Statistics - Box plots and whiskers

[Whickerman]

When mapping box plots with whiskers, can the upper and lower values exceed the actual data range? ie is the calculation 1.5 times the interquartile range plus/minus the 25th/75th percentiles limited by the data range?

[factor-fiction]

Eh?
The Box and Whisker diagram shows the minimum and maximum values, the median, and the upper and lower quartile values. The upper quartile (75th percentile) cannot by definition be above themaximum value, and the lower quartile can't be below the minimum.

[factor-fiction]

Have a look here.
http://en.wikipedia.org/wiki/Box_plot
I think you are referring to outliers .

[Whickerman]

thanks but I'm referring to whiskers, which can also be mapped on the graph with the box plot

Whiskers are calculated as 1.5 time the interquartile range, then added to the 75th percentile and taken away from the 25th percentile.
Where the data range is wide it means the whiskers can technically be wider than the data range, so there are no outliers in that case?

[Whickerman]

example here:
http://www.statcan.ca/english/edu/power/ch12/p lots.htm

the case i'm dealing with is odd, as the whiskers are wider than the data range

[dundurn]

I'm sorry, but the example you've just quoted makes no mention of multiplying anything by 1.5. The 5 figure plot used to get the box plus whiskers is simply the lowest value, q1, median, q3 and highest value. The concept of a whisker extending beyond the value range just doesn't occur in this example.

[factor-fiction]

Hello Dundurn,
I'm not too familiar with Box and Whisker plots as they weren't used when i was at school but I've seen my kids do them.
Yes, normally the box plots the upper quartile, median and lower quartile and the whisker (doesn't look likea whisker to me) shows the maximum and minimum of the range of data. That's as far as it goes with GCSE maths.
But if you look on the wikipedia page and other sites it seems the format is adjusted when there are outliers which don't fit in with the rest of the data. If any values lie outside a certain range (75th percentile plus 1.5 x interquartile range ) then they are considered to be outliers and are ignored for the purpose of the whisker diagram.
But I'm still not sure what Whickerman is saying as he seems to be saying I've missed the point. Whickerman is I think saying the whiskers can be wider than the overall data range, bu this only makes sense if his data range excludes the outliers.

[factor-fiction]

Correction- I don't think Whiskerman's point can EVER be true. The whiskers can't be longer than the data range

[Whickerman]

Thanks all - appreciate your help