1. Variability

2. Variability Central tendency locates middle of distribution
How are scores distributed around that point?
Low variability vs. high variability
Ways to measure variability
Range
Interquartile range
Sum of squares
Variance
Standard deviation

3. Why Variability is Important
Inference
Reliability of estimators
For its own sake
Consistency (manufacturing, sports, etc.)
Diversity (attitudes, strategies) 49 58
100 61 97 55 31 52 13 43
178
154
136
103 94
181 91
109 46 28 34
175 37
139
106 19
184 88
112 64 76
172
157 22 16 85
142
130
151 67
121
124
160 40
145
127 82
133
169
166
163
118
148 79 25
115 73
187 70
Average = 49 58
100 61 97 55 31 52 13 43
178
154
136
103 94
181 91
109 46 28 34
175 37
139
106 19
184 88
112 64 76
172
157 22 16 85
142
130
151 67
121
124
160 40
145
127 82
133
169
166
163
118
148 79 25
115 73
187 70
Average = 87.25 49 58
100 61 97 55 31 52 13 43
178
154
136
103 94
181 91
109 46 28 34
175 37
139
106 19
184 88
112 64 76
172
157 22 16 85
142
130
151 67
121
124
160 40
145
127 82
133
169
166
163
118
148 79 25
115 73
187 70
Average = 121
m = 100
100
105
108 99 97
106
104 93 96
109 95 92
103
102 94 98
101
107
Average = 99.5
100
105
108 99 97
106
104 93 96
109 95 92
103
102 94 98
101
107
Average = 99.13
100
105
108 99 97
106
104 93 96
109 95 92
103
102 94 98
101
107
Average = 100.5 M
87.3
121
111.6 M
99.1
100.5
99.5

4. Range Distance from minimum to maximum Sample range depends on n
More useful as population parameter
Theoretical property of measurement variable
E.g. memory test: min and max possible
Rough guidelines, e.g. height
Measurement unit or precision
X = [66.2, 78.6, 69.6, 65.3, 62.7]
78.6 – = 16.0
78.65 – = 16.0
(11 – 1) + 1 = 11

5. Interquartile range Quartiles Interquartile range
Values of X based on dividing data into quarters
1st quartile: greater than 1/4 of data
3rd quartile: greater than 3/4 of data
2nd quartile = median
Interquartile range
Difference between 1st and 3rd quartiles
Like range, but for middle half of distribution
Not sensitive to n  more stable
6 – 3 = 3
X = [1,1,2,2,2,3,3,4,4,4,4,5,5,5,5,6,6,6,6,6,7,7,7,8]
1st quartile = 3
3rd quartile = 6

6. Sum of Squares Based on deviation of each datum from the mean: (X – m)
Square each deviation and add them up m
115 88 94
108
122
133
145
729 = 272
441 = 212
49 = 72 27 21 7 7 18 30
2 = 49
2 = 324
2 = 900

7. Variance Most sophisticated statistic for variability
Sum of squares divided by N
Mean Square: average squared deviation
729 = 272
441 = 212
49 = 72 m
115 88 94
108
122
133
145 27 21 7 18 30
2 = 49
2 = 324
2 = 900

8. Standard deviation Typical difference between X and m
Again, based on (X – m)2
Variance is average squared deviation, so sqrt(variance) is standard deviation
X = [5, 3, 7, 6, 4, 6, 8, 7, 4, 2, 3, 5]
m = 5
Square-root
Average
X – m = [0, -2, 2, 1, -1, 1, 3, 2, -1, -3, -2, 0]
(X – m)2 = [0, 4, 4, 1, 1, 1, 9, 4, 1, 9, 4, 0]
Square
Average

9. Why squared difference?
Could use absolute distances, |X - µ|
Would be more intuitive: average distance from the mean
Squares have special mathematical properties
Can be broken into different parts
Central Tendency:
Common to all scores
Sum of Squares:
Differences among scores