Partnership for

Performance
University of California/IBM Consulting Group

Data Analysis

Credit to
Mears, Peter. Quality
Improvement Tools &
Techniques. NY:
McGraw-Hill, Inc. , 1995.
 Table of Contents
n Balanced Analysis of Data
n Strategic Goals

n Discrete and Continuous Data

n Data Displays

Pareto Histogram Scatter

Run Control Pie
Bar Horizontal Bar Frequency
Radar Hi-L0-Mean or Median
Mixed Displays

University of California/IBM Consulting Group Partnership for Performance

 Balanced Analysis of Data

Financial Quality
Efficiency and Future Viability
Productivity
Credit to Norton & Kaplan's Balanced Scorecard

University of California/IBM Consulting Group Partnership for Performance

 Balanced Analysis of Data
n Cost of Process, Net Income,
Growth in Assets, etc. n Financial
n Satisfaction with Service,
Compliance with Standards, Test tQuality
Scores, Awards, etc.

tNumber Processed, Number tEfficiency,

Educated, etc.
Productivity
tResearch Activity, Investment in
Training, Literacy in evolving
technologies, Infrastructure, etc. tFuture Viability

University of California/IBM Consulting Group Partnership for Performance

 Strategic Goals

Data and measures

Data
related to strategic
goals Measure Measure

Measure Goals Data

Data Measure

University of California/IBM Consulting Group Partnership for Performance

Strategic Goals

Measurements and
data need to be
related to strategic Increase Decrease
Enrollment
goals to be evaluated. Enrollments
by 4% in 2 Yrs.
by 5%/Year

No Growth Open Law

in Enrollment School, Close
Three Years Theater Dept.

University of California/IBM Consulting Group Partnership for Performance

 Discrete & Continuous Data

Continuous (Variable) Discrete (Attribute)

tMeasurable data n Countable

tPrecision dependent n Specific number of

on measuring units (no decimals or
instrument further breakdown
tData can be possible)
subdivided-length, n Defective/nondefective

weight, height n Pass/fail; Go/no go

University of California/IBM Consulting Group Partnership for Performance

 Discrete & Continuous Data

Continuous Data Discrete

Weight/Height /Thickness l No. Unsatisfied

l Average Service Time
Customers
l Rejections
l Attitude Scale

(Lickert Type) l No. of Transactions

l Average Failure Time

University of California/IBM Consulting Group Partnership for Performance

 Pareto Chart

n Separates the vital few from the trivial many

t Problems
t Symptoms

t Causes

n 80-20 rule: 80% problems from 20% causes

n Type of frequency chart

t Bars arranged in descending order from left

University of California/IBM Consulting Group Partnership for Performance

 Pareto Chart
Lab Inventory Problems 1994 100%

60 95%
85%

50 70%

50%
40
Wrong Part #
Wrong Paperwork
% Bad Inventory Items 30 Scratched or Dirty
Broken
Other
20

10

University of California/IBM Consulting Group Partnership for Performance

 Pareto Chart
University Theft Insurance Complaints 1994
Number of Occurrences 100
%
140
93%
83%
120
67%
100
40%
Claim Pay Delays
80 Claim Pay Amount
Complaint Categories Insurance Rates
60 Insurance Coverage
Other
40

20

University of California/IBM Consulting Group Partnership for Performance

 Pareto Chart Cautions

n Categories need to be mutually exclusive;

not
t Injuries
t Broken Bones

n Categories of problems need to be similar;

not
t Inventory (reduction problem)
t Rejects (zero-based problem)

t Sales (increase problem)

University of California/IBM Consulting Group Partnership for Performance

 Histogram

n Gives a picture of the population

n Displays continuous data

n Displays variables in a data set

n Develops logical groupings

n Patterns of variability reveal underlying facts

about process
n Pattern of variability can suggest theories to

test

University of California/IBM Consulting Group Partnership for Performance

 Histogram
Library Fines Paid - 3/4/95

14

12

10

$14.50
8 $34.50
Number of Fines Paid $54.50
6 $74.50
$94.50

0
Dollar Range of Fines Paid
University of California/IBM Consulting Group Partnership for Performance
 Histogram Steps
n Compute range (R) of data: R = Xmax -Xmin =
n Compute the approximate number of classes (k): k= n

n Compute approximate class width (h): h = R/k =

n Determine the unit of measurement: m =

n Finalize the class width (k): h =

n Compute lower boundary of first class (L1):

(L1) = Xmin - (m/2) =

n Determine lower boundaries of remaining classes:

L2 = L1 + h =
n Construct frequency table

University of California/IBM Consulting Group Partnership for Performance

 Histogram Frequency Table

Class Intervals Tally Frequency

Total

University of California/IBM Consulting Group Partnership for Performance

 Left-Handed Histogram
(Skewed Right)

20

18

16

14

12

10 Click here to create chart

8

University of California/IBM Consulting Group Partnership for Performance

 Comb Histogram

20

18

16

14

12

10 Click here to create chart

8

University of California/IBM Consulting Group Partnership for Performance

 Bimodal Histogram

25

20

15

10

University of California/IBM Consulting Group Partnership for Performance

 Histogram With Outlier

25

20

15

10

University of California/IBM Consulting Group Partnership for Performance

 Scatter Diagram (Relationship)

n Shows relationship between two variables

n Has high visual impact

n Helps generate ideas

n Cause (independent variable) X axis

n Effect (dependent variable) Y axis

University of California/IBM Consulting Group Partnership for Performance

 Scatter Diagram
Travel Reimbursements 1/1/94 - 6/30/94

14

12

10

8
# Expense Receipts # Expense Receipts
6

0
0 20 40 60 80 100 120 140
Days to Process
University of California/IBM Consulting Group Partnership for Performance
 Scatter Diagram
n Strong Positive or Negative Correlation
t High motivation and productivity
t Speed and traffic accidents

t Hours of training and mistakes

t Maintenance expenditures and breakdowns

n No Correlation
t Rise in one quantity produces either rise or fall
in another quantity

University of California/IBM Consulting Group Partnership for Performance

 Scatter Diagram - No Correlation

Number of Resident Students

25

20

15
Number of Annual
Credits
10

0
0 2 4 6 8 10 12 14 16 18
Resident Students
University of California/IBM Consulting Group Partnership for Performance
 Run Chart

n Visual summary of data

n Shows output of process over time

n Shows trends over time

n Easy to understand

n Use Consistency, Good Labels and Standard

Methodology

University of California/IBM Consulting Group Partnership for Performance

 Run Chart
Admission Office Complaints Record
January 1994

350

300

250

200
Number of Complaints Number of Complaints
150

100

50

0
1988 1989 1990 1991 1992 1993

University of California/IBM Consulting Group Partnership for Performance

 Run Chart
n Title and Date of Preparation
n Y Axis (Vertical) Quantity or frequency of

what is being displayed. Must be labeled

n X Axis (Horizontal) Usually represents

time.
n Key (Legend) Explains the items on the

chart
n Scale Should permit data movement to be

clearly seen
University of California/IBM Consulting Group Partnership for Performance
 Run Chart

n Scale Suggestions
t Divide Y axis into equally spaced intervals
l Range covers existing data points and a

cushion
t Round off the values being plotted

l Try to use no more than 3 digit number

t Scale does not have to start at zero

l Truncated scales emphasize changes in data

points

University of California/IBM Consulting Group Partnership for Performance

 Run Chart - Lack of Scale
Differentiation
Admission Office Complaints Record
January 1994

350

300

250

200
Number of Complaints Number of Complaints
150

100

50

0
1988 1989 1990 1991 1992 1993

University of California/IBM Consulting Group Partnership for Performance

 Run Chart

n Keep chart simple

t Don't include more than two lines of data
n Don't change scale from chart to chart if
making comparisons among charts
n Explain major point of chart in accompanying

text
n Concentrate on "Pain" on internal reports

n Reflect positive in external reports

University of California/IBM Consulting Group Partnership for Performance

 Run Chart
Admissions and Complaints
January 1994

400

350

300

250
Number Complaints
# Admissions 200
Number Applicants
150

100

50

0
1988 1989 1990 1991 1992 1993

University of California/IBM Consulting Group Partnership for Performance

 Run Chart - Too Complex
Admissions and Complaints
January 1994

400

350

300

250
Number Complaints
Number Applicants
Number 200
Number Admitted
150 Planned Targets

100

50

0
1988 1989 1990 1991 1992 1993

University of California/IBM Consulting Group Partnership for Performance

 Control Chart

n Visual display of process

n Helps organize process into management

size
n Promotes greater understanding and control

of process
n Shows is process is in statistical control

n Show trends over time

n UCL and LCL = + or - 3 standard deviations

University of California/IBM Consulting Group Partnership for Performance

 Control Chart
n TYPES
t Continuous Variables
l X-Chart are computed on process average

l R-Chart are based on the difference between

the highest and lowest values or range

t Discrete (Attribute) Data

l P-Chart reflects percentages of items

l C-Chart reflects a constant event

University of California/IBM Consulting Group Partnership for Performance

 Control Chart Continuous Variables
(Average of Small Samples)

20

18

16
UCL
14

12
X
10 Avg Days of Process
LCL
8

4
J F M A M J J A S O
2

University of California/IBM Consulting Group Partnership for Performance

 Control Chart Continuous Variables
(Range R-chart)

6
UCL
5

4
R AVG
3

LCL
1 J F M A M J J A S O N D J F

University of California/IBM Consulting Group Partnership for Performance

 Control Chart Unstable Process

40

UCL
35

30

AVG
25

20
LCL
15

10
J F M A M J J A S O N D J
5

University of California/IBM Consulting Group Partnership for Performance

 Control Chart RUN

30

UCL
25

20
AVG

15
LCL
10

5 J F M A M J J A S O N D J

University of California/IBM Consulting Group Partnership for Performance

 Control Chart Trends

30

UCL
25

20
AVG

15
LCL
10

5 J F M A J J A S O N D J F

University of California/IBM Consulting Group Partnership for Performance

 Control Chart Hugging

25

UCL
20

AVG
15

LCL
10

5
J F M A M J J A S O N D J

University of California/IBM Consulting Group Partnership for Performance

 Control Chart Cycling

30

UCL
25

20
AVG

15
LCL
10

5 J F M A M J J A S O N D J F M

University of California/IBM Consulting Group Partnership for Performance

 PIE Chart

n Divided circle with pieces arranged in

descending order
n Shows percent each item makes up of the

whole
n Compute "pieces" by multiplying percent

times 360 degrees to aid in plotting.

University of California/IBM Consulting Group Partnership for Performance

 Pie Chart
Paperwork Complexity 1994 Analysis

Travel Claims
58.4%

Other
8.3%

Personnel
Purchases 12.5%
20.8%

University of California/IBM Consulting Group Partnership for Performance

 Bar Chart

n Shows data as separated rectangles (bars) of

equal width and of different heights
n Compares differences in quantities for

discrete data
n X axis show categories and Y axis shows the

measurement

University of California/IBM Consulting Group Partnership for Performance

 Bar Chart
Paperwork Complexity, 1994 Analysis

70

60

50

40
Percent Non-Value
Paperwork
30

20

10

0
Travel Claims Purchases Personnel Other

University of California/IBM Consulting Group Partnership for Performance

 Horizontal Bar Chart

n Compares "before" with "after" using

separate rectangles (bars) of equal width and
of different lengths
n Compares differences in quantities for

discrete data
n X axis shows measurement, Y axis shows

categories

University of California/IBM Consulting Group Partnership for Performance

 Horizontal Bar
Food Service Customer Satisfaction
January and November 1994 Surveys

Food Qty

Before Training
Service Category Quick Service
After Training

Friendly Staff
Not Happy Very Happy

0 1 2 3 4 5 6 7 8
Satisfaction Scale
University of California/IBM Consulting Group Partnership for Performance
 Frequency Chart

Used with discrete data to show its

distribution
Compares differences in quantities for
discrete data
X axis shows categories, Y axis shows
quantity

University of California/IBM Consulting Group Partnership for Performance

 Frequency Chart
Correct Answers on Test
January 1994 Analysis

14

12

10

8
High
Number Students
Low
6

9 11 13 14 15
2 10

0
14 12 10 8 6 4 2
Number Correct
University of California/IBM Consulting Group Partnership for Performance
 Radar Chart

n Circle divided into "slices" representing

questions. Answers plotted along slices
from center.
n Represents relative strengths of important

activities.
n A line from circle's center represents an

activity.
n Useful for displaying Lickert scales.

University of California/IBM Consulting Group Partnership for Performance

 Radar Chart
Food Quality
7
Food Service 6
Survey 1/19/95 5
4
3
Price 2 Quick Service

1
0

Serving Size Friendly Staff

University of California/IBM Consulting Group Partnership for Performance

 Hi-Lo-Median or Mean Chart

n Comparison of performance data

n Reflects the highest, lowest and either

median or mean
n Compare own institution to other individual

institutions or average of a group of

institutions

University of California/IBM Consulting Group Partnership for Performance

 Hi-Lo-Median or Mean Chart

% Students with Internet Access from Residence

January 1994

35

30

25

20 High
Percent Students Low
15 Mean

10

0
ABC College DEF College Group 1 Group 2

University of California/IBM Consulting Group Partnership for Performance

 Selection of Display - Food Service
Customer Satisfaction

Food Quality
7 8
6
7
5
4 6
3
5
Price 2 Quick Service
1994
1 Score 4
1995
0
3

0
Serving Size Friendly Staff

ly
y

ck

e
ze
lit

ic
nd
ui

Si
ua

Pr
ie
Q
Q

Fr
1994 1995
University of California/IBM Consulting Group Partnership for Performance
 Selection of Display - Lost Work Days
per 100 Faculty & Staff FTE
FTE
Lost Work Days
Lost1994
Workdays All Institutions

3000 300

250
2500

200
2000
150 ABC College

1500
100

1000 50
lost Work Days
1994 Year
500 0

ge
um
um
n
n

ia
ea

le
ed

im

im
M

ol
M

in

ax
0

C
M

C
B
0 50 100 150 200 250 300

A
Work Days Lost/100 FTE Work Days Lost/100 FTE
University of California/IBM Consulting Group Partnership for Performance