RESS - Ebook Flashcards
What type of data do bar charts use?
Categorical (ordinal and nominal) and Numerical discrete
What type of data do pie charts use?
Categorical (ordinal and nominal)
What type of data do histograms use?
Numerical Continuous
In a histogram what does the area of the bar and the height of the bar mean?
Area: Frequency
Height: Frequency/Class width
What is standard deviation?
Average distance of all the points from the MEAN!
Larger the SD- bigger distribution of data
With what type of data is SD used
Normally distributed
What is IQR?
Q1 - Median - Q3
Between Q1 and Q3 is the IQR which is where 50% of the distribution lies
With what type of data is IQR used?
Skewed data
Define Incidence
Incidence is the number of new cases over a specific time period
Number of new cases/ Number at risk
Define Prevalence
Prevalence is the number of cases of disease at a particular time point
Number of people with a disease/ number of people in a population
Define Mortality
Mortality is the number of people dying from a disease
Number of people who die from a disease/Number of people who die in general
Define Case Fatality
number dying in period from disease / number with the disease in period
What is risk?
Synonymous with probability
Number of new cases/ Number at risk
What is a risk ratio?
Shows relative risk in an exposed and unexposed group
When is risk ratio used?
Cohort studies to show relative risk!
What is odds?
Number of time an event occurs / Number of time it doesn’t occur
What is odds ratio?
Shows relative risk
When is odds ratio used?
Case control studies and RCT to show relative risk!
What two things can determine relative risk?
Odds ratio (RCT and case control) and Risk ratio (cohort studies)
What does a RR of 1 show?
means risk in exposed = risk in unexposed so no benefit or harm
What does RR <1 show?
means risk in exposed < risk in unexposed exposure is protective
What does RR >1 show?
means risk in exposed > risk in unexposed exposure is harmful
What does Gaussian distribution show?
Normal distribution in the shape of a bell curve
What are features of normal distribution?
Values symmetric around the mean
Greatest frequency is at the mean
Normal distribution is symmetric around the mean median and mode
What are features of a normal distribution curve?
SD is how broad the bell is
Area under the curve is 1
Mean = 0
SD= 1
How do you make probability statements about a variable using a distribution curve?
Transform the distribution curve to a standard normal distribution curve
What is standard error?
Used to calculate confidence intervals of the mean (Standard deviation needs to be known to calculate standard error)
(Is a measure of the precision of the sample mean)
The larger the sample size the lower the standard error
What is a confidence interval
A range of feasible values that the true (mean) value is believed to be within (usually 95% probability)
What does it mean if a confidence interval crosses 0?
If a confidence interval crosses this is not statistically significant as the effect of the variable could be 0
How is the SE and thus confidence interval calculated if you have known SD?
Known SD if >200 samples
Confidence interval calculated using critical values from standard normal distribution
How is the SE and thus confidence interval calculated if you have estimated SD?
Estimated if sample size <200
Confidence interval calculated using students t distribution
What has a wider range 90% CI or 95%
95% (as to be 95% sure the ranges will be wider)
What is a null hypothesis?
Opposite to hypothesis; assumes nothing happens
What is an alternative hypothesis?
Has two possible outcomes e.g. drug will decrease hypertension or increase hypertension
What assumptions are necessary for the test statistic?
1 - The variable; change in bp has to be normally distributed
2- The samples are independence i.e. from different patients
What is the p value?
The probability that the null hypothesis is true
Lower the P value the less likely the null hypothesis is true
When is the p value statistically significant?
When it is <0.005
Are confidence intervals adequate for hypothesis testing?
Yes
What is parsons correlation coefficient?
r
Measure of linear correlation between two numerical values
R values =0, >0, <0, 1 and - 1
=0 Linear relationship >0 Positive correlation <0 Negative correlation 1 perfect correlation -1 perfect negative correlation
What value of correlation is treated with suspicion
<0.7
When should pearsons coefficient not be used?
No linear relationship
Outliers
Subgroups i.e. not numeric
One or both variables are not normally distributed
When pearsons coefficient can’t be used what is used instead?
Speerman rank correlation
When is superman rank correlation used?
Ordinal data
Small sample size
Data not normally distributed
In linear regression analysis what is on the X and Y axis
X axis: independent variable
Y axis: Dependent variable
What requirements are needed for regression analysis?
Linear relationship
Normally distributed
What is the chi squared test?
Association between two categorical variables
What does the chi squared test compare?
Contingency tables from; results observed and results expected if null hypothesis was true
What conditions are required fro chi squared test?
4 cells should be more than 1
Three of these should be more than 5
If conditions are not met for a chi squared test what else can be used?
Fishers exact test
Sensitivity
How well a test detects a condition
Specificity
How well a test correctly excludes those without the condition
Positive predictive value
Probability someone has the condition if they test positive
Negative predictive value
Probability someone doesn’t have the condition if they test negative
Failure Event
The time a person leaves a study is known
Censory
Leaving a study before the event occurs or enters late
Left Censory
Disease present before the entered (not sure what happened before they entered)
Right censory
People in the study did not reach a failure before the end (i.e for HIV drug an individual left country or died from non HIV related cause)
Survival Function
Chance of survival until a certain time
Hazard Function
Chance of an instantaneous failure
Kapplan meir plot
Shows survival as a series of steps
It steps down every time a failure is observed
