Does anyone know how to create the notches in the vertical box whisker
charts?

An example can be found in figure 10 of the following article:

http://www.qualitydigest.com/oct97/html/excel.html

Thank you!!

Hi,

The technique describes how to use an xy-scatter to construct the boxes.

You need to add a few more xy pairs in order to reduce the width of the
Median line and form the notches.

Cheers
Andy

wrote:
Does anyone know how to create the notches in the vertical box whisker
charts?

An example can be found in figure 10 of the following article:

http://www.qualitydigest.com/oct97/html/excel.html

Thank you!!

On Jun 15, 8:47 am, Andy Pope wrote:
Hi,

The technique describes how to use an xy-scatter to construct the boxes.

You need to add a few more xy pairs in order to reduce the width of the
Median line and form the notches.

Cheers
Andy



wrote:
Does anyone know how to create the notches in the verticalboxwhisker
charts?

An example can be found in figure 10 of the following article:

http://www.qualitydigest.com/oct97/html/excel.html

Thank you!!- Hide quoted text -

- Show quoted text -

Can you elaborate on the technique?

Use this set of xy pairs to draw a notched version

X Y
Min 1 3
Min 1 5
Min 1 4
25th 3 4
25th 3 7
TopNotchLeft 5.342592593 7
TopNotchMid 5.5 6.5
TopNotchRight 5.657407407 7
75th 8 7
75th 8 1
BottomNotchRight 5.657407407 1
BottomNotchMid 5.5 1.5
BottomNotchLeft 5.342592593 1
25th 3 1
25th 3 4
Skip
Median 5.5 1.5
Median 5.5 6.5
Skip
75th 8 4
Max 10 4
Max 10 5
Max 10 3

The formula for X value of NotchLeft is
=MEDIAN($A$1:$A$50)-(1.7*((1.25*(PERCENTILE($A$1:$A$50,0.75)-PERCENTILE($A$1:$A$50,0.25))/(1.35*50))))

The formula for X value of NotchRight is
=MEDIAN($A$1:$A$50)+(1.7*((1.25*(PERCENTILE($A$1:$ A$50,0.75)-PERCENTILE($A$1:$A$50,0.25))/(1.35*50))))

All other formula are as William W. Dorner's example.

Cheers
Andy

wrote:
On Jun 15, 8:47 am, Andy Pope wrote:

Hi,

The technique describes how to use an xy-scatter to construct the boxes.

You need to add a few more xy pairs in order to reduce the width of the
Median line and form the notches.

Cheers
Andy



wrote:

Does anyone know how to create the notches in the verticalboxwhisker
charts?

An example can be found in figure 10 of the following article:

http://www.qualitydigest.com/oct97/html/excel.html

Thank you!!- Hide quoted text -

- Show quoted text -


Can you elaborate on the technique?

On Jun 16, 8:27 am, Andy Pope wrote:
Use this set of xy pairs to draw a notched version

X Y
Min 1 3
Min 1 5
Min 1 4
25th 3 4
25th 3 7
TopNotchLeft 5.342592593 7
TopNotchMid 5.5 6.5
TopNotchRight 5.657407407 7
75th 8 7
75th 8 1
BottomNotchRight 5.657407407 1
BottomNotchMid 5.5 1.5
BottomNotchLeft 5.342592593 1
25th 3 1
25th 3 4
Skip
Median 5.5 1.5
Median 5.5 6.5
Skip
75th 8 4
Max 10 4
Max 10 5
Max 10 3

The formula for X value of NotchLeft is
=MEDIAN($A$1:$A$50)-(1.7*((1.25*(PERCENTILE($A$1:$A$50,0.75)-PERCENTILE($A$*1:$A$50,0.25))/(1.35*50))))

The formula for X value of NotchRight is
=MEDIAN($A$1:$A$50)+(1.7*((1.25*(PERCENTILE($A$1:$ A$50,0.75)-PERCENTILE($A$*1:$A$50,0.25))/(1.35*50))))

All other formula are as William W. Dorner's example.

Cheers
Andy



wrote:
On Jun 15, 8:47 am, Andy Pope wrote:

Hi,

The technique describes how to use an xy-scatter to construct the boxes.

You need to add a few more xy pairs in order to reduce the width of the
Median line and form the notches.

Cheers
Andy

wrote:

Does anyone know how to create the notches in the verticalboxwhisker
charts?

An example can be found in figure 10 of the following article:

http://www.qualitydigest.com/oct97/html/excel.html

Thank you!!- Hide quoted text -

- Show quoted text -

Can you elaborate on the technique?- Hide quoted text -

- Show quoted text -

You ROCK!!!

On Sat, 16 Jun 2007, in microsoft.public.excel.charting,
Andy Pope said:

The formula for X value of NotchLeft is
=MEDIAN($A$1:$A$50)-(1.7*((1.25*(PERCENTILE($A$1:$A$50,0.75)-PERC
ENTILE($A$1:$A$50,0.25))/(1.35*50))))

The formula for X value of NotchRight is
=MEDIAN($A$1:$A$50)+(1.7*((1.25*(PERCENTILE($A$1: $A$50,0.75)-PERC
ENTILE($A$1:$A$50,0.25))/(1.35*50))))

Any particular reason for preferring
PERCENTILE(<range,0.75)-PERCENTILE(<range,0.25)
to
QUARTILE(<range,3)-QUARTILE(<range,1)
?

(I'm such a fond user of quartiles that I sometimes use them instead on
MIN, MAX and MEDIAN, because the five values are so simple to copy down
a column next to the numbers 0-4)

--
Del Cotter
NB Personal replies to this post will send email to ,
which goes to a spam folder-- please send your email to del3 instead.

Hi Del,

No preference just using the same formula as the example the OP was having
problems with.

Cheers
Andy

--

Andy Pope, Microsoft MVP - Excel
http://www.andypope.info
"Del Cotter" wrote in message
...
On Sat, 16 Jun 2007, in microsoft.public.excel.charting,
Andy Pope said:

The formula for X value of NotchLeft is
=MEDIAN($A$1:$A$50)-(1.7*((1.25*(PERCENTILE($A$1:$A$50,0.75)-PERC
ENTILE($A$1:$A$50,0.25))/(1.35*50))))

The formula for X value of NotchRight is
=MEDIAN($A$1:$A$50)+(1.7*((1.25*(PERCENTILE($A$1 :$A$50,0.75)-PERC
ENTILE($A$1:$A$50,0.25))/(1.35*50))))

Any particular reason for preferring
PERCENTILE(<range,0.75)-PERCENTILE(<range,0.25)
to
QUARTILE(<range,3)-QUARTILE(<range,1)
?

(I'm such a fond user of quartiles that I sometimes use them instead on
MIN, MAX and MEDIAN, because the five values are so simple to copy down
a column next to the numbers 0-4)

--
Del Cotter
NB Personal replies to this post will send email to
,
which goes to a spam folder-- please send your email to del3 instead.

On Fri, 15 Jun 2007, in microsoft.public.excel.charting,
Andy Pope said:
The technique describes how to use an xy-scatter to construct the boxes.

You need to add a few more xy pairs in order to reduce the width of
the Median line and form the notches.

While I have as much reverence for the late John Tukey as the next
person, I don't see that boxes and whiskers as such are necessary these
days, except that they're a familiar idiom that the graph viewer will
usually recognise.

And even that isn't true for notched boxes, which I don't think many
people have seen. Certainly most couldn't interpret without them a
guide; I never even knew until reading that article just now what the
notches were supposed to represent-- I thought they were just meant to
enphasise the median in some way.

If we abandon the need to copy Tukey's shapes, doing this stuff in Excel
immediately gets a lot easier. Here's my idea of a boxless "box" and
whisker distribution chart, with circled outliers and an error range
around the median, all just using the standard Excel symbol shapes.

http://i146.photobucket.com/albums/r264/del_c/
infographics/not_boxplot.gif

It would be simple to substitute circles, diamonds, or half-ticks, and
alter the thickness or colour of the Excel error bars, to suit your
preferences, and I think the point comes across even though they're not
the traditional boxes.

--
Del Cotter
NB Personal replies to this post will send email to ,
which goes to a spam folder-- please send your email to del3 instead.

Del -

Your chart allows plenty of different quantities to be shown, but I suspect
it may become cluttered, and at least for now, it's unfamiliar, and forces a
lot of back and forth between the chart and the legend. Don't knock a
"familiar idiom".

The box plot is pretty much self-explanatory especially since it is
familiar, and the difference between the box itself and the whiskers is
immediately recognizable (compared to your multiple error bars colored
different shades of gray, which is slower to be interpreted). If you could
make whiskers of various line lengths, that might help.

I agree that the notched box plot must be rather obscure, as I've never seen
it used in any real display of information.

- Jon
-------
Jon Peltier, Microsoft Excel MVP
Tutorials and Custom Solutions
Peltier Technical Services, Inc. - http://PeltierTech.com
_______


"Del Cotter" wrote in message
...
On Fri, 15 Jun 2007, in microsoft.public.excel.charting,
Andy Pope said:
The technique describes how to use an xy-scatter to construct the boxes.

You need to add a few more xy pairs in order to reduce the width of
the Median line and form the notches.

While I have as much reverence for the late John Tukey as the next
person, I don't see that boxes and whiskers as such are necessary these
days, except that they're a familiar idiom that the graph viewer will
usually recognise.

And even that isn't true for notched boxes, which I don't think many
people have seen. Certainly most couldn't interpret without them a
guide; I never even knew until reading that article just now what the
notches were supposed to represent-- I thought they were just meant to
enphasise the median in some way.

If we abandon the need to copy Tukey's shapes, doing this stuff in Excel
immediately gets a lot easier. Here's my idea of a boxless "box" and
whisker distribution chart, with circled outliers and an error range
around the median, all just using the standard Excel symbol shapes.

http://i146.photobucket.com/albums/r264/del_c/
infographics/not_boxplot.gif

It would be simple to substitute circles, diamonds, or half-ticks, and
alter the thickness or colour of the Excel error bars, to suit your
preferences, and I think the point comes across even though they're not
the traditional boxes.

--
Del Cotter
NB Personal replies to this post will send email to
,
which goes to a spam folder-- please send your email to del3 instead.

Duh, every now and then even proved experts (even with a degree in
statistics) say something that leaves me deeply perplexed.

Less than five minutes of googling reveals these five fine examples:

- Exhibit 1 (electronics / mobile phone manufacturing, 1996)
http://ieeexplore.ieee.org/iel3/4031...rnumber=561268

- Exhibit 2 (experimental / cognitive psychology, 1996)
http://faculty.washington.edu/jmiyam...f%20pref. pdf

- Exhibit 3 (physical anthropology / monkeys, 2002)
http://www.hopkinsmedicine.org/FAE/CBR2002AJPA.pdf

- Exhibit 4 (entomology / PhD thesis on honeybee parasites, 1994)
[I mean, I've published in truly numerous fields of medicine, psychology,
statistics, computer science, nuclear physics, physiotherapy, management,
phylosophy and more and what not, but is this topic exotic or what?!]
http://doc.rero.ch/lm.php?url=1000,4...se_RickliM.pdf

- Exhibit 5 (a physicist teaching maths presenting grade distribution at an
exam, 2006/7)
[trust me from plenty of experience with such people that being a physicist
and/or teaching mathematics otherwise tends to preclude knowledge and
understanding of statistics]
http://www.maths.qmul.ac.uk/~ob/MAS2...docs/stats.pdf

[if URLs are broken across lines, please put them together in your browser]

Note that I've selected only freely downloadable publications, while from an
academic institution with subscription to various online services from major
scientific publishers there are literaly dozens more readily available
examples!

Of course, your definition of "real display of information" might exclude
any kind of scientific or even technical publication, thus meaning only
"business" stuff and the general press. (Though I sincerly hope that it is
not what you meant.) In that case, my objection should be disregarded.

Regards,

Gaj Vidmar, PhD
Univ. of Ljubljana, Fac. of Medicine, Inst. of Biomedical Informatics

"Jon Peltier" wrote in message
...
Del -

Your chart allows plenty of different quantities to be shown, but I
suspect
it may become cluttered, and at least for now, it's unfamiliar, and forces
a
lot of back and forth between the chart and the legend. Don't knock a
"familiar idiom".

The box plot is pretty much self-explanatory especially since it is
familiar, and the difference between the box itself and the whiskers is
immediately recognizable (compared to your multiple error bars colored
different shades of gray, which is slower to be interpreted). If you could
make whiskers of various line lengths, that might help.

I agree that the notched box plot must be rather obscure, as I've never
seen
it used in any real display of information.

- Jon
-------
Jon Peltier, Microsoft Excel MVP
Tutorials and Custom Solutions
Peltier Technical Services, Inc. - http://PeltierTech.com
_______


"Del Cotter" wrote in message
...
On Fri, 15 Jun 2007, in microsoft.public.excel.charting,
Andy Pope said:
The technique describes how to use an xy-scatter to construct the boxes.

You need to add a few more xy pairs in order to reduce the width of
the Median line and form the notches.

While I have as much reverence for the late John Tukey as the next
person, I don't see that boxes and whiskers as such are necessary these
days, except that they're a familiar idiom that the graph viewer will
usually recognise.

And even that isn't true for notched boxes, which I don't think many
people have seen. Certainly most couldn't interpret without them a
guide; I never even knew until reading that article just now what the
notches were supposed to represent-- I thought they were just meant to
enphasise the median in some way.

If we abandon the need to copy Tukey's shapes, doing this stuff in Excel
immediately gets a lot easier. Here's my idea of a boxless "box" and
whisker distribution chart, with circled outliers and an error range
around the median, all just using the standard Excel symbol shapes.

http://i146.photobucket.com/albums/r264/del_c/
infographics/not_boxplot.gif

It would be simple to substitute circles, diamonds, or half-ticks, and
alter the thickness or colour of the Excel error bars, to suit your
preferences, and I think the point comes across even though they're not
the traditional boxes.

--
Del Cotter
NB Personal replies to this post will send email to
,
which goes to a spam folder-- please send your email to del3 instead.

Gaj -

I'm not an entomologist, nor have I read much on the anthropology of
monkeys. I've worked in scientific research as a metallurgist (for my
doctorate and a dozen years of employment following that), and as an
engineer in manufacturing. Maybe not the widest mathematical background, and
I'm not degreed in statistics (though I've taken a graduate level course or
three). I've encountered thousands of box and whisker charts and their
variants, but I've never seen a notched box chart used in the heat of
battle.

- Jon
-------
Jon Peltier, Microsoft Excel MVP
Tutorials and Custom Solutions
Peltier Technical Services, Inc. - http://PeltierTech.com
_______


"Gaj Vidmar" wrote in message
...
Duh, every now and then even proved experts (even with a degree in
statistics) say something that leaves me deeply perplexed.

Less than five minutes of googling reveals these five fine examples:

- Exhibit 1 (electronics / mobile phone manufacturing, 1996)
http://ieeexplore.ieee.org/iel3/4031...rnumber=561268

- Exhibit 2 (experimental / cognitive psychology, 1996)
http://faculty.washington.edu/jmiyam...f%20pref. pdf

- Exhibit 3 (physical anthropology / monkeys, 2002)
http://www.hopkinsmedicine.org/FAE/CBR2002AJPA.pdf

- Exhibit 4 (entomology / PhD thesis on honeybee parasites, 1994)
[I mean, I've published in truly numerous fields of medicine, psychology,
statistics, computer science, nuclear physics, physiotherapy, management,
phylosophy and more and what not, but is this topic exotic or what?!]
http://doc.rero.ch/lm.php?url=1000,4...se_RickliM.pdf

- Exhibit 5 (a physicist teaching maths presenting grade distribution at
an
exam, 2006/7)
[trust me from plenty of experience with such people that being a
physicist
and/or teaching mathematics otherwise tends to preclude knowledge and
understanding of statistics]
http://www.maths.qmul.ac.uk/~ob/MAS2...docs/stats.pdf

[if URLs are broken across lines, please put them together in your
browser]

Note that I've selected only freely downloadable publications, while from
an
academic institution with subscription to various online services from
major
scientific publishers there are literaly dozens more readily available
examples!

Of course, your definition of "real display of information" might exclude
any kind of scientific or even technical publication, thus meaning only
"business" stuff and the general press. (Though I sincerly hope that it is
not what you meant.) In that case, my objection should be disregarded.

Regards,

Gaj Vidmar, PhD
Univ. of Ljubljana, Fac. of Medicine, Inst. of Biomedical Informatics

"Jon Peltier" wrote in message
...
Del -

Your chart allows plenty of different quantities to be shown, but I
suspect
it may become cluttered, and at least for now, it's unfamiliar, and
forces
a
lot of back and forth between the chart and the legend. Don't knock a
"familiar idiom".

The box plot is pretty much self-explanatory especially since it is
familiar, and the difference between the box itself and the whiskers is
immediately recognizable (compared to your multiple error bars colored
different shades of gray, which is slower to be interpreted). If you
could
make whiskers of various line lengths, that might help.

I agree that the notched box plot must be rather obscure, as I've never
seen
it used in any real display of information.

- Jon
-------
Jon Peltier, Microsoft Excel MVP
Tutorials and Custom Solutions
Peltier Technical Services, Inc. - http://PeltierTech.com
_______


"Del Cotter" wrote in message
...
On Fri, 15 Jun 2007, in microsoft.public.excel.charting,
Andy Pope said:
The technique describes how to use an xy-scatter to construct the
boxes.

You need to add a few more xy pairs in order to reduce the width of
the Median line and form the notches.

While I have as much reverence for the late John Tukey as the next
person, I don't see that boxes and whiskers as such are necessary these
days, except that they're a familiar idiom that the graph viewer will
usually recognise.

And even that isn't true for notched boxes, which I don't think many
people have seen. Certainly most couldn't interpret without them a
guide; I never even knew until reading that article just now what the
notches were supposed to represent-- I thought they were just meant to
enphasise the median in some way.

If we abandon the need to copy Tukey's shapes, doing this stuff in
Excel
immediately gets a lot easier. Here's my idea of a boxless "box" and
whisker distribution chart, with circled outliers and an error range
around the median, all just using the standard Excel symbol shapes.

http://i146.photobucket.com/albums/r264/del_c/
infographics/not_boxplot.gif

It would be simple to substitute circles, diamonds, or half-ticks, and
alter the thickness or colour of the Excel error bars, to suit your
preferences, and I think the point comes across even though they're not
the traditional boxes.

--
Del Cotter
NB Personal replies to this post will send email to
,
which goes to a spam folder-- please send your email to del3 instead.

On Tue, 19 Jun 2007, in microsoft.public.excel.charting,
Jon Peltier said:
Your chart allows plenty of different quantities to be shown, but I suspect
it may become cluttered,

Possibly, but the example I showed was bound to look a little cluttered
compared to a simple pair of notched boxes, due to the sheer number of
data points on that graph, and my not taking the trouble to clean up the
legend. This version looks less cluttered, I hope.

http://i146.photobucket.com/
albums/r264/del_c/infographics/not_boxplot2.gif

and at least for now, it's unfamiliar, and forces a
lot of back and forth between the chart and the legend. Don't knock a
"familiar idiom".

Very true.

--
Del Cotter
NB Personal replies to this post will send email to ,
which goes to a spam folder-- please send your email to del3 instead.

If we are introducing different visuals for the error of the median, why not
retain the "familiar idiom" of the box chart, and merely add a visual to
represent the error. This could be a line across the bar (like your red
markers in either version of your chart) or some other type of marker. It
avoids reinvention of the entire wheel.

- Jon
-------
Jon Peltier, Microsoft Excel MVP
Tutorials and Custom Solutions
Peltier Technical Services, Inc. - http://PeltierTech.com
_______


"Del Cotter" wrote in message
...
On Tue, 19 Jun 2007, in microsoft.public.excel.charting,
Jon Peltier said:
Your chart allows plenty of different quantities to be shown, but I
suspect
it may become cluttered,

Possibly, but the example I showed was bound to look a little cluttered
compared to a simple pair of notched boxes, due to the sheer number of
data points on that graph, and my not taking the trouble to clean up the
legend. This version looks less cluttered, I hope.

http://i146.photobucket.com/
albums/r264/del_c/infographics/not_boxplot2.gif

and at least for now, it's unfamiliar, and forces a
lot of back and forth between the chart and the legend. Don't knock a
"familiar idiom".

Very true.

--
Del Cotter
NB Personal replies to this post will send email to
,
which goes to a spam folder-- please send your email to del3 instead.

On Jun 19, 8:26 am, "Jon Peltier"
wrote:
Del -

Yourchartallows plenty of different quantities to be shown, but I suspect
it may become cluttered, and at least for now, it's unfamiliar, and forces a
lot of back and forth between thechartand the legend. Don't knock a
"familiar idiom".

Theboxplot is pretty much self-explanatory especially since it is
familiar, and the difference between theboxitself and the whiskers is
immediately recognizable (compared to your multiple error bars colored
different shades of gray, which is slower to be interpreted). If you could
make whiskers of various line lengths, that might help.

I agree that the notchedboxplot must be rather obscure, as I've never seen
it used in any real display of information.

- Jon
-------
Jon Peltier,MicrosoftExcelMVP
Tutorials and Custom Solutions
Peltier Technical Services, Inc. -http://PeltierTech.com
_______

"Del Cotter" wrote in message

...



On Fri, 15 Jun 2007, inmicrosoft.public.excel.charting,
Andy Pope said:
The technique describes how to use an xy-scatter to construct the boxes.

You need to add a few more xy pairs in order to reduce the width of
the Median line and form the notches.

While I have as much reverence for the late John Tukey as the next
person, I don't see that boxes and whiskers as such are necessary these
days, except that they're a familiar idiom that the graph viewer will
usually recognise.

And even that isn't true for notched boxes, which I don't think many
people have seen. Certainly most couldn't interpret without them a
guide; I never even knew until reading that article just now what the
notches were supposed to represent-- I thought they were just meant to
enphasise the median in some way.

If we abandon the need to copy Tukey's shapes, doing this stuff inExcel
immediately gets a lot easier. Here's my idea of a boxless "box" and
whiskerdistributionchart, with circled outliers and an error range
around the median, all just using the standardExcelsymbol shapes.

http://i146.photobucket.com/albums/r264/del_c/
infographics/not_boxplot.gif

It would be simple to substitute circles, diamonds, or half-ticks, and
alter the thickness or colour of theExcelerror bars, to suit your
preferences, and I think the point comes across even though they're not
the traditional boxes.

--
Del Cotter
NB Personal replies to this post will send email to
,
which goes to a spam folder-- please send your email to del3 instead.- Hide quoted text -

- Show quoted text -

Wow!! I did not know my initial request generated such an exchange of
academic and technical passion. Just to let you know, I have decided
that if I need to generate a significant number of notched box whisker
charts, I am going to obtain the latest version of SigmaPlot (v 10
being the latest version) in order to generate them.

Jon, if your add-in can generate notched box whisker plots to the
scope and degree that SigmaPlot can, please let me know!!!

Bruce

Nope, I have not ever tried to generate notched box plots. There is no easy
way in an Excel chart to integrate a non-rectangular shape into the chart
series. I will give it further thought, of course, since other people find
it a more important feature than I thought it was.

If you want to show outliers as individual points outside the span of the
whiskers, stay tuned, because an unreleased version of the utility can
handle this.

- Jon
-------
Jon Peltier, Microsoft Excel MVP
Tutorials and Custom Solutions
Peltier Technical Services, Inc. - http://PeltierTech.com
_______


wrote in message
ps.com...
On Jun 19, 8:26 am, "Jon Peltier"
wrote:
Del -

Yourchartallows plenty of different quantities to be shown, but I suspect
it may become cluttered, and at least for now, it's unfamiliar, and
forces a
lot of back and forth between thechartand the legend. Don't knock a
"familiar idiom".

Theboxplot is pretty much self-explanatory especially since it is
familiar, and the difference between theboxitself and the whiskers is
immediately recognizable (compared to your multiple error bars colored
different shades of gray, which is slower to be interpreted). If you
could
make whiskers of various line lengths, that might help.

I agree that the notchedboxplot must be rather obscure, as I've never
seen
it used in any real display of information.

- Jon
-------
Jon Peltier,MicrosoftExcelMVP
Tutorials and Custom Solutions
Peltier Technical Services, Inc. -http://PeltierTech.com
_______

"Del Cotter" wrote in message

...



On Fri, 15 Jun 2007, inmicrosoft.public.excel.charting,
Andy Pope said:
The technique describes how to use an xy-scatter to construct the
boxes.

You need to add a few more xy pairs in order to reduce the width of
the Median line and form the notches.

While I have as much reverence for the late John Tukey as the next
person, I don't see that boxes and whiskers as such are necessary these
days, except that they're a familiar idiom that the graph viewer will
usually recognise.

And even that isn't true for notched boxes, which I don't think many
people have seen. Certainly most couldn't interpret without them a
guide; I never even knew until reading that article just now what the
notches were supposed to represent-- I thought they were just meant to
enphasise the median in some way.

If we abandon the need to copy Tukey's shapes, doing this stuff inExcel
immediately gets a lot easier. Here's my idea of a boxless "box" and
whiskerdistributionchart, with circled outliers and an error range
around the median, all just using the standardExcelsymbol shapes.

http://i146.photobucket.com/albums/r264/del_c/
infographics/not_boxplot.gif

It would be simple to substitute circles, diamonds, or half-ticks, and
alter the thickness or colour of theExcelerror bars, to suit your
preferences, and I think the point comes across even though they're not
the traditional boxes.

--
Del Cotter
NB Personal replies to this post will send email to
,
which goes to a spam folder-- please send your email to del3
instead.- Hide quoted text -

- Show quoted text -

Wow!! I did not know my initial request generated such an exchange of
academic and technical passion. Just to let you know, I have decided
that if I need to generate a significant number of notched box whisker
charts, I am going to obtain the latest version of SigmaPlot (v 10
being the latest version) in order to generate them.

Jon, if your add-in can generate notched box whisker plots to the
scope and degree that SigmaPlot can, please let me know!!!

Bruce