-
median
- (n+1)/2, the median will be between these two numbers
- least susceptible to outliers, not useful for large ranges
-
standard distribution
- can be transformed from normal distribution
- with zero as the mean and st. dev of 1
-
normal distribution
all central tendency (mode, mean, median) are the same
-
skewed distributions
neg skewed distribution with tail on the left (mean towards the left, then median and then mode)
-
bimodal distribution
two or more peaks
-
range
- heavily affected by outliers
- approx te st. dev as 1/4 of the range
-
interquartile range
- Q1=multiply n by 1/4
- Q3=multiply n by 3/4
- if whole number, quartile is the mean of the value at this position and the next highest, round up in other cases)
- IQR=Q3-Q1, range is 1.5(IGR) below Q1 or above Q3
-
outlier
- more than 3 st. dev away from mean
- key to determining whether or not the mean is an appropriate measure, could be measurement error
- throw out or not (DET before study)
-
independent events
outcomes do not impact each other
-
mutually exclusive outcome
cannot occur at the same time (one cannot flip head and tail in one throw)
-
exhaustive
no other possible outcomes
-
probability
- "and" multiply the probabilities
- "or" add the probabilities and subtract the probably of both happening together
-
probability of 10 births having at least one boy
1-probability of having all girls
-
p value and significance level (alpha)
if p value is greater than sign alpha, then we fail to reject the null hypothesis
-
type 1 error
report difference between two populations when there isn't one
-
type 2 error
we report no difference when there is a difference
-
confidence
report no difference when there is no difference
-
power
report difference when there is difference
-
confidence interval
- reverse of hypothesis testing; instead of using p score, we use 95% confidence (30 is the mean)
- 30-(3)(1.96) to 30+3(1.96)
- 95% confidence that true mean value is between those two levels
- 3 is standard dev.
-
histograms and bar graphs
histograms show distribution (mode) of variables, while bar graphs compare variables
-
box and whisker
- Q1 and Q3 (Q2 median in middle line)
- range is 1.5(IQR)
-
correlation
could be direct, inverse but does not imply causation
-
respect for persons
- honesty
- informed consent
- confidentiality
-
ethics of medicine
- beneficence
- nonmaleficence
- autonomy
- justice
-
justice
- selection of research topic and execution of research
- morally relevant differences (appropriate reasons to treat people differently): transplant more likely to give to younger individual
- no difference in perceived benefit between individuals, then all individuals should assume equal risk; particular population is likely to benefit that pop should assume higher proportion of risk
-
equipoise
if it becomes evidence that one treatment option is clearly superior before a study is done, then study must stop, because providing inferior treatment is a net harm.
-
popluation
complete group of every individual that satisfies the attributes of interest
-
parameter
information that is calculated using every person in a population
-
internal validty
support for causality
-
external validity (generalizability)
samples are representative of target population
-
studies need to have what in order for intervention to be allowed
statistically significantly (not the result of random chance) and clinical significance (notable or worthwhile change in health status as a result of our intervention
|
|