04-07 boxplotsimulation

Upload: ly-le

Post on 05-Apr-2018

223 views

Category:

Documents


0 download

TRANSCRIPT

  • 8/2/2019 04-07 BoxPlotSimulation

    1/3

    Normal Boxplots

    48.58 66.16 69.71 68.45 39.57

    32.77 58.67 58.04 50.62 52.99

    34.12 56.49 33.15 2.37 61.95

    28.97 50.69 71.80 47.83 19.82

    62.48 45.41 57.46 46.26 45.67

    74.34 58.27 45.50 51.49 44.26

    47.07 26.03 71.69 40.48 61.76

    55.28 60.14 31.28 52.90 82.5656.33 7.02 47.70 60.01 56.42

    38.18 57.30 44.00 27.64 15.29

    Min Q1 Q2 Q3 Max

    Item X Y Sample 2.37 39.80 50.66 58.57 82.56 Item X Y

    Min 2 1.5 Expected 15.10 39.88 50.00 60.13 84.90 Min 15 4.5

    Q1 40 1.5 Q1 40 4.5

    Q1 40 2.5 Q1 40 5.5

    Q3 59 2.5 Q3 60 5.5Q3 59 0.5 Q3 60 3.5

    Q1 40 0.5 Q1 40 3.5

    Q1 40 2.5 Q1 40 5.5

    Q3 59 1.5 Q3 60 4.5

    Max 83 1.5 Max 85 4.5

    3

    Q2 51 0.5 Q2 50 3.5

    Q2 51 2.5 Q2 50 5.5

    Q3 59 2.5 Q3 60 5.5

    Q3 59 0.5 Q3 60 3.5

    Expected Box Plot Setup

    Sample of 50 Items

    Sample Box Plot Setup

    Mean 50

    St. Dev. 15

    0 20 40 60 80 100

    Sample

    Normal

    Source See Rick Hesse, "Box and Whiskers Plots," Decision Line, March, 1997,

    pp. 17-18. You can change the display format of sample or quartiles to see more

    decimals. Expected statistics are based on the normal distribution. The min and

    max are F-1[(i-0.5)/n] using i = 1, i = 50, and n = 50. For further explanation, see

    Ralph B. D'Agostino and Michael A. Stephens, Goodness of Fit Techniques

    (Marcel Dekker, Inc., 1986, p. 25.

    Press F9 for anew sample.

    Note You can

    change the meanand st. dev.

    Exercises (1) Examine the normal boxplot (shown in blue). In a normal

    distrbution, the quartiles are 0.675 standard deviations from the mean, so

    the box is narrower than the range. Is this true in most of the samples, as you

    press F9 a few times? (2) As you press F9 a few times, compare the samplequartiles (shown in red) to the normal boxplot (shown in blue). How stable

    are the quartiles from sample to sample? (3) Watch the end points of the

    sample boxplot as you press F9 a few times. Are the end points more or lessstable than the quartiles? (4) In general, did your samples usually resemble

    the expected normal boxplot? (4) press F9 ten times. What percentage of thetime did you get a boxplot that definitely did notlook like the normalboxplot? What is the implication?

    Learning Stats

    Copyright 2007

    by

    Th e McGraw-Hill Companies

    This spreadsheet is intended solely

    for educational purposes by licensed

    users ofLearningStats . It may not

    be copied or resold for pro fit.

  • 8/2/2019 04-07 BoxPlotSimulation

    2/3

    Uniform Boxplots

    28.60 63.72 83.72 35.68 31.90

    40.92 40.06 72.56 22.64 37.28

    72.20 36.49 84.69 46.42 67.94

    29.70 66.20 81.60 71.07 71.86

    42.79 36.07 40.75 64.35 40.96

    27.82 79.25 41.96 58.85 27.56

    67.89 69.32 40.31 29.38 65.45

    54.70 22.74 69.11 38.55 65.8153.99 72.78 79.10 47.68 76.22

    56.85 71.17 66.26 22.22 83.31

    Min Q1 Q2 Q3 Max

    Item X Y Sample 22.22 37.60 55.78 70.63 84.69 Item X Y

    Min 22 1.5 Expected 16.00 33.25 50.50 67.75 85.00 Min 16 4.5

    Q1 38 1.5 Q1 33 4.5

    Q1 38 2.5 Q1 33 5.5

    Q3 71 2.5 Q3 68 5.5Q3 71 0.5 Q3 68 3.5

    Q1 38 0.5 Q1 33 3.5

    Q1 38 2.5 Q1 33 5.5

    Q3 71 1.5 Q3 68 4.5

    Max 85 1.5 Max 85 4.5

    3

    Q2 56 0.5 Q2 51 3.5

    Q2 56 2.5 Q2 51 5.5

    Q3 71 2.5 Q3 68 5.5

    Q3 71 0.5 Q3 68 3.5

    Expected Box Plot Setup

    Sample of 50 Items

    Sample Box Plot Setup

    Min 16

    Max 85

    0 20 40 60 80 100

    Sample

    Uniform

    Source See Rick Hesse, "Box and Whiskers Plots," Decision Line , March,

    1997, pp. 17-18. You can change the display format of sample or quartiles to

    see more decimals. Expected statistics are based on the uniform distribution.

    Press F9 for anew sample.

    Note You can

    change the minand max.

    Exercises (1) Examine the uniform boxplot (shown in blue). How is the

    position of the quartiles and end points determined? (2) As you press F9

    a few times, compare the sample quartiles (shown in red) to the uniform

    boxplot (shown in blue). How stable are the quartiles from sample to

    sample? (3) Watch the end points of the sample boxplot as you press F9

    a few times. Are the end points more or less stable than the quartiles?

    (4) In general, did your samples usually resemble the expected uniform

    boxplot? (5) press F9 ten times. What percentage of the time did you

    get a boxplot that definitely did notlook like the uniform boxplot?What is the implication?

    Learning Stats

    Copyright 2007

    by

    The McGraw-Hill Companies

    Th is spreadsheet is intended solely

    for educational purposes by licensedusers ofLearningStats . It may not

    be copied or resold for pro fit.

  • 8/2/2019 04-07 BoxPlotSimulation

    3/3

    Skewed Boxplots2

    1.35 1.24 8.30 0.46 1.38 Slight

    1.11 0.93 0.18 0.11 0.39 Medium

    0.74 0.22 0.34 0.33 0.30 Strong

    7.00 11.85 0.73 3.47 1.22 D.F. 2

    1.70 2.12 2.53 0.45 1.33 Control 3

    0.48 1.44 0.39 2.60 0.65

    1.43 0.03 2.39 0.06 0.16

    1.21 0.36 0.52 0.28 0.360.60 1.01 7.72 4.79 4.420.70 0.46 0.23 0.21 4.18

    Min Q1 Q2 Q3 Max

    Item X Y Sample 0.03 0.36 0.74 1.63 11.85 Item X Y

    Min 0.03 1.5 Expected 0.02 0.58 1.39 2.77 9.21 Min 0.02 4.5

    Q1 0.36 1.5 Q1 0.58 4.5

    Q1 0.36 2.5 Q1 0.58 5.5

    Q3 1.63 2.5 Q3 2.77 5.5Q3 1.63 0.5 Q3 2.77 3.5

    Q1 0.36 0.5 Q1 0.58 3.5

    Q1 0.36 2.5 Q1 0.58 5.5

    Q3 1.63 1.5 Q3 2.77 4.5

    Max 11.85 1.5 Max 9.21 4.5

    3

    Q2 0.74 0.5 Q2 1.39 3.5

    Q2 0.74 2.5 Q2 1.39 5.5

    Q3 1.63 2.5 Q3 2.77 5.5

    Q3 1.63 0.5 Q3 2.77 3.5

    Degree of Skewness

    Expected Box Plot Setup

    Sample of 50 Items

    Sample Box Plot Setup

    0 5 10 15 20

    Sample

    Expected

    Source See Rick Hesse, "Box and Whiskers Plots,"Decision Line, March,

    1997, pp. 17-18. You can change the display format of sample or quartiles to

    see more decimals. Expected statistics are based on the chi-square

    distribution. The min and max are F-1[(i-0.5)/n] using i = 1, i = 50, and n =

    50. For further explanation, see Ralph B. D'Agostino and Michael A.

    Stephens, Goodness of Fit Techniques (Marcel Dekker, Inc., 1986, p. 25.

    Press F9 for anew sample.

    Note You can

    change theskewness.

    Exercises (1) Examine the expected boxplot (shown in blue). Examine the

    position of the median within the box, and the length of the each tail. Does itsappearance reflect the desired degree of skewness (slight, medium, strong)?(2) As you press F9 a few times, compare the sample quartiles (shown in red)

    to the expected boxplot (shown in blue). How stable are the quartiles from

    sample to sample? (3) Watch the end points of the sample boxplot as youpress F9 a few times. Are the end points more or less stable than the

    quartiles? (4) In general, did your samples usually resemble the expected

    boxplot? (5) press F9 ten times. What percentage of the time did you get aboxplot that definitely did notlook like the expected boxplot? What is the

    implication?

    Learning Stats

    Copyright 2007

    by

    Th e McGraw-Hill Companies

    Th is spreadsheet is int ended solely

    for educational purposes by licensed

    users ofLearningStats . It may not

    be copied or resold for pr ofit.