Hi,
Could this be a job for the bihistogram? There was a discussion on this list
earlier. Austin Nichols did post a code to create one.
hth
Rajesh
-----Original Message-----
From: [email protected]
[mailto:[email protected]] On Behalf Of Nick Cox
Sent: 29 September 2008 21:02
To: [email protected]
Subject: Re: st: overlapping histograms
Do you mean something like this?
sysuse auto
twoway histogram mpg if foreign, start(10) width(2) bfcolor(none)
blcolor(red)
|| histogram mpg if !foreign, start(10) width(2) bfcolor(none)
blcolor(navy) barw(1.9) legend(off)
Jeph Herrin wrote:
>
> Yes, a colleague specifically wants histograms, and he knows
> his audience, presumably. More pointedly, he also has a
> figure just like he wants, from an unknown source, which is
> clearly created in Stata. However, the solution there is to
> narrow the bars in the top histogram, while leaving those in
> the back full width. This works well visually - the eye is
> tricked into seeing both distributions at once, despite a great
> deal of overlap, except I'm not happy that the bars are not
> the same width. So I'm trying to improve on it, while sticking
> close to what it is.
>
> Thanks to others for their suggestions.
>
> cheers,
> Jeph
>
>
>
>
>
> Nick Cox wrote:
>> Jeph may be acting under orders, i.e. someone is insisting on
>> something like this, but the problem that one obscures the other
>> presumably arises because your distributions are not too different and
>> is likely to persist with any small modifications to the recipe.
>>
>> In addition to other solutions suggested, -qqplot-, -qplot- (SJ),
>> -distplot- (SJ), -stripplot- (SSC), -beamplot- (SSC) are some commands
>> I would use to compare distributions graphically long before I tried
>> to superimpose histograms.
>>
>> Nick
>> [email protected]
>>
>> Jeph Herrin wrote:
>>>
>>> I've been trying to create a graph which shows
>>> two overlapping histograms - I'm trying to show
>>> how the distribution for a quantity if interest
>>> has shifted from time A to time B.
>>>
>>> Putting two histograms on one graph is not the
>>> problem
>>>
>>> twoway (histogram myvarA) (histogram myvarB)
>>>
>>>
>>> but the problem is that one obscures the other. I
>>> can make the one on top "clear" (specify -color(none)-)
>>> but because the bars line up it still difficult to
>>> distinguish them. So I'm looking for a way to do
>>> either
>>>
>>> 1. make the overlapping areas a darker color than
>>> either histogram is alone
>>> 2. shift the bars for one histogram so that the one
>>> behind can be seen ever so slightly
>>>
>>> Is there a way to do either? Is there a better solution?
>>
>> *
>> * For searches and help try:
>> * http://www.stata.com/help.cgi?search
>> * http://www.stata.com/support/statalist/faq
>> * http://www.ats.ucla.edu/stat/stata/
>>
> *
> * For searches and help try:
> * http://www.stata.com/help.cgi?search
> * http://www.stata.com/support/statalist/faq
> * http://www.ats.ucla.edu/stat/stata/
>
*
* For searches and help try:
* http://www.stata.com/help.cgi?search
* http://www.stata.com/support/statalist/faq
* http://www.ats.ucla.edu/stat/stata/
*
* For searches and help try:
* http://www.stata.com/help.cgi?search
* http://www.stata.com/support/statalist/faq
* http://www.ats.ucla.edu/stat/stata/
