Term 3 math charts

Table of Contents
Formulas.........................................................................................................................................................................................4
Statistical tests................................................................................................................................................................................6
Choosing main Statistical tests........................................................................................................................................................7
Choice of appropriate statistical significance test............................................................................................................................7
Data Relationships..........................................................................................................................................................................9
Interpret Correlation Coefficient......................................................................................................................................................9
H0 VS Ha/1.........................................................................................................................................................................................9
Type 1 vs type 2.............................................................................................................................................................................10
Clinical significance VS statistically significant...............................................................................................................................10
Definitions

z-value Think SD of normal

curve
p-value
α Level of statistical Think SD of normal
significantce curve
probability you will
reject the null if its
true
Critical region / Above critical value
rejection region
Statistical power
Clinical sig Difference gonna
improve wellbeing of
pts
 Formulas
Formula Variables
Equation of straight Y=0 +1x1 0 = Y intercept (value of Y when x =0)
line
1 = gradient (slope)
Standard deviation = population standard deviation
population = sum of…
= each value
= population mean
= number of values in the population

Standard deviation = sample standard deviation

sample = sum of…
= each value
= sample mean
= number of values in the sample
Variance σ² = ( Σ (xi-μ)² ) / N σ²= pop varience
xi= value of ith element
μ= population mean

𝒁 = x̅− 𝝁 /𝝈
N= population size

𝝈 = population/sample standard deviation

Z score x̅ = Individual value

𝝁= Population Mean
Gives you critical values
 SEM = 𝝈 /n
𝝈 = population standard deviation
Standard error of SEM= standard error of the sample
mean
n= sample size

Margin of error= Z * 𝝈 /n 𝝈=Population standard deviation

Confidence level Confidence level = 1 − α
Margin of Error

CL=x̄ ± Z(𝜎/𝑛 )
n= sample size
Confidence Limits sample mean (x̄ )

𝝈 = population standard
CL=x̄ ± Z(SEM) (Z =a critical value or sds

deviation
n= sample size
Confidence interval Range of data around
confidence limit
 Statistical tests
Predictor (IV) Outcome (DV) Varience
Chi-Square Categorical Categorical Fischer’s Exact test-
compares 2 or For small sample size (<5)
more categorical McNemar Test- Running twice on same
variables samples
Fischer’s Exact test- Categorical Chi-squared for For Chi-squared for For
small sample size small sample size
(<5) (<5)

McNemar Test- Categorical Chi-squared Chi-squared Running

Running twice on twice on same
same samples samples
Student T-Test Categorical continuous Comparing means of Unpaired T-Test
2 groups Paired T-Test
Man-Whitney test
Wilcoxon signed rank test
Paired T-Test Categorical
Man-Whitney test Categorical

Wilcoxon signed Categorical

rank test
Analysis of Variance Categorical Kruskal-Wallis test- used to compare
(ANOVA) means of 2 groups (nonparametric)
Kruskal-Wallis test- Categorical used to compare Nonparametric
 means of 2 groups
(nonparametric)
Pearson’s Correlation Need 2 Continuous Scatterplots
Coefficient Continuous Produces r
Variables
Coefficient of r squared
determination
Regression analysis
Spearman’s Rank
Correlation

Choosing main Statistical tests

Predictor data Outcome data type Test
Exploring differences
Categorical
Categorical 2 groups
Categorical more than 2 groups
Exploring associations and prognostications
Continuous and/pr multiple

Choice of appropriate statistical significance test

First variable Second variable Appropriate test of significance Examples
Continuous (C) Continuous (C) Pearson’s Correlation Coefficient (r); Age (C) and measured systolic BP(C)
linear regression
Ordinal (O) ANOVA or Spearman’s Correlation Age (C) and levels of satisfaction O
(rho)
 Dichotomous unpaired Two-Sample t-test systolic BP (C) and stroke (yes/no) O
(DU)
Dichotomous paired (DP) Paired t-test Difference in systolic BP (C)
Nominal (N) ANOVA
Ordinal (O) Ordinal (O) Spearman’s Correlation (rho)
Dichotomous unpaired Mann-Whitney U-test
(DU)
Dichotomous paired (DP) Wilcoxon signed rank test
Nominal (N) Kruskal-Wallis test-
Dichotomous (D) Dichotomous unpaired Chi-Square; Fischer’s Exact probability
(DU) test
Dichotomous paired (DP) McNemar Chi-Square test
Nominal (N) Chi-Square test
Nominal (N) Nominal (N) Chi-Square test
 Data Relationships

Correlation
Monotonic
Relationships
Linear correlation
Covariance
Parametric
NonnPArametric

Interpret Correlation Coefficient

Pearson’s r Strength of relationship
0.90-1.0 (+ or -) Very strong
0.70-0.90 (+ or -) Strong
0.50-0.70 (+ or -) moderate
0.30-0.50 (+ or -) Weak
0.00-0.30(+ or -) Very weak/ negligible

H0 VS Ha/1
H0 Ha/1
Assumption, status quo, nothing new Rejection of the assumption
Assumed to be true/ a given Rejection of the given
Negation of research question Rejection of the question to be “proven”
Always contains an equality (=,≥,≤,) Does not contain equality (≠,<,>)

Type 1 vs type 2
Type 1 error Type 2 error
Lower significance Decreases risk Increases risk
Increasing power Increases risk Decreases risk

Clinical significance VS statistically significant

Clinical significance
yes no
Statistical significance yes Typically assumes the groups, Consider the sample size may be too large
outcomes, or treatments are different
No Consider sample size too small Typically assumes the groups, outcomes, or
treatments are NOT different