Attached is a sample suvey that has been tabulated.Pray tell me what the Pareto is telling me which the pie is not.I see no reason to spend time doing a pareto when I can make the same deductions from a pie chart.
Or am I doing the wrong thing??
Please excuse my ignorance if you find this too dumb.

Sorry-here is the attachment.This a random survey of a school cafetaria.

Just in case the other one did not make sense,a new one.
Again the same question-
What is the Pareto telling me that the pie chart cannot???

Nitin,

In principle, the Pie Chart would show you the distribution of data across parameters ( looks like complaints in your case) while Pareto Chart would also find the cumuative distribution. It is a well known theory formulated by Pareto that 80% of the effects are caused by only 20% of the causes, or in such cases 80% of the trouble is caused by 20% of the causes. A Pareto Chart will help you to identify those top 20% of the causes and prioritize the course of action.

I have perhaps reiterated what you already know.

Analyzing your spreadsheet, it looks like there are only 5 categories that share a fair amount of % of complaints each ( between 15 to 25%) giving an appearance that all categories are contributing almost equally to the complaints. In such a case I would try to resolve my survey into more specific and more categories of complaints. Frame more specific questions within each category. This should bring the picture out more clearly. Higher the resolution of the measurement system, more the information you will be able to get.

Hope this helps.

I was composing the reply while you posted the second Excel file.

You can see yourself that your second survey brings out the 80-20 law clearly.

Hi Ravi,
I am not disputing the law.What I am saying is that to anybody who is aware of the "law' the same deduction can be made by looking at the Pie chart.What additional information does the Pareto Chart give me to help me zero in or prioitise my efforts?Looking at the Pie,I would for example decide to go after the big slices first.Looking at the Pareto,it tells me exactly the same thing.So why all this cumulative percentage conversion,charting plotting etc ?

Sure, the pie chart would indeed tell you which slices are big and which are small.

It pays to measure numerically though, rather than describing the slices as merely big or small. The question of how big a slice should be described as big enough for attention becomes subjective. You will have to do a quick mental arithmetic to arrive at the cumulative 80% of the slices.

Once I have the cumulative frequency curve in a Pareto Chart, the method I commonly use is to project the 80% mark from the Y axis to the curve, and drop the intrecept down on the x axis from that point. All bars to the left of this dividing line can be treated as important, since they will be addressing 80% of the effects. Once this is done, I turn my attention to another process where another 80% of pressing matters demand attention.

I find this a very effective way to prioritize.

Let's say the Pareto Chart makes the 80-20 law more 'visible'.

Spot on Ravi. Another consideration is that a Pareto is easier to handle when there are larger numbers of catagories. The more slices you make in a pie, the harder it becomes to read it accurately. Besides there is no real challenge in spelling "pie".

Dave

Charts are just visual aids. The moment you start arranging your data in descending order looking for top 20% troublemakers, you are already following the Pareto principle.

Pie charts are generally used to to compare parts or 'slices' to the 'whole', the basic purpose of Pareto is to 'rank' the items.

Atul said it all when he spoke of the Pareto Principle - I dont believe there is actually such a thing as a Pareto Chart.

The Pareto Principle is the 80:20 rule that we all know and love. Traditionalists might represent this as a histogram in descending order, however as the original poster rightly points out a pie chart can also achieve this.

Thanks for enlightening me.However ,one more question-Is there any specific type of distribution/data a Pareto Principle would be more relevant to?
If this histogram(1st chart) is to be converted to a Paynter Chart,how would I calculate the cumulative percentage,will it be based on overall total of all or based on the individual total for each data series?

Here is Dr. Juran´s Article published in 1975.:bigwave:

Pareto combine in the same graph two info: Numbers and Cumulated percentages vs. Events.
His major scope is to visualize very easy: Major contributors (max.2-3), Main contributors ( 80%).
If you have lots of parameters on the X axes, you can use a scroll bar to see the rest.
Seeing “column heights” increases the visual impact.
We adapted “Pareto” to “Pareto Like”, see attachment, to increase the difference of the column heights.

At first vu, the info seems to be the same in Pie graph.
But please take a look to the attachment. Pareto is linear, Pie –polar.

We asked our operators 3 years ago and they preferred Pareto. So, from that time, all our SQC applications are designed to get automatically “Pareto Like” graphs for ‘defects”.

vio