Naplex: Biostatistics

Front 1

are outliers reflected in the mean value (arrange all numbers, pick the middle one)

Back 1

no, it better represents data that is skewed...ex professor ranking, if class is small median is better

Front 2

null hyp

Back 2

states no difference in 2 groups (drug a = drug b)

Front 3

P value

Back 3

probability that obtaining a test statistic as extreme as the one that was actually observed, assuming null hypothesis is true.

Front 4

type 1 error

Back 4

null hyp is tue, yet rejected in error. prob of type 1 error is represented by alpha. is it is 5%then 5% of the time the null hypothesis will be rejected in error, falspe positive (court finds an innocent person guilty and sends them to jail)

Front 5

type 2 error

Back 5

null hyp is false yet accepted in error... we conclude there is no difference when there actually is a difference. prob of making a type 2 error is Beta

Front 6

2 types of variables?

Back 6

discrete, continuous rando

Front 7

def of discrete variable?

Back 7

nominal and ordinal:

Front 8

nominal?

Back 8

can be divided into groups or categories (marital status, gender, ethnicity)

Front 9

ordinal?

Back 9

usually a scoring system (ex likert scale) cannot be measured/quantified. ex a trauma score of 4 is not necessarily twice as big as a score of 2

Front 10

continuous random variable?

Back 10

ex height weight, can presumably take an infinite amound of possible values, can be fractions

Front 11

statistical power?

Back 11

prob that the null hyp will be rejected if it is false. a higher statistical power means that we can be more certain that the null hypothesis was not rejected incorrectly or achieve a lower type 1 error

Front 12

confidence interval?

Back 12

statistical range with a specified probability that a given parameter lies within the range or said another way, the porportion of times that an estimate will be correct. if CI is 95% it means that 95% of the time, you are confident that the parameter will fall in this range

Front 13

Odds ratio?

Back 13

ex. if there are 100 people, 40 people had an event and 60 didnt...40/60=0.66...odds ratio of 1 is no difference in the groups

Front 14

Relative risk or risk ration

Back 14

ratio of risk in the treatment group to the risk in the control group

Front 15

standard deviation def

Back 15

the SD is a statistic that tells you how tightly all the various examples are clustered around the mean in a set of data

Front 16

if SD is normal distribution?

Back 16

68% of values are within 1 SD, 95% are within 2 SDb

Front 17

when is SD appropriate?

Back 17

if normal distribution continuous data

Front 18

interquartile range (IQR)

Back 18

difference between the 1st quartile (25th percentile) and the 3rd quartile

Front 19

Standard error of the mean (SEM)

Back 19

estimate of the standard deviation, taken from a sample group

Front 20

case control study? look forward or back?

Back 20

retrospective, cased used of stubjects with the intervention and controls, cases compared to the controls

Front 21

cohort study? look forward or back?

Back 21

prospective, study group (cohort) follwoed over time and outcomes are copared to a subset that were not exposed to the intervention

Front 22

controlled clinical trial

Back 22

study group compared in a controlled setting

Front 23

Gold standard for doing a clinical drug trial is a randomized placebo controlled double blinded multicenter trial with adequate statistically power (p value under 0.05)

Back 23

ture

Front 24

cross sectional study?

Back 24

for ex. if people who have smoked for a given period of time have a higher incidence of heart diseaseat 2 years

Front 25

frequency distribution

Back 25

used for lists of data ranked from lowest to highest

Front 26

histogram

Back 26

tabulated frequencies, shown as bars

Front 27

what is regression

Back 27

ability for 1 or more variables to predict another (risk of GI bleeding in pts on high dose chronic nsaids)

Front 28

multiple regression

Back 28

involves many predictors that can affect a variable, and there will be many r values. this means that regression can be used to hold one value constant, so that is doesnt affect other variables. regression models to control for confounding variables is common in drug studies