Skewed right distribution

A distribution skewed right has most of its measured points on the left side of dis-

tribution and fewer on the right, suggesting that more data points are in the lower

range of the X axis.

Random distribution

A random distribution has no recognizable pattern and does not reveal much

information.

Statistics

The resulting numbers from the computation of ‘mean’ and ‘standard deviation’

from sample data is called ‘statistics’, in other words it is a measure that describes

the characteristics of a sample. Statistics is the science that deals with collection,

collation, classification, and analysis of data or information contained in ‘sample’

and is used to make inferences about population parameters that are unknown.

Statistics can be subdivided into three broad categories (see Table 7.6). Of the

three types of statistics, ‘descriptive statistics’ is the study of organizing, summa-

rizing, and displaying data, ‘statistical inference’ derives information of a popula-

tion based on the characteristics of a sample, and ‘probability’ is reaching conclu-

sions about a sample by studying the population.

RELEVANCE OF DATA AND DATA COLLECTION

In TQM philosophy, generating accurate data and using it for continual improvement

is sacrosanct. It is a pity that dishonest managers manipulate data to serve their

216 MACRO STRATEGY

Table 7.6: Statistics used in TQM

THREE AREAS OF STATISTICS NORMALLY USED IN TQM

Descriptive Statistics Inductive Statistics Probability

Describes the characteristics Draws conclusions on the Study of a

sample

based on

of a product or process

population

based on information knowledge of the

population

based on data

collected

available about a

sample

(the reverse of statistical

about it. inference)

Organizing, presenting The

sample

has to be Make statements about the

and displaying data in representative of the likelihood of a

sample

having

meaningful patterns for

population

certain features based on

study through: Marketing uses statistical information about the

population

.

z Arrays inference to find the right

z Frequency distributions, sample size to conduct

z Frequency Polygons market surveys on a new

z Pareto diagrams product.

z C-E diagrams, etc.

vested interests. Data management is the primary element for correct deployment

of statistical techniques, and ensuring transparent data collection is an important

characteristic of TQM. Data is the primary element on which statistics and statis-

tical techniques are built, and hence top management’s active interest and super-

intendence is required to ensure that there is no laxity or manipulation in data

management. Data generated by properly implemented statistical techniques give

clues and guidance to anticipate and correct non-conformity and reduce variability

in the system.

Data is used in all areas of statistics viz. descriptive, inductive and probability.

Data is also the basis on which corrective and preventive actions are taken and

reviewed, using measurement and benchmarking. Process control, variation control,

Six Sigma initiatives and even Balanced Scorecards are based on data analysis,

and the success or failure of such initiatives are also indicated by data. Customer

satisfaction or delight is indicated by data. Data is fundamental to the TQM axiom

that states that which cannot be measured, cannot be improved.

Appropriateness and accuracy of raw data is crucial as data are processed

through statistical steps until useful information is generated. Unfortunately there

is a lot of indiscipline in data planning, collection and analysis. TQM philosophy

says, ‘Speak with data and act on data’. Data reflect the real state of affairs, give

warning through trends, and throw up opportunities for improvement. Whatever

is done or not done needs to be captured in time and analysed for continuous

improvement.

Accomplishment is only authenticated by data. Top management has to lay a

lot of stress on the need for honest and efficient data management and a work

culture where data is not something alien or fearsome but a valuable input for

decision-making and improvement. Top management should take extra care to

train and motivate people in data collection and relevant statistical techniques.

Management should also institutionalise systems and procedures to ensure

calibration, training and checking of measuring instruments and sensors to avoid

faulty data.

Frequency Distribution

The frequency distribution graph is a fundamental tool that gives a representative

picture of the nature and extent of total variation and studies the variation after

improvements. The ‘central limit theorem’ discovered by Walter A. Shewhart, the

statistician and quality philosopher of the 1920s, establishes that if measurements

from a process or machine are graphed, they often form a bell-shaped frequency

distribution, which has been likened to a London bobby’s hat. This distribution

curve is also called normal or Gaussian distribution. If a distribution of measurements

is graphed from a chance (constant cause) system, it does not matter whether the

shape of the distribution of individual values (population) is triangular or trape-

zoidal, the averages of different sized samples will have a similar central tendency

and will be distributed in a bell-shape. This bell-shaped distribution occurs fre-

quently in business and nature. A normal curve has three important characteristics.

PROCESS ORIENTATION 217

Get *Total Quality of Management* now with the O’Reilly learning platform.

O’Reilly members experience live online training, plus books, videos, and digital content from nearly 200 publishers.