Descriptive Statistics

null hypothesis

-assertion that there is no relationship between exposure and disease

alternative hypothesis

-assertion that there is some relation between exposure and disease

significance testing

-play of chance- must always be considered alternative explanation

-tells likelihood of finding results because of play of chance

p value

-all tests of statistical significance lead to probability or p value

-usually set at 0.05

-if probability = or < 0.05, then you can safely reject the null hypothesis

-if p value > 0.05, chance cannot be excluded as likely explanation and the null is not rejected, state findings not statistically significant

alpha, type I error

-false positive

beta, type II error

-false negative

-most common error in dental literature

-1- beta (power) = probability of correctly concluding that groups differ

statistical power

-power = 1-beta

-higher the power, better the chances of finding the tx benefit if there is one

-increase number of subjects, increase power

-influenced by: determining factors (size of tx, alpha error (0.05), beta error (0.2)), interrelationship of alpha/beta error, impact of sample size (overpowered for large sample size, underpowered for small sample size), meta-analysis (increase statistical power)

incidence

-number of new cases of specific disease occurring in a defined population during a specified time period

-gives clues to the research into etiology and pathogenesis of disease

-helps with study of distribution of disease and taking action to control disease

-useful in evaluating the efficacy of preventative and therapeutic measures

prevalence

-how common is some factor at a given time

-typically measured in cross-sectional sample

-useful in estimating the magnitude of disease or health problems in a community

-identify potential high-risk populations

-useful in administration and planning purposes

-prevalence = incidence * mean duration

positive skew, negative skew, normal curve

standard deviation

-measure of variability of sample

standard error

-variability in the mean within the whole population

-SE = SD/square root of N

-tighter range than SD

dependent v independent variable

-dependent- outcome variable

-independent- factor manipulated in experiment, cause

types of variables

-categorical/nominal: non-ranked, dichotomous (binary)

-ordinal: have directionality

-numerical/continuous: have directionality, values have numeric sense