statistics and data analysis

Question 1

what is a nominal variable

Answer 1

data in the form of labels or names
but without a natural order or ranking
e.g. eye colour
it is a type of categorical variable

Question 2

what is an ordinal variable
give an example

Answer 2

a named variable
which has a natural order e.g. ‘level of satisfaction’ or ‘degree of pain’ or ‘socioeconomic status’ or ‘workplace status’

Question 3

what is an interval variable
give an example

Answer 3

named variable
with a natural order
AND equal intervals between variables
e.g. temperature, credit scores, SAT scores

but NOT a ‘true zero’ value (that’s a ‘ratio’ variable)
for temperature, you can have negative degrees celcius, so there’s no zero that the value cannot go beneath
you cannot have a credit score of negative, or of zero

Question 4

what is an ratio variable
give an example

Answer 4

named variable
with a natural order
AND equal intervals between variables
AND a ‘true zero’ value
e.g. height, weight, length

e.g. your height could technically be zero but NOT negative
equally with weight

Question 5

how to calculate standard error of the mean

Answer 5

SD / square root of sample size

Question 6

how to calculate BMI

Answer 6

kg / (m^2)
weight / height in m / height in m

Question 7

obesity class 3 is

Answer 7

40+

Question 8

obesity class 2 is

Answer 8

35 - 39.9

Question 9

obesity class 1 is

Answer 9

30 - 34.9

Question 10

overweight is BMI

Answer 10

25 - 29.9

Question 11

Normal BMI is

Answer 11

18.5 - 24.9

Question 12

underweight BMI is

Answer 12

<18.5

Question 13

how to calculate odds ratio

Answer 13

Divide the probability that the event will occur by the probability that it will not occur
In other words, it’s a ratio of successes (or wins) to losses (or failures)

Odds ratio = p / (1 - p)

Question 14

how to calculate standard deviation

Answer 14

square root of the variance

Question 15

what does a low p-value mean

Answer 15

there is strong evidence to reject the null hypothesis
i.e. it’s likely that your results are due to a true effect and not just due to chance

so a low p-value is usually exciting for your experiement :)

Question 16

what is a t-test

Answer 16

assesses whether the means of two groups are statistically different from each other

Question 17

if you are comparing 2 different diets (for example), which test would you use?
if you are comparing 3 diets, which test would you use?

Answer 17

2 - two sample t-test
3 - ANOVA

Question 18

What does a p value of 0.05 mean?

Answer 18

The p value is a proportion: if your p value is 0.05, that means that 5% of the time you would see a test statistic at least as extreme as the one you found if the null hypothesis was true.

Question 19

what is a Forest plot?

Answer 19

the way you represent data in a meta-analysis

Question 20

Forest plots vertical line is…
what value is the line at?

Answer 20

the line of no effect
if a RR or OR, the line is at 1.0
for AR or SEM, the line is at 0

i.e. no difference between 2 interventions
or no association between an exposure or an outcome

Question 21

Forest plot horizontal line at each point is…

Answer 21

the confidence interval for that study in the meta-analysis
(a Wider confidence interval is Worse.
Or a Longer line is Less reliable data)

Question 22

on a Forest plot, a bigger box means…

Answer 22

a larger sample size
as these are more weighted, the confidence interval tends to be narrower too.

Question 23

the diamond-shaped box at the bottom of the Forest plot is…
the lateral tips of the diamond represent…

Answer 23

the weighted average of all the studies

the lateral tips of the diamond are the confidence interval of the weighted average of all the studies

Question 24

If the diamond at the bottom of a Forest plot touches the vertical line, that means…

Answer 24

the results of the meta-analysis show the thing being studies is not statistically significant

Question 25

If the diamond at the bottom of a Forest plot is to the left/right of the no effect line, what does it mean for each side (left vs right)?

Answer 25

LEFT of the line - there are fewer episodes of interest in the treatment group RIGHT of the line - there are MORE episodes of interest in the treatment group

Question 26

when do you used a chi squared test? What is the most commonly used chi2 test?

Answer 26

Pearson's chi2 test. Testing a hypothesis regarding a relationship between 2 CATEGORICAL variables with no rankings

Question 27

what is a t-test

Answer 27

To formally test whether or not there is a statistically significant difference between two POPULATION MEANS. e.g. We want to know if diet A or diet B leads to greater weight loss

Question 28

what does a likelihood ratio actually mean?

Answer 28

assess the utility of a diagnostic test. And how likely it is that a patient has a condition. LR+ is about positive test results LR- is about negative test results You can use LR to calculate post-test odds, meaning how suspicious you are that a person has a disease based on the result.

Question 29

post-test odds =

Answer 29

post-test odds = pre-test odds x LR

Question 30

difference between odds and probability in an example where n=10, diseased = 4, and non-diseased = 6

Answer 30

p(disease) = FRACTION of the total = 40% odds of disease = RATIO of sthg happening vs not happening = 4/6 = 67%

Question 31

as a rule of thumb, according to a video I watched, what was a strong LR+

Answer 31

>5

Question 32

difference between paired and unpaired t-test

Answer 32

Paired t-test: Used to compare the means of two samples when each individual in one sample also appears in the other sample. Unpaired t-test: Used to compare the means of two samples when each individual in one sample is independent of every individual in the other sample.

Question 33

define 'incidence'

Answer 33

The incidence can be thought of as the number of new cases occurring in a given time. The incidence equals the number of newly affected individuals divided by the number of people at risk for the disease for a given duration

Question 34

how to calculate prevalence

Answer 34

Prevalence equals the total number of cases divided by the total number of at risk (differs from the incidence in that incidence is NEWLY affected individuals and over a certain period of time, e.g. a year)

Question 35

Suppose we have two choices of shirt to wear at a party then the degree of freedom is... Now suppose we have to again go to the party and we can not repeat the shirt then the choice of shirt we are left with is One then in this case the degree of freedom is...

Answer 35

1 = there are 2 choices, so there's 1 degree of freedom for the second party, dF = 0, as we do not have any choice on which shirt to choose. in a chi2 test, dF = (number of columns -1) x (number of rows -1)

Question 36

A study has been designed to investigate whether a certain drug plus physiotherapy treatment is better than drug treatment alone in the management of rheumatoid arthritis. After randomising the patients a small proportion of the drug plus physiotherapy group decide to drop out of the study or omit some treatment sessions specified in the research protocol. What is the correct way of analysing the subsequent data? A. Assume the patients have withdrawn their consent B. Exclude these patients from all analysis C. Extend the trial recruitment to make up the numbers D. Include these patient outcomes in the drug plus physiotherapy group E. Interview the patients and report their group separately

Answer 36

D is correct - include them in the data This is the principle of 'intention to treat'. Intention to treat helps to reduce bias by sticking to the original allocation of treatment and analysing the patient in that treatment group even (and concentrate for this bit), even if they don't get the treatment! e.g. it is possible that the physiotherapy intervention was harmful to the patients and this is why they left.

Question 37

false negative calculation

Answer 37

Out of everyone who has the disease, how many tested negative so it's almost like calculating specificity, but instead of true positives on the top you do false negatives that's hard to remember, though, as because it's talking about the TEST result, I thought I needed to put a denominator of everyone who tested negative (but no, that's for negative PREDICTIVE value)

Question 38

is a type 1 error a false positive or a false negative

Answer 38

false positive = type 1 false negative = type 2

Question 39

formula for calculating first quartile

Answer 39

Q1 = (n+1)/4 - th value so if n = 11, then the data point for Q1 is the third number (when put in ascending order) Q2 = (n+1)/2 - th value Q3 = 3(n+1)/4 - th value