Quality Circle Forum of India

presents

Problem Solving Techniques

- a simplified revised practical guide

What is Data collection?

Data collection is nothing but the collection of the required information in figures for the statistical analysis of a problem. This provides a sound basis for decision making and corrective action.

Anything we do we would like to evaluate to find out whether we are doing it rightly.

For conventional function, our evaluation is normally based on past experience

Sir, I think we should rely only on facts before taking any decision

But basis or guidance for decision is facts.

Did I not tell you that the prices of my stocks are going to sky rocket. My sixth sense never fails me.

Experience and sixth sense may be necessary

However, these are liable to invite error in judgment due to the decision being too personal or out of erroneous memory.

To avoid this, it is important to get the facts in terms of numbers or quantity.

TYPE OF DEFECT

NUMBER OF FANS

CUM

% OF TOTAL

CUM %

F. G. A. B. C. E. D. H.

Regulator Bend in Screws Motor winding Bend in wings Cracks in Central rod Scratches Capacitor Others

40 25 20 15 14 12 10 14

40 65 85 100 114 126 136 150

27% 17% 13% 10% 9% 8% 7% 9%

27% 44% 57% 67% 76% 84% 91% 100%

Total

150

100%

These numbers are called data. They are expression of facts of any activity or feature in numerical terms.

MILEAGE : 55 Km/litre

Examples: Fuel efficiency of a scooter,

Clock speed 233 Mhz.

Clock speed of a computer

CUSTOMER COMPLAINTS 3000 complaints in four months

Volume of customer complaints

100 degree Celsius

Boiling point of water

PURPOSE OF DATA COLLECTION

Data are collected for a variety of purposes. In industry, the following are more common.

QUALITY CONTROL DEPARTMENT

Quality Evaluation Raw materials, semi finished products and finished products

Process Control To assure that products turned out meet specified requirements of customer/design.

GOOD DETERGENT HIGH PRICE

GOOD DETERGENT LOW PRICE

Improvement Trials - To improve product

quality, reduce costs and increase productivity

Problem Solving For information through out the problem solving process by Quality Circles

Total sales of VIJAY CHAIN OF SUPERMARKETS (in crores)

12 10 8 6 4 2 0 TOTAL SALES 1991 1992 1993 1994 1995

Compare Performance From time to time measure improvement in production, profit, quality, competitiveness etc.

TYPES OF DATA Generally data are of two types: a) Measurement data b) Attribute data

Measurement data
Diameter of a Steel Wire measured using a screwgauge 1st Diameter of the steel wire (in cm) 2nd 3rd 4th

0.156

0.152

0.153

0.156

Weight, diameter, tensile strength, % carbon, reaction time etc. Basic observation obtained by using an instrument or measuring process.

Measurement data
SOLDER BATH TEMPERATURE TIME 0800 0900 1000 1100 1200 1300 1400 TEMPERATURE 62 61 59 58 60 64 63 C C C C C C C

These data are continuous, they are also called continuous variable. A variable is one which varies i.e takes different values. Time, for instance is a continuous variable.

Attribute Data
SEPTA LIMITED
NUMBER OF TRUCK TYRES REJECTED AND PASSED BY THE QUALITY CONTROL DEPARTMENT DURING THE MONTH OF JANUARY 1996

TYRES REJECTED TYRES PASSED

JANUARY 1996

5000

22000

Data obtained by classification into two or more categories.

Attribute Data
SEPTA LIMITED NUMBER OF TRUCK TYRES REJECTED AND PASSED BY THE QUALITY CONTROL DEPARTMENT DURING JAN-JUN 1996

TYRES REJECTED JAN FEB MAR APR MAY JUN 1996 1996 1996 1996 1996 1996 5000 2000 7000 1000 2000 8000

TYRES PASSED 22000 25000 20000 26000 25000 19000

These data are in whole numbers and are called discontinuous or discrete variables.

Attribute Data
Collection of information or generation of information as Dr. Juran says begins and ends with questions.

Attribute Data
WHY
WHEN

WHAT
WHO

WHERE
HOW Skill in collection of information lies in asking the right questions using the 5 - Ws & 1 - H.

CHECK SHEET An easy and effective way of collecting data is by using well designed formats. A check sheet is a simple data recording form specially designed so that data can be interpreted readily from the sheet itself.

SOLDER BATH TEMPERATURE

CHECKED BY : DATE:
TAKE READING OF TEMPERATURE NEAREST TO DEGREE.

TIME

TEMP (deg. cel.)

TIME

TEMP (deg. cel.)

Items to be checked are pre-printed so that recording becomes very easy.

Check sheets are of the following types:

a) b) c) d) e) f)
g)

Process control information checks Product quality assurance checks Defective item checks Defect location checks Defective cause checks Process variation distribution checks Others

PROCESS CONTROL INFORMATION CHECK

PROCESS CONTROL INFORMATION CHECK

This format enables to record process control checks.

PROCESS CONTROL INFORMATION CHECK

SOLDER BATH TEMPERATURE

CHECKED BY :Preeti DATE:18-6-98
TAKE READING OF TEMPERATURE NEAREST TO DEGREE. TIME TEMP (deg. cen.) TIME TEMP (deg. cen.)

0800 0900 1000 1100 1200

60 62 59 58 59

1300 1400 1500 1600 1700

61 58 63 63 59

Note: Power failure from 12.00 to 12.15 hours

PRODUCT QUALITY ASSURANCE CHECK

PRODUCT QUALITY ASSURANCE CHECK

This check sheet is useful in the final test of the product at the end of manufacturing line

PRODUCT QUALITY ASSURANCE CHECK

The purpose of the check is not the check of only the final assembly or operation but to carry out a comprehensive check of quality that has been built through the previous stages.

PRODUCT QUALITY ASSURANCE CHECK

Actual Observation No PARAMETER Space 1 2 3 4 5

1. 2. 3.

Axial clearance Cleanliness of Crank S Surface finish : : :

0.1.0.45 Clean 0.4

6.3 OK 300

0.18 OK 320

0.22 Not OK 325

0.17 OK 315

0.21 OK 300

:
24. Paint Finish Glossy OK OK OK OK OK

An example of format used for checks on diesel engine at final end is shown here.

PRODUCT QUALITY ASSURANCE CHECK

Permanent records of the product quality assurance check sheets.

It is clear that all facts are made without fail. This way there is no chance of forgetting and the check sheets remain as a permanent record that can be referred back later whenever necessary

CHECK LIST TYPE CHECK SHEET

CHECK LIST TYPE CHECK SHEET

Look at Check Sheet for maintenance work

CHECK LIST TYPE CHECK SHEET

Frequent checks must be made to keep equipment work efficiently and trouble free.

CHECK LIST TYPE CHECK SHEET

Some checks are made daily, others weekly, still others monthly or at fixed intervals.

CHECK LIST TYPE CHECK SHEET

With even the slightest absent mindedness, it is easy to forget a check up or even whether a check was really made or not

CHECK LIST TYPE CHECK SHEET

M/L No.: Location: Component Check Points Shop manager : Supervisor: Months J F M A M J J A S O N D

Drive Gear Box Bearing Transmitter ELECTRIC Motor Control Wiring Instrumentation

Noise Vibration Temperature Operation

Operation Age

In such cases, a check sheet helps to carryout work with certainty; it will also reveal forgotten test items. Example: Preventive maintenance of a plastic moulding machine

DEFECTIVE ITEMS CHECKS

DEFECTIVE ITEMS CHECKS

To reduce the number of defectives, it is first necessary to know the type of defects and their magnitude

DEFECTIVE ITEMS CHECKS

Since each defect may have different causes, it is of no use just to list the total number of defects.

DEFECTIVE ITEMS CHECKS

SI No. 1. 2. 3. 4. 5. * * DEFECT BARE METAL OVERFLOW DUST LESS GLOSSY PAINT PEEL TALLY //// //// //// //// //// //// //// //// //// //// //// /// 58 //// //// //// //// //// //// //// //// //// // //// //// //// //// //// //// //// //// //// //// //// //// //// //// //// //// //// //// //// //// //// //// //// //// //// //// //// //// //// //// // 47 80 40 32

*
17. OTHERS //// 4

Let us look at the tally of inspection results on painted refrigerator doors as they come out of the painting booth.

DEFECTS LOCATION CHECK SHEET

DEFECTS LOCATION CHECK SHEET

This kind of check sheet is meant to be placed in the drawing or sketch of the product so that location of defects can be identified and investigated.

DEFECTS LOCATION CHECK SHEET

CHECK SHEET (IN PICTURE) DEFECT IN CHIPS

FUNCTIONAL REJECT

PARAMETRIC REJECT

Defects in chips were located as shown here.

DEFECTS LOCATION CHECK SHEET

Clues to control of defect in chips can be obtained from this

DEFECTS LOCATION CHECK SHEET

Such check sheets are helpful in controlling defects in castings, glassware, plastic components etc.

DEFECTS LOCATION CHECK SHEET

As can be seen, this type of check sheet leads to quick action and is an important tool in process analysis.

DEFECTIVE CAUSE CHECK SHEET

DEFECTIVE CAUSE CHECK SHEET

This type of check sheet is used to generate information on causes and their correspondence to the observed defects.

DEFECTIVE CAUSE CHECK SHEET

50 45 40 35 30 25 20 15 10 5 0

East West North

... . .. ...... .. .. ...... .. .. .

Before

Data may have to be later analysed by stratification or scatter diagram to get clues.

After

DEFECTIVE CAUSE CHECK SHEET

Consider the example which shows recording format for defectives in bakelite knobs categorised by workers, machines, day of the week and type of defects.

DEFECTIVE CAUSE CHECK SHEET

M/C Operator Monday AM oo x PM ox x Tuesday AM PM xx Wednesday Thursday Friday AM PM AM ooo oooo xxx xx ooo PM AM PM o oooo xx oo

1
2 1

.
oxx

ooo

.
xx

..
ooo

..
x

.
xx ooo x x Blow hole Poor finish oo

..
x ooo

B
2
o Surface scratch

Distorted

Others

By a mere glance, it will be possible to identify which worker or which machine and on which day there is more proneness to produce defectives.

DEFECTIVE CAUSE CHECK SHEET

This is a rough analysis for cause and effect relationship.

PROCESS VARIATION DESCRIPTION CHECK

Measurements taken on a product or rejects of inspection at incoming material stage or inprocess can straightaway be recorded in a check sheet to reveal the variation in quality.

Process Variation Checksheet

Results of inspection on sack-hole thickness of 'nozzles' are recorded in a data sheet as shown.

The check sheet shows quickly where is the process set, what is the extent of variation and what is the conformance to specification limits

34 out of 45

Obviously, machine has been set nearer the upper specification limit and hence a large number that is 34 out of 45 are above USL.

0.83 - 0.90

Observed process variation is only 0.83 to 0.90 compared with specification band of 0.75 to 0.85.

Such check sheets are handy in discovering patterns of different vendors supplying the same part. Inferences can be drawn by just looking at the information filled check sheets.

This type of check sheet can be examined for the following information. Does the pattern resemble a symmetric shape, like single peak or are there two or more peaks?

Are the values skewed to one side and are there isolated values?

What is the conformance to specifications limits?

Specification

-3s

+3s

Is the centre of the distribution close to the target value given in specification

Specification

-3 s

+3 s

Is the width of the distribution greater than the specification band?

DATA SHEET

This is a simple tabular form for recording data. Here additional processing is needed for interpretation.

POWER CONSUMPTION (kWh) OF REFRIGERATOR

1.97 2.24 2.10 2.43 2.52 2.31 1.87 2.13 1.86 1.83 2.08

2.49 1.75

2.29 2.07 1.98

2.24 2.20 1.96 1.76 1.96 2.05 1.92 1.92 2.04 2.02

2.42 2.52 2.16 1.91 2.00 1.92 2.14 2.04 2.16 2.02

2.16 2.32 1.94 2.06 2.29 2.25 2.08 2.00 1.74 2.23

2.89

2.45 2.02 2.16 2.08

2.13 2.04 2.33 2.36 2.17 2.08 1.96 2.07

1.95 2.00 2.23 1.95 2.05 2.17 2.17

2.08 2.34 1.95 2.64

HOW TO INTERPRET COLLECTED DATA

Before coming to any conclusion from the data collected, it is wise to verify that the data collection was done on the basis of what was required.

Look at the indications of the data collected. Is there any evidence of bias in the collection process, In the number of observations collected, in the number specified etc.

POWER CONSUMPTION (kWh) OF REFRIGERATOR

1.97 1.75 1.98 1.76 2.31 2.08 1.95 1.92 2.04 2.02

2.49 2.24 1.96 1.91 2.16 2.23 1.92 2.04 2.16 2.23

2.29 2.07 2.16 2.06 2.00 2.05 2.14 2.00 1.74 2.07

2.24 2.20 1.94 2.04 1.96 1.92 2.08 2.08 1.96 2.08

2.42 2.52 2.13 2.52 2.00 2.25 2.17 1.86 1.83 2.64

2.16 2.32 2.43 2.02 2.29 2.36 2.13 2.34 1.95 2.17

2.89 2.10 2.45 1.95 2.33 1.87 2.08 2.05 2.17 2.02

The data should be able to answer the questions we ask as the data have been collected on the basis of what we wanted.

It may not be able to answer other questions that we did not anticipate.

SOLDER BATH TEMPERATURE

CHECKED BY : Preeti DATE: 18-6-98

TAKE READING OF TEMPERATURE NEAREST TO DEGREE. TIME TEMP (deg. cel.) TIME TEMP (deg. cel.)

0800 0900 1000 1100 1200

60 62 59 58 59

1300 1400 1500 1600 1700

61 58 63 63 59

Usually, a check sheet, with its simple collection and analysis format is intended to provide quick answers to simple questions

A complete data sheet however will often support many levels of analysis and stratification, provided they have been anticipated in the design.

ERRORS IN INTERPRETATION

Errors in Interpretation are as follows:

Omit the data from the metros.That should change the picture in our favour.

Exclusion Bias : make sure that no area is left out and what is included has aproper representation in the collection of sample,

Sir, the format is ready. But we have not tested it for accuracy.

Dont bother, nobody will notice. MD wants it to be introduced immedietly. We will look into the problem later.

Interaction, Perception, Operation Biases

Any one of this can happen, if the formats were not properly and adequately prepared and tested in working condition. Proper training and instruction should be ensured to avoid any or all of such biases.

Sir, we could complete the survey of only 30 segments out of 45 planned. What do we do ?

We have no time. Client wants it urgently. Fill up an average figure for the balance. That will do.

Non Response Bias: never try to fill up with average in places of missing data. Sampling plan is based on different types of population. Missing data can bias the result.

Estimation bias: This bias is beyond our scope. The very system and formulae used to collect data (sample size, types of question etc. may be wrong, causing certain types of biases. You cannot do anything about it unless and until you were the designer of the system.

What makes you think Arundati Roys prediction that BJP will win in UP wont come true.

I say that because the survey was not conducted properly. The questions asked were not properly framed & sample size was too small.

Estimation bias

HOW TO COLLECT DATA

Data can be collected in the following ways:

Formulate good questions that relate to specific information needs of the project.

Consider appropriate data analysis tools and be certain the data needed for the analysis are being collected. Wherever possible, collect continuous (variable) data.

Define comprehensive data collection points.

Select an unbiased collector

Understand data collectors and their environment

M/C Operator

Monday
AM PM

Tuesday
AM PM

Wednesday
AM PM

Thursday
AM PM

Friday
AM PM

1
2 1

B
2
o Surface scratch x Blow hole Poor finish

Distorted

Others

Design data collection forms

Prepare the instructions

Test data collectors.

Audit the collection process and validate the results.

BENEFITS OF DATA COLLECTION

Data collection have the following benefits: Right decision can be made

Errors due to subjective feeling or personal bias are avoided.

Agreement on decisions necessary rather than different or subjective opinion

Measurement understandable to all

Assessments of magnitude of improvements

I just had a look at the data collected. Is the quality control department sleeping !!

Discovery of causes affecting quality and productivity

USEFUL POINTS FOR DATA COLLECTION

Useful points are as follows:

Be clear in mind about objective

Prepare the question which will make it clear and precise

Collect data relevant to that

Reading in Machine A

1.97

Use data sheet, check sheet and/or check list as per the requirement

2.49 2.89 2.52 2.16 1.76 2.02 1.96 2.23 1.87 2.17 2.00 2.04 1.95 2.08

2.29 1.75 2.32 1.94 1.91 1.95 2.00 2.05 1.95 2.13 2.08 2.16 2.17 2.64

2.24 2.24 2.10 2.13 2.06 2.31 2.29 1.92 1.92 2.08 1.86 1.74 2.02 2.17

2.42 2.07 1.98 2.43 2.04 2.16 2.33 2.25 2.14 1.92 2.34 1.96 2.23 2.02

2.16 2.20 1.96 2.45 2.52 2.00 2.08 2.36 2.08 2.04 2.05 1.83 2.07

Keep them simple and easy and reduce opportunities for error.

Capture data for analysis reference and tractability

FORM

Form should be self explanatory

Analyse the collected data for its credibility and relevance (for whom, when or who collected the data are important.)

Present data in a way that clearly answers the question.

SUMMARY
Most quality improvement tools depend on reliable, accurate data. If such data are not available, they must be collected. The datacollection process must be driven by the information questions we formulate based on our information needs. In short, know what question you are trying to answer before you collect any data.

Prepared by G. Vijaykumar