Biostatistics Flashcards
Branch of statistics that summarizes information about a group based on the actual data collected
Descriptive
Branch of statistics where conclusions are prepared based on the data collected, with predictions made that go beyond the sample analyzed but are generally related to different but related situations
Inferential
number that describes the population
Parameter
number that describes the sample
Statistic
reflects reproducibility or true exactness
Precision
reflects closeness to the true value
Accuracy
characteristic that varies between different individuals (gender, blood type)
Between-subject variable
characteristic of an individual that varies with time (blood pressure, heart rate)
Within-subject variable
Data type involving separate, indivisible categories
Discrete
Data type involving infinite number of possible values that fit between any two adjacent values
Continuous
Scale of measurement: Names or classification data, no order, arbitrary labels
Nominal
Scale of measurement: Ordered data, ranks
Groups in sequence (one is better than the other)
Comparative quality or rank order
Ordinal
Scale of measurement: Ordered categories with all intervals equal
Categories are randomly assigned
Exact different among groups (how much better)
Quantitative, mean and standard deviation
Interval
Scale of measurement: Ordered categories with a fixed absolute zero
Interval with TRUE zero (real difference)
Ratio
Skew where left tail is longer; mass of the distribution is concentrated on the right of the curve
Negative skew
Skew where right tail is longer; mass of the distribution is concentrated on the left of the curve
Positive skew
Peak sharpness where there is a normal distribution
Mesokurtic
Peak sharpness where peakedness is greater than normal distribution
Leptokurtic
Peak sharpness where peakedness is less than normal distribution
Platykurtic
Percentage of data points exist that within 1 standard deviation from the mean
68%
Percentage of data points exist that within 2 standard deviations from the mean
95%
Percentage of data points exist that within 3 standard deviations from the mean
99%
How to calculate joint probability of two independent events
P1 x P2
How to calculate joint probability of two mutually exclusive events
P1 + P2
How to calculate joint probability of non-mutually exclusive events
(P1 + P2) - (P1 x P2)
Do larger sample sizes have wider or narrower confidence intervals?
Narrower
Z-score used for 90% confidence interval
1.645 ~ 1.5
Z-score used for 95% confidence interval
1.96 ~ 2.0
Z-score used for 99% confidence interval
2.576 ~ 2.5
Equation to calculate confidence interval
Mean +/- Z * (stand. dev. / square root n)
Equation for standard error of the mean
Standard deviation / square root of sample size
Statistical test used on 2 interval groups (e.g. height and weight) to show a linear relationship
Pearson correlation
Statistical test used on 2 nominal groups (e.g. single vs married)
Chi-squared
Statistic test used for 1 interval and 1 nominal group (2 groups only)
E.g. blood pressure by gender
t-test
Statistic test used for 2 groups (1 interval and 1 nominal) that are linked data pairs
E.g. blood pressure before and after NBME
Matched pairs t-test
Statistical test used for 2 or more groups (at least 1 nominal and 1 interval)
E.g. pain improvement by two drugs in two age groups
One-way ANOVA
Statistical test used to 2 or more groups, 2 nominal variables
E.g. blood pressure in men and women from US and Canada
Two-way ANOVA
Type of error involving incorrect rejection of a true null hypothesis (false positive)
Chance is alpha value (P value)
Type I error
Chance of type I error is given by this
Alpha value (P value)
Type of error involving incorrectly retain a null hypothesis (fail to reject) a false negative
Probability is beta (1 - power)
Type II error
Chance of type II error is given by this
Beta (= 1 - Power)
likelihood a study will detect an effect then there is one to be detected
Statistical power
Statistical power increases as this type of error decreases
Type II
5 steps in evidence-based medicine
Ask, Acquire, Appraise, Apply, Audit
PICO(T) is used for this step in evidence-based medicine
Ask: convert information need into an answerable clinical question
In evidence-based medicine, P in PICO describes this
Problem or population (patient)
In evidence-based medicine, I in PICO describes this
Intervention
In evidence-based medicine, C in PICO describes this
Comparison
In evidence-based medicine, O in PICO describes this
Outcome
