Measures
of Central Tendency
mode = most common
value (think: hip, popular, fashionable)
-
computing mode with graphical
techniques
median = middle value
(think: median of a street)
mean = balancing
point in histogram
-
best guess of a value if
no other information about distribution
-
what people commonly refer
to as the average
sensitivity to outliers
percentile
quartile
Distribution Shapes
unimodal
bi/multimodal
symmetrical
-
uniform (special case) =
all values have same number of observations
positively skewed
(skewed to right)
negatively skewed (skewed
to left)
numerical clue to skew: mean
> or < median