Mean
Addition of all samples divided by number of samples
Median
Point above which have of the observations fall
- not affected by outliers
- more accurate than the “average” in skewed distributions
Mode
Most common observed variable
Right Skewed Distribution
Tail is on the right hand side
Left Skewed Distribution
Tail is on the left hand side
Measures of spread (2)
1) Range
2) Percentile
3) Variance
4) Standard Deviation
Box and Whisker plot contains: (7)
1) High outlier
2) Maximum whisker
3) Upper quartile (Q3)
4) Median
5) Lower quartile (Q1)
6) Minimum whisker
7) Low Outlier
Variance
S^2
is the sum of the squares of differences from the mean / degrees of freedom minus 1
S^2 = (difference)^2 + (difference)^2 / (n-1)
Standard Deviations
S
- square root of the sample variance
- estimates average variation of n-values from the mean
- tells us how much variability is expected among individuals
Binary or dichotomous data
numbers or % in each category
yes/no style answers
Nominal Data
number and % of subjects in each category
Ordinal Data
Numbers and % of subjects in each category
- median
- smallest and largest
- values/range
Quantitative Data
Graph and visualize the distribution
- mean/median/mode
- smallest and largest
- values/range
- percentiles
- variances
- standard deviations
How are confidence intervals derived and interpreted
Ex) 95% CI
- The interval from ___ to ___ has a 95% chance (probability) to contain the true population mean
- greater sample size = smaller CI
Concept of hypothesis testing and steps (3)
- Hypothesis testing involves comparison of groups
1) Test statistic (t-distribution, z-dist, f-dist)
2) P-value
3) p-value compared to alpha (0.05, 0.1, etc)
Interpretation of p-values
P-value is an indication of the “data occurring” if the null was true
p < alpha = reject Ho
p > alpha = Do not reject Ho
How are Chi-squared tests derived from 2x2 tables
2x2 tables are expanded to show observed vs. expected numbers
Chi-squared uses differences in observed vs. expected numbers to calculate chi-squared statistic
Alternate hypothesis
Ha: two groups are different
What does it mean when a CI for means/risk difference contains 0?
Means it is non-significant
What does it mean when CI for odds ratio/risk ratio contains 1?
Non-significant
What does it mean when CI for odds ratio/risk ratio contains 1?
Non-significant
Parametric Tests (3)
When data is normally distributed
1) T-test
2) ANOVA
3) Regression
Non-Parametric Tests (2)
When data is not normally distributed
1) Wilcoxon rank test
2) Kruskal-wallis test
T-test
Parametric test
- Test whether the mean of a sample or population is different from a particular value
- one sample or one group OR
- two groups (2 sample t-test)
