Use LEFT and RIGHT arrow keys to navigate between flashcards;
Use UP and DOWN arrow keys to flip the card;
H to show hint;
A reads text to speech;
36 Cards in this Set
- Front
- Back
Incidence & prevalence
|
Incidence: new cases in a population during the time period of a study. Incidence rate is the # of new diagnoses over the TOTAL population
Prevalence: # of people with a disease at a given point in time |
|
Rate
|
# of events that occur during a specific time period / average # of people at risk for the event during the time period of study
|
|
Three types of rates
|
Crude: # of deaths in entire population
Specific: # of deaths in specific group Adjusted: crude rates adjusted to control for other variables |
|
Type I vs. type II error
|
Type 1: alpha error, false positive
Reject the null when it is true Type 2: beta error, false negative Accept the null when it is indeed true |
|
What is the alpha value?
|
Associated to confidence level of the test, indicates the probability that a type I error will occur
Compare the alpha value with the p-value. p-value < alpha value --> reject the null p-value > alpha value --> accept the null (fail to reject it) |
|
What does a p-value represent?
|
Probability that the result is based on chance alone
|
|
What are sensitivity and specificity?
|
Sensitivity: ability of test to detect disease when disease is present. The higher sensitivity, less false negative.
Specificity: ability of test to detect no disease when disease is not present. The higher the specificity, the less false positive. |
|
What are the types of sampling?
|
Random: each member in population has same chance of being chosen
Stratified: break up population into groups, then select from those groups Systemic: picked via system, ss#, etc Convenience: use volunteers or readily available subjects |
|
What is bias?
|
Differential error that weakens true associations, produces false associations, or distorts direction of associations
Occurs when a nonrandom allocation is used for study |
|
What are validity and reliability?
|
Validity: accuracy of scale to measure what it's supposed to
Reliability: repeatability |
|
The 5 studies classified by data type are:
and their data includes: |
Cross-sectional: all data comes from 1 point in time
Good to see associations, but not causality Retrospective: data is collected from the past Observational study: no changes by experimenter Experiemental: randomly divide sampe into experimental and control groups |
|
Summarize a case study
|
Reports on 1 person with novel or rare disease
Observational data |
|
Summarize a cross-sectional survey
|
Survey of a sample at a given point in time via interviewers
Quick, easy, but data is only from 1 time Cross-sectional data |
|
Summarize a cohort study
|
Cohort is a clearly defined group of persons, usually have a common factor
Retrospective cohort: recruit study group, get baseline data, continue to follow Prospective cohort: use historical data to get group, follow members up through present to see what outcomes occur |
|
Case-control studies
|
Subjects placed in groups whether or not they have a disease. Used to study rare diseases with many risk factors at once.
Compare cases vs. non-cases or controls Nested case-control: a case-control study nested in a cohort study. Take a cohort, and choose cases vs. non-cases within the group |
|
Clinical trial
|
Gold standard
Tests treatment vs. placebo Double blind: participants and administrators don't know any info on the study to prevent bias Randomized Controlled: intervention group gest tx, control group gets placebo |
|
Positive and negative predictive values
|
Positive: proportion of people with a positive test result that have the disease
Negative: proportion of people with negative test results that do not have disease |
|
Negative or positive confounder
|
Negative: weakens or masks a true association
Positive: makes an association that doesn't really exist |
|
What is effect modification or interaction
|
Direction OR strength of 2 variables differs according to a third variable
Females an coffee |
|
What are the two main types of data?
|
Discrete/categorical
Continuous You can take continuous data and convert it to discrete or categorical data |
|
What are the scales of measurement? NOIR
|
Nominal - categories in no order, pie chart
Ordinal - ordered categories with unequal intervals btw, bar graph Interval - equal intervals btw data points with no true zero, histogram or frequency polygon Ratio - equal intervals btw data points with true zero, histogram or frequency polygon |
|
Descriptive statistics do what?
|
Describe a sample
|
|
What are measures of central tendency?
|
Mean - average
Median - middle # Mode - most commonly occurring number |
|
What are measures of dispersion?
|
Distribution - bell-shaped curve that can be skewed
Skewed right - tail falls right, mound falls left. Large outliers pull the mean to the right Skewed left - tail falls left, mound falls right. Large outliers pull the mean to the left |
|
What is the equation for variance?
|
Look at it
|
|
What is variance?
|
How far the data is spread out
|
|
What is SD?
|
Average deviation from the mean, regardless of direction
|
|
T-test compare what?
|
Two groups or two sets of data. Compares the variances and SD's with p-value to determine if it's significant
Paired t-test will compare 2 groups with same characteristic, or 1 group with 2 scores or data points Independent t-test will compare 2 independent groups on differences of the man value of a variable |
|
Assumptions of t-test include
|
Continuous data on interval or ratio scale
Normal distribution of data Variances are equal unless sample size is large |
|
What is ANOVA/ANCOVA?
|
ANOVA: compares mean values for MORE than 2 groups at the same time. Tests variance in groups to determine if the null hypothesis or alternate hypothesis is true.
Tells that there's a diff, but NOT where the diff lies via f-statistic and p-value ANCOVA = ANOVA minus cofounders |
|
What is regression?
|
Prediction of scores on Y from scores on X
Linear - variables have linear association |
|
How do odds ratios relate to risk?
|
OR = 1, no difference
OR < 1, less risk OR > 1, increased risk If confidence interval is above 1 for both range, it's good |
|
What is logistic regression?
|
Tests 2 variables with an outcome that is dichotomous
|
|
What is chi-square test?
|
Tests 2 groups frequency or rate observed to see if there's statistically significant diff
|
|
What does OR show
|
Risk or odds of a particular outcome
|
|
Parametric vs. non-parametric statistics
|
Non-parametric are used when the data is nominal or ordinal, categorical outcome
|