Can you convert ordinal data into interval data? Why, or why not?

3858
2
10-25-2016 07:34 AM
nolumanyanocamagu
New Contributor II

Can you convert ordinal data into interval data? Why, or why not?

0 Kudos
2 Replies
DanPatterson_Retired
MVP Emeritus

Ordinal data are data that are ranked.  The 'spacing' between the ranks is meaningless, ie, rank 1 isn't twice as good as rank 2.  Rank 1, just is based on some criteria

Interval data are data which are measured using some scale, where the spacing between the values has meaning.  For example, degree celcius... the spacing between 10 and 11 degrees is the same as between 11 and 12 etc.  The zero point is arbitrary (ie degrees celcius, F .... elevation, etc etc)

Ratio data are like interval data, except they have a real zero point (degrees Kelvin, Rankin, etc)

So to back to your data question, it is not possible to convert ranks back to interval data unless you knew how the data were classified and what criteria determined the threshold to go from one rank to another.

If this is known, then you can get an idea

Rank  some variable

1         10 something +

2          5  something to < 10

3          2  something to < 5

4          < 4

But I would caution taking it back beyond that level.  I have all too often seen people trying to calculate the "average rank"... skip it... just record the results

Rank  count

1         5

2         10

3         3

4         1

All you can say is rank 2 is kindof sort of the most popular.  I will spare you more on this topic and the potential misuse of numbers...

nolumanyanocamagu
New Contributor II

   Thank you very much !