
Parameter
numerical value that would be calculated using all units in the population
Sample
subset of “units”
Estimate/statistic
numerical value that is calculated using all units in the sample
Sampling unit
the individual items or elements of the population that are to be sampled (individual people, census tracts, houses)
Sampling Frame
a list of all units in the population (census tracts, class list, entire households)
Probability sampling
each unit in the population has a known, non-zero chance of being selected from the population to be included in the sample
Simple random sample
randomly selected individuals from the population such that each individual has the same change of being selected
Stratified random sample
divide population into subgroups and take simple random sample from each subgroup
Systematic random sample
randomly choose a starting unit; sample ___th unit after the starting unit
Cluster sample
sample clusters of individuals (households, clinics, schools) and use all individuals in the cluster
Multistage random sample
sample in stages (1st = school, 2nd = classroom, 3rd = student)
Convenience Sample
samples that aren’t selected probabilistically
Generalizability
a measure of how useful the results of a study are for a broader group of people or situations
When is a sampling method bias?
if it produces results that systematically differ from the population
When is a sampling method unbias?
if over repeated sample, the average of the estimates from those sample can be expected to equal the parameter in population of interest
Selection bias
the sampling procedure systematically includes or excludes a portion of the population
Social desirability or response bias
subjects may not answer truthfully
Hawthorne effect
alteration of behavior by the subjects of a study due to their awareness of being observed
Low variability / high precision (bull’s eye)
estimates tightly cluster around each other
Unbiased (bull’s eye)
estimates cluster around the bull’s eye
Margin of error (MOE)
measure of the precision of an estimate and gives a plausible range of values for the parameter
What provide a higher level of evidence, experimental or observational studies?
Experimental
Experimental Study
Exposure or treatment is controlled by the researcher
Replication
Same treatments are assigned to different sampling units to help assess variation in the responses
Randomization
ensure that uncontrolled factors do not bias the experimental results
Observational Study
exposure or treatment is not controlled by the researcher. Researchers collect data from an existing situation
Confounder
a variable that changes the estimate of the association between the main independent variable of interest (exposure) and the dependent variable (outcome) by 10% or more
Mediator
variable in the causal pathway that is associated with exposure and outcome
Case Reports
Describe characteristics of a few individual patients
Ecological Study
a study of risk-modifying factors and health or other outcomes based on populations defined either geographically or temporally
Cross-sectional Study
Samples populations and measures diseased/not diseased and exposed/not exposed at the same time
Case-control Study
Samples diseased and not diseased individuals separately and then measures exposed/not exposed
Odds Ratio
odds of exposed among those with disease/odds of exposure among those w/o disease
OR of exposure comparing diseased to not diseased =
OR of disease comparing exposed to unexposed
Relative risk and odds ratio will always estimate the same direction of association
OR < 1 if and only if RR < 1
OR > 1 if and only if RR > 1
OR = 1 if and only if RR = 1
Odds ration is always at least as extreme as the relative risk
If RR < 1, then OR </= RR < 1
If RR > 1, then OR >/= RR > 1
When the risk of disease is relatively low, the OR and RR are similar
RR = OR
Cohort Study
Enroll healthy participants and follow over time to see who develops disease
Randomized Clinical Trial
Study population —> randomize —> split into intervention and control —> measure outcome
Meta-analysis
a way of combining results from multiple existing studies
Pre-clinical work
testing of a drug in-vitro and in animals to get a sense of toxicity, metabolism, PK, dosing
Phase 1 CT
drug given to usually healthy volunteers to determine safety and appropriate dosing levels. no controls
Phase 2 CT
further explore safety and get an initial idea of efficacy. control group usually used
Phase 3 CT
large, definitive RCT designed to show safety and efficacy. usually double blind, placebo controlled
