Comparing Histograms - Bimodal Distribution?

[cable6]

Hi all

I have done the normal distribution calculation for the data attached. Its the measurement of length generated by a set of winders one situated in the North and the other one from South.Eventhough I calculated and submitted the report my manager is not happy with the way I have done and he is saying it as Bimodal distribution.Can anyone give a good comparison of the histogram I have attached

Regards and thanks in advance

 

[Miner]

Re: Comparing Histograms

Please be more specific in your question and attach the raw data if possible. What is your manager's goal?

The information that you have attached appears to be four distinct distributions from four different process streams. Without the raw data and more process information, it is difficult to speculate further.

If these are four different process streams, you can reduce the total variation by adjusting the four mean values closer to the same target value.

 

[Steve Prevette]

Re: Comparing Histograms

Hi all

I have done the normal distribution calculation for the data attached. Its the measurement of length generated by a set of winders one situated in the North and the other one from South.Eventhough I calculated and submitted the report my manager is not happy with the way I have done and he is saying it as Bimodal distribution.Can anyone give a good comparison of the histogram I have attached

Regards and thanks in advance

You've got a lot going on in that chart. Not only do you appear to be comparing North and South, but also Old and New. Which is the important comparison?

Which of the 4 curves does your manager think is bimodal?

It may help for readability (if you must keep all 4 curves) to plot the histograms as line charts rather than bar charts.

I do notice you have shifted all to percentage, but you may want to portray the actual counts since if the sample sizes are different, there may be different interpretations.

If you could provide the raw data in an excel spreadsheet attachment it could help us help you.

 

[M Greenaway]

Its not bi-modal, you appear to have four distinct distributions on one chart which gives the appearance of bi-modal distributions.

Maybe a box plot would give a better comparative picture ?

 

[Jim Wynne]

Its not bi-modal, you appear to have four distinct distributions on one chart which gives the appearance of bi-modal distributions.

I agree--I suspect that the OP's boss looked at the chart briefly and came to the bimodal (multi-modal, more accurately) conclusion.

Maybe a box plot would give a better comparative picture ?

It might be that just four separate histograms would do, but if combining the data in one chart is desirable, I think a line chart would be the answer. Another one might be printing four separate histograms on transparencies and overlaying them for comparison, if only a "picture" of the charts is necessary.