Quantitive analysis Flashcards
why is statistics important?
- way to describe and measure biology
- allows us to estimate an unknown quantity using our sample
- test hypotheses
is stats math?
kinda? helps us to define what we know, how we know it and how strong the evidence is
what do we use descriptive stats for
summarize and describe the data
describe patterns
tables, graphs and charts
- the first step of making inferences
what are the 4 types of descriptions of data and their two categories
Categorical
1. nominal
2. ordinal
continuous
1. interval
2. ratio
What is nominal data?
- at type of categorical data
1. non - ordered categories - sex
- blood type etc
what is ordinal data ?
type of categorical data
- ordered categories
eg: house hold income - lowest to highest
eg: age
how do we display nominal data?
- frequency tables
- bar graphs
how do we display/describe continuous data?
- frequency table
- histogram
- parameters - mean, median etc
what is a frequency table?
Create artificial groups and report the frequency
* e.g. age
* This is not very informative and throws out information
what does a histogram show?
the shape of the data,
how to describe continuous variables?
- when measuring central tendency
a)mean
b) media
c)mode - measuring of variability or spread
a)range
b)percentiles
c) variance/standard deviation
what is a central tendency?
the average distribution.continuous variable
summarizes the data with a single value
what can the mean be distorted by?
outliers
what is the median?
middle of data
order from smallest to largest and average two middle points
what is the mode?
number that occurs most often in the data
can be more than 1
not impacted by outliers
when to use mode, median and mean
mode - rare
mean - if your sample is normal - bell
median - if your sample is skewed
what is the range?
smallest to largest values
what are the percentiles? -hard
- a measure of the dispersion for one variable that indicates the % of cases at or below a score or point
eg: the median is the 50th percentile
eg: 25th percentile is the value at which 25% of the distribution have that value or a lower one
often report quartiles - 25th, 50th, 75th
what is standard deviation/variance?
- the measure of dispersion for one variable that indicates an “average distance” between the values and the mean
what is generally used to show continuous data visually?
boxplots
What do boxplots show? - parts of it
- the median
- the IQR - intervals q1 q2 etc
- the range or the outliers
if the median is the centre of a boxplot what does it mean
that you have a bell-shaped curve
what can boxplot diagrams be used for?
- to compare different groups
what are the types of prevalence?
- point prev
- period prev
- lifetime prev
what is incidence density/ rate
the number of new cases in a given time period
divided by
total person-time of follow-up (sum)
USed for longitudinal cohort studies
- each person contributes time based upon how long they were followed -
what happens when individuals are censored in data ?
we exclude any data on them that was gathered after the censoring point, if any
what is the unit for incidence density
generally per 100 person-years because you times the number you get by 100
what does mortality rate show?
- how many people die from a disease in a given time
what does case fatality rate mean
CFR = the proportion of people WITH the disease who die within a given time, again usually one year
what is a issue with MR’s and CFR’s? `
we do not know if the number is high or low except if comparing to other disorders
- do not always know how many people have the disorder
One way around this issue is to look at death rates for a
specific disorder in relation to deaths from all causes…
What is the proportional Mortality Rate? PMR
the number of deaths due to a disorder in a given time
divided by ………..
number of deaths from all disorders
- makes sense only when we compare the PMR from one disorder to another disorders PMR
How can the PMR for a disease increase over time? 2 reasons
- more people may be contracting and dying from the disorder
- fewer people may be dying from other causes
What is the Standardized Mortality Rate?
the ratio of the number of observed deaths from a given cause in a given time to the expected number of deaths
- this measure does not have to be compared to others
how do we interpret SMRs? Standardized mortality rates
- an SMR less than 1 = people with the disorder are dying lower than the expected rate
- an SMR more than 1 = higher than the expected rate
when does using the general population not make sense as a reference group?
when comparing groups with different age structures
- need to use - age-standardized mortality rates
what do measures of impact find? m
how dangerous the disease is
- relative risk or odds
what is the relative risk
the ratio of…
1. the risk of death in the exposed group - a divided by a+b
2. the risk of death in the unexposed group - c divided by c+d
relative risk is the ratio of these two risks
exposed risk divided by unexposed risk
what are the limitations of relative risk?
the relative risk is the same regardless of prevalence of the disorder
what is relative risk reduction RRR
improvement as a percent relative to the comparison group
risk (treatment) - risk (control)
divided by…………
risk (control)
Absolute risk reduction ARR
- decrease in the risk of the treatment group compared to the control group
risk treatment - risk control
What is the number needed to treat? NNT
the number of people who need to be treated to have one additional positive outcome NNT = 1/0.1 = 10
eg: need to treat 10 people for one cure
What is the odds ratio?
this is used in case-control studies to assess the relationship between the exposure and the disease
- the ratio of the odds of exposure among the cases compared to the odds of exposure amount the controls
what does PICO stand for?
population
intervention
comparison
outcome