Stats / Data Flashcards
Whats an ogive
A continous cumulative frequency curve whereby the central point had 50% of values above it and 50% of values below it
i.e. the interquartile range lies between the first and third quartiles
Types of data from a sample
Qualitative
Quantitative
Quantitive data can be of two types
Numerical
Categorical
Numerical data can be either
Discrete - distinct values within a range
Continuous - any values within a range
Categorical data has three subcategories
Binary (two levels only)
Ordinal (more than two levels with a specific order i.e. low medium high)
Nominal (more than two levels with no order e.g. ethnicity)
What is the Chi square test
A statistical test that can be used to determine wether observed frequencies are significantly different from expected frequencies
Outcome allows to accept or reject the null hypothesis
Used to compare discontinous categorical data
When is a binominal test used
To see if observed test results differ from what is expected
Used when there are two possible outcomes e.g. success or failure and you have to estimate what the probability of success is
When is the paired t-test used
when a group is measured twice i.e. each individual has two repeated measures
Whats an unpaired t test for
looks at the differences between two groups on a variable of interest
a normal distribution and normal standard deviation are assumed
What is regression analysis
Generates an equation to describe the statistical relationship between one or more predictor variables and a response variable
Used to determine hwo one variable correlates with another
When is a box and whisker plot used
Depicts a range of values including a sample minimum, lower quartile, medium, upper quartile and sample maximum
Box extends from border of the lower quartile to the border of the upper quartile, with the median depicted as a horizontal line across it
How to work out the relative risk?
Incidence of the disease in the exposed group DIVIDED BY incidence of the disease in the NON exposed group
What has the biggest impact on interpretation of meta analyses and why
Publication bias
There is a tendency to only submit positive trials and for journals to favour publication of positive studies
This can lead to an over estimate of effect size of a particular intervention vs a single, large, well conducted study
How to do a calculation of number needed to treat
Big number take away small number
Divide into 100
How to calculate sensitivity
Dividing the true positives by the true positives plus the false negatives
When is the null hypothesis rejected
When p value is smaller than or equal to the significance level
What is the power of a study
The probability of correctly rejecting the null hypothesis when it is false
What is used to demonstrate publication bias in meta analysis
Funnel plot
How to work out the standard error of the mean
Standard deviation DIVIDED BY Square root (number of patients)
How to calculate specificity
Number of true POSITIVES DIVIDED BY (Number of true positives PLUS Number of false negatives)
What is a cohort study
Two or more selected acording to exposure to a particular agent
Followed up to see how many develop disease or other outcome
Usual outcome measure is relative risk
What is a case control study
Patients with a particular condition are identified and matched with controls
Data then collected on past exposure to possible causative agent for condition
What is a cross sectional survey
Provide a snapshot
sometimes called prevalnce studies
Provide weak evidence of cause and effect
Relative -
Absaloute -