Question | Answer |
epidemiologic research focuses on | possible asoication |
Independent variables | variables set/determined by investigator |
dependent variables | the effects that depend upon the independent variables |
Qualitative | description, word not number |
Quantitaive | use rigid, continuous measurement scale |
the order of different type of data (from low to high) | nominal ==> binary (dichotomous) ==>ordinal (ranked) ==> continuous (dimensional) ==> ratio |
Which three types of data are discrete data | nominal, binary, ordinal |
can lower order data be expanded into high order date? | No. Only the higher order data contains more information cn collapsed into lower order data |
the importance of statistic in clinical setting | it helps us predict what might happen in the future of our patients |
normal distribution | the trendline of population data which exhibit "bell-shaped" |
example of non-parametric data | nominal, binary (dichotomous), ordinal |
Parametric data | data that can be easily described by "mean, mode and median" |
parameter | data that are characteristics of that population that help to describe or define the population |
pencentiles | percentage of observation below the indicated point when all the observations are ranked in descending order. Mean = 50th percentile |
a measure of dispersion, lowest to highest value | range |
measure of variability of data about the mean (sum of squared deviation from the mean) | variance |
square root of variance-a smaller number used to describe the amount of "spread" in the frquency distribution | standard deviation (SD) |
horizontal stretching of a frequency distribution to one side or the other ==> create a long, "thin" tail of the data distribution | Skewness |
Vertical stretching or flattening of the frequency distribution | Kurtosis |
type of hypothesis testing which proceed form general to specific | Deductive |
Type of hypothesis testing proceeds from the specific to the general | Inductive |
: there is statistically significant difference between 2 groups | alternative hypothesis |
: there is statistically significant difference between 2 groups | null hypothesis |
is the probably of incorrectly reject H0 when it is actually correct. | alpha |
the probably of incorrectly “fail to reject” (accept) H0 when it is actually incorrect | beta |
: the probably that 2 group are different with respect to the data measurement | p value |
what is p-value when we say the difference is statistical significant | p<0.05 |
what is p-value when we say the difference is statistical insignifciant | p>=0.05 |
alpha is what type of error | Type I error |
Beta is what type of error | Type II error |
what is the conventional value for alpha | 0.005 |
what is the formula for statistic power | 1-beta |
how do we calculate 95% Confidence interval (CI) | mean+/- (1.96 x Standard error) |
what can we say about two population if 95% CI does not overlap for the populations | we can be "95% certain" that 2 populations are difference. |
What can we say about 2 population if 95% CI does overlap | we are not 95% certain that the populations are different. Therefore there is no statistically significant different between the outcomes of 2 populations |
What does 95% CI tell us about a data | the interval of data that include 95% of data measure |
if value 1 fall in the 95% CI of RR, what is the conclusion? | there is no statistically significant difference in 2 population |
If value =1 does not fall in the 95% of RR, what is the conclusion about 2 population? | there is Statistically significant difference between 2 populations |
When RR =1, what does it tell us about the 2 populations? | RR=1, Risk 1 = Risk 2, no difference between 2 population |