Hi!
I have a problem that I solved long ago, but now I don't remember
how I did it! I have some grouped data, the number of children saw
by GPs in a year. The possible answers are for the number of
children that each GP saw are: 0-5, 6-10, 11-15 and >15.
Hence, what I have is the number of GP that say between 0-5 children,
6-10 children and so on.
I want to estimate the total number of children that went to
a GP in a year. A possibility, suggested by "Statistical Methods"
by Snedecor and Cochran, is to take the middle point of the each
category (16 in the last category) and estimate the total number
of children. However, there must be a more sensible way.
For example, if this case happens
Category N
0-5 5
6-10 90
11-15 100
surely the middle point is not sensible for the 0-5 category (nor
for the 6-10 one) since these five GP that tick the 0-5 category
are closer to 5 than to 0. Is there better way then?! What to do with
the extreme category!? Use a conservative approach and chose 16
as the multiplying factor? As I remember I used some graphical
display (sorry for the poor display, but it was the best I
could get)
2
|---|
| \/| 3
| /\|---| The number represent the categories
1 |/ || | 1 : 0-5
|---| || | 2 : 6-10
| | || | 3 : 11-15
| | || |
-------*----------
0 5 |10 15
|
>7.5 (middle point bigger than 7.5)
Is this approach correct and was it mentioned in the literature!?
Thanks for all the help
Miguel Goncalves
Institute of Hearing Research
Nottingham
UK
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
