Stats Flashcards
% Annual preterm births in US
10%
(7.3% late (34+), 2.8% early)
% women raped in lifetime in US
1 in 5 (19%)
type I error
find there is a difference when there really isn’t
false positive
- reject the null incorrectly
- typically use .05 (or .01)
type II error
find no difference when there really is
false negative
power associated with what kind of error
type II error
- sample size
what affects power of a study
sample size
significance level desired
standard deviation
effect size
symbol for type 1 error
alpha
symbol for type 2 error
beta
symbol for power
1 - beta
% under the curve for normal bell for 1 standard deviation
68%
% under the curve for normal bell for
2 standard deviation
95%
% under the curve for normal bell for 3 standard deviation
99%
in normal distribution curve, mean, median, and mode are
all the same
what does confidence interval mean
there is that % of the samples that fall somewhere within the range
what is the null hypothesis
there is no difference
- you want to disprove this
what is the alternative hypothesis
what you want to prove
what is sensitivity
likelihood that if you have the disease, the test is positive
(if you have disease, we’ll pick it up)
what is specificity
likelihood that if you don’t have the disease, you’ll test negative
(if you don’t have disease, it won’t pick it up)
NPV
of those who test negative, how many don’t have the disease
PPV
of those who test positive, how many have the disease
false positive RATE
1 - specificity
false negative RATE
1 - sensitivity
what values depend on disease prevalence
npv & ppv
T-test used to compare
means of two samples of normally distributed data
Anova used to compare
means of more than two samples of normally distributed data
Chi-square used to compare
rates, frequencies, or proportions
Mann-Whitney U Test used for what kind of data
not normally distributed data
what is power
ability to find a difference if one exists
confidence interval calculation
mean +/- 2 standard error of the mean
(SEM is standard devation / sqr rt sample size)
overall lifetime risk ovarian cancer
1 in 70 (1.7%) for no additional risks
what % ovarian CA have BRCA defects
20-25%
BRCA1 breast cancer risk
65-74%
BRCA1 ovarian cancer risk
39-58%
BRCA2 breast cancer risk
65-74%
BRCA2 ovarian cancer risk
13-29%
Lynch syndrome (HNPCC) cancer risks
colon cancer 70%
endometrial cancer 13-57%
ovarian cancer 1-38%
epithelial ovarian CA 5y survival
Stage I: 76%-93%
Stage II: 60%-74%
Stage III
IIIA: 41%
IIIB: 25%
IIIC: 20%
Stage IV: 11%
endometrial hyperplasia risk of progression to cancer
penny nickle dime quarter
1% simple without atypia
3% complex without atypia
8% simple with atypia
29% complex with atypia
leiomyosarcoma survival rate
40% at 5y (poor despite early stages)
Incidence of HPV
75-80% of sexually active people will acquire HPV before age 50
difference between survey and questionnaire
questionnaire is a specific tool known as an instrument, relating to construct of interest, usually gives you a score
validity
extent to which instrument measures what it is supposed to measure
- aka accuracy