Categorical: 
Qualitative 
 Nominal 
· Labelled ·
No quantity or order 
 Ordinal 
· Numbered ·
Variable increments 
Numerical: 
Quantitative 
 Discrete 
·
Whole numbers only 
 Continuous 
· Any value · Constant increment · Interval data: false zero point (i.e. 20°C isn’t twice as hot as 10°C) · Ratio data: true zero point (e.g. 2/52 is twice as old as 1/52) 

2 groups 
>2 groups 

Paired 
Unpaired 
Paired 
Unpaired 

Parametric 
Student T 
Student T 
ANOVA 
ANOVA 
Nonparametric Nominal 
McNemar 
Chi 
Cochrane Q 
Cochrane Q 
Nonparametric Ordinal / ratio 
Wilcoxon Rank Sum 
Mann Whitney U 
Friedman 
Kruskell Wallis 
N.B. Parametric tests are used when data a) are numerical b) follow a distribution
Mean 
· Sum / number of observations · More affected by outliers 
Median 
· Half the observations are higher, other half are lower · Less affected by outliers 
Mode 
· Most frequently occurring value · Rarely relevant 
Interquartile range 
· Middle 50% of observations · Often represented on box and whisker plot o Box: 25, 50 and 75 o Whiskers: 10 and 90 
Standard deviation 
· SD = √variance ·
Variance = ε(xẋ)^{2}/(n1), where 
Standard error 
·
SE = SD/√n, where · Indicates how far the sample mean is likely to be from the true mean · Used to derive confidence interval 
95% confidence interval 
· CI = mean ± 0.95(SD/√n) ·
Indicates both magnitude and precision of difference · If the confidence interval crosses 1.0, result is insignificant 
Example:

Pain 
No pain 
Fentanyl 
A 
B 
Nothing 
C 
D 
Relative risk 
· Risk of pain in fentanyl group = A/(A+B) = R_{F} · Risk of pain in nothing group = C/(C+D) = R_{N} · Relative risk = R_{F}/R_{N} · The risk of the event in the intervention group compared with the risk of the even in the control group” 
Odds ratio 
· Odds of pain in fentanyl group = A/B = O_{F} · Odds of pain in nothing group = C/D = O_{N} · Odds ratio = O_{F}/O_{N} · “A ratio of event to nonevent in the intervention group compared with the control group” 
Hazard ratio 
· Hazard ratio is the relative risk of an event happening at time t · i.e. risk of pain now cf. risk of pain at some point 
Number needed to treat 
· Absolute risk reduction = R_{N}  R_{F} · NNT = 1 / (absolute risk reduction) 

Screening 
Diagnostic 
Target 
Everyone 
Symptomatic; or Positive screening test 
Nature 
Noninvasive 
Invasive 
Thresholds 
High sensitivity (few false neg) 
High specificity (few false pos) 
Cost 
Cheap 
Expensive 

Difficult ETT 
Not difficult ETT 
Predictive values: 
High MP score 
a)True +ve 
b)False +ve 
PPV: a/(a+b) 
Low MP score 
c)False ve 
d)True ve 
NPV: d/(c+d) 

Sensitivity: a/(a+c) 30% 
Specificity: d/(b+d) 90% 

Likelihood ratios 
Positive LR = sensitivity / (1specificity) = 3 Negative LR = (1sensitivity) / specificity = 0.78 
Receiver operator characteristic 
· Relationship between sensitivity, specificity, test quality · ↑AUC associated with high utility

Youden’s J statistic 
· J = sensitivity + specificity  1 · J = point of maximum divergence of the curve · Single representation of a test’s utility, between 0 to 1 
P value 
· Probability of finding this result by chance if the null hypothesis were true · i.e. probability of this being a false positive result Problems: · Does not take into account prior probability · Does not quantify effect size (cf. confidence interval) · 0.05 not low enough if the stakes are high · Inappropriate for multiple comparisons (see Bonferroni correction) 
Type 1 error 
· False positive ·
Threshold (α value) usually set 5% 
Type 2 error 
· False negative ·
Threshold (β value) usually set at 20% 
Power 
· The ability to detect a difference if there is one · Power = 1 – β value = usually 80% Many determinants, e.g. · Sample size · Magnitude of difference · Threshold 
Fragility index 
· Number of patients whose change in status would turn a significant result into a nonsignificant result 
Bonferroni correction 
· Used if assessing for several outcomes simultaneously · Divide α (p = 0.05) by the number of tests being performed 
Type 
Explanation 
Mitigation 
Hawthorne effect 
Being studied = feel good 
Control group 
Selection 
Samples studied does not reflect the population 
Forced participation (i.e. no solution) 
Response 
Questions encourage a certain type of answer 
Control group 
Recall 
Patient reports symptoms differently depending upon allocation 
Patient blinding 
Detection 
Intervention and control group assessed differently 
Assessor blinding 
Observer 
Data collector able to be subjective about the outcome 
Assessor blinding Hard outcomes 
Publication 
Negative results don’t get published 
Clinical trial registries 
Feedback welcome at ketaminenightmares@gmail.com