qs Flashcards
When do you not reference?
- When it is considered common knowledge
- Where it is your own opinion
- When it is your own data / research
How many authors should be listed as ‘et al’
6/7
What is stratified sampling?
- Sample set into groups
- Then chosen at random from these groups
- Ensures all populations make up the sample and are represented
What is quota sampling?
- Sample taken from a stratified population until a pre-assigned quota in each stratum is represented
- Not random
When is quota sampling useful?
- Quick, cheap
- When detail isn’t important
what is cluster sampling
- simple random sampling within each cluster
- used when ‘natural’ but relatively homogenous grouping are evident in a population (e.g., regions in the UK)
what is categorical data
- normal or ordinal
- distinct categories
- do not have a numerical value
how can ordinal data be made numerical?
assigning a rank scale value to each category
ordinal vs nominal data
nominal: no natural order e.g., sex
ordinal: have natural order e.g., poor, good, excellent
what type of data would the number of cases of MRSA on a particular ward be classed as?
numerical discrete
what type of data is age?
numerical continuous
what is relative frequency
the % of the total frequency accounted for by particular variables
what does a low s.d indicate?
the data points lie close to the mean
how is incidence rate calculated
number of new cases occurring in a set period of time / number of people at risk in this time (e.g., exposed people)
what is included in the epidemiological triad
time
place
person
diseases with short duration incidence and prevalence rates
high incidence
low prevalence
how is mortality rate calculated
number of people who die from the disease / number of people who die from all causes in a set period of time
how is data adjusted for (e.g., for age, gender)
Stratum specific rates - population split into stratum (age,gender)
Incidence calculated in each, weighting then assigned to each group & data adjusted accordingly
how is case fatality calculated
No. Of people who die for a disease In period / no. Of people with the disease
how are odds ratios calculated
Odds for disease of exposed group / odds of disease of unexposed group
what are dichotomous variables
The variable takes 1 of 2 forms (that are co merely different) e.g. Dead or alive
what type of data are histograms used for?
numerical continuous
what is the use of IQR
Able to form a graphical representation of probability distribution (box plot)
In normally distributed data what measure would you use to look at spread of data ?
mean and standard deviation
what measure of distribution should be used in non normally distributed data
median and IQR (so that it is not affected by outliers) plotted on a histogram
what does a 95% confidence interval mean?
If the experiment were repeated 100 times the results would fall into that range 95 times
what is standard error a measure of?
How reliable the sample population mean is from that of the population, so measures precision
The CI should not cross the value where the variable is said to have no effect. Where are the points of no impact in ratio and difference
ratio = 1 difference = 0
When is correlation the appropriate statistical analysis
When both the outcome and exposure are numerical
If the outcome is numerical, and the exposure is categorical (independent groups) what is the appropriate statistical analysis?
T test
A hypothesis proposed after analysis is which type of hypothesis?
Inductive hypothesis - explanation as why two sets of info are related to one another
What is a deductive hypothesis?
A hypothesis proposed before analysis - predict what the relationship will be. You think that at least 2 variables are related to each other
If data is positively skewed, which way will its ‘tail’ be on a box plot?
tail to the right
When refining a search, what function does adding an asterisk (*) to the end of a word have?
Denotes a truncated ending where alternative endings would be accepted
e.g. Teen* would show up teenage, teenagers, teenager, teens
What are Boolean operators ? Give examples
Words used to refine searches, e.g. AND, OR, NOT
what type of study is a census
Cross sectional - looks at entire population at defined time assessing prevalence
what is responsiveness
Whether a measure can detect real change (over time?)
what is regression analysis
Statistical process for estimating the relationships among variables
What is a type 1 error ?
The incorrect rejection of a true NULL hypothesis- false positive for alternative hypothesis (too optimistic)
What is a type 2 error ?
Failure to reject a false NULL hypothesis- false negative for alternative hypothesi s(Too pessimistic)
What is chi squared used for ?
Determine association between categorical variables
How is standard error calculated ?
Standard deviation/square root of sample size
What happens after you propose a hypothesis?
Test it - then reject and modify or not reject
Define population
Every member of a defined group of interest
What is the only true random sample?
Simple random sample - population members are chosen purely a random
What is interval sampling?
When you take samples at set intervals
e.g drugs company take a sample of every 100th drug produced to check that it’s being made correctly
What type of variable is the number of a beds on a ward?
Numerical discrete
What type of variable is weight or height?
numerical continuous
For what type of data would you use a bar chart?
Categorical and discrete metric variables
For what type of data would you use a pie chart?
Categorical, may be used for numerical data
What is cumulative frequency?
the running total
How do you calculate frequency density?
Freq density = frequency / class width
examples of measures of spread
range, IQR, s.d
examples of measures of central tendency
mean, median, mode
Negative skew would result in the greatest spread in what quartile on a box plot?
lower quartile
Which type of skew would give these results; Q1-Q2 > Q2-Q3?
negative skew
What phase of a clinical trial assesses effectiveness and dosage of drugs on a few hundred patients WITH the disease?
phase 2
What is a consensus study?
A consensus statement developed by professionals via a group consensus process that is intended to advance health professional and/or public understanding of a targeted health problem.
In what type of paper/article/study would you write a short paper discussing any of these 4 things; updating readers on new methods, providing an overview of the topics in an issue, press opposing opinions or clarifying positions?
editorial
What do p-values mean?
P = probability of obtaining the results of the test given that the null hypothesis is true
How do you calculate prevalence?
= Number of people with a disease at a certain time / number of people in the population that time
How to calculate risk?
Risk = number of new cases / number at risk
In what type of study do you use a risk ratio?
cohort - use to find risk factors of a disease
In what type of study would you use odds ratio?
case-control study
What is relative risk reduction?
RRR = 100 - RR. Essentially its the difference the new treatment makes to the condition (compared to old/untreated group)
how to calculate the odds of an event
odds = probability of an event (disease) / probability of the event NOT occurring
how to calculate odds ratio
OR = odds of disease for exposed group / odds for disease in unexposed group.= probability of disease occurring in exposed group / probability of disease NOT occurring in unexposed group
Do larger or smaller studies have a larger CI?
Smaller studies = large CI
Large studies = small CI (larger studies reduce spread of data and inc the accuracy of the result in relation to population)
Which type of health prevention is this; prevention of disease by treating clinical cases, reducing disability and maximising function?
Tertiary prevention
define reliabliity
gives the same result on retesting
r = -1 then what is the correlation?
perfect negative correlation
‘r’ is what coefficient?
Pearson correlation coefficient. Values lie between -1 & 1
If T statistic in T test is more than critical value = reject or accept the null hypothesis?
Reject the null hypothesis
What term described this definition ‘can detect real changes when they occur’?
responsiveness
What three key things should health outcomes be?
valid
reliable
responsive
Type of efficacy described as ‘production that matches the consumer demand’?
allocative efficacy
Term used to described this statement ‘concerned with the fairness or justice of the distributions of costs and benefits’
Equity - another criterion for allocating resources, who benefits may matter to society.
Type of efficacy described as ‘obtaining maximum output from a set of given resources’?
Technical efficacy - is concerned with how best to deliver a programme, or achieve a given objective
When do you not need ethical approval?
audits / evaluations
Benefits of knowing health outcomes?
- identify treats and procedures that work and which are less effective
- identify national and international variation
- monitor performance against targets and over time
covariates
confounders or competing exposures
what are confounders
A covariate that precedes both the ‘exposure’ and the ‘outcome’
what are competing exposures
covariates that compete with the exposure for the outcome
what is a mediator
a variable which comes after the exposure but before the outcome
what are clinical guidelines and quality standards
Clinical guidelines constitute recommendations for strengthening practice, while quality standards are benchmarks for auditing ‘best practice’
