Hi everyone,
I am trying to determine what scale of measurement to give to this variable: On how many days of the week do you discuss Politics with people?
The answers that people can give are: no days; one day; two days; three days; four days; five days; six days; seven days.
Initially I thought it was a ratio variable (on spss) as its in relation to time, but I tried to carry out a scatterplot with age, and the distribution comes up in 7 different lines. Im wondering if it needs to be categorised as a ordinal variable? If anyone knows though that would be very helpful!
Many thanks,
Marnie
Marnie, as you think, it is a ratio scale variable. But it is also discrete because it can take only integer values, and bounded, because its values are limited to 0, 1, ... 7. Here are other examples: Number of children in a family: discrete ratio scale, left bounded by 0 or the number of credit cards that a client can have from a bank if it has established a limit of 5 (for example).
You have a probability mass function f(x) = Pr(X = x) for x = 0, 1, 2, ... 7 and f(x) = 0 elsewhere. To that point, it is the same as a multinomial probability function, but with the order, it has a discrete distribution function given by F(x) = Pr(X
Jorge is right. It definitively ratio scale. You have defined intervals (1 day differnce is one day difference, this has the same meaning on any number of days), and you have a real zero point (0 days means null, nothing, absence).
That the variable is discrete and bounded has nothing to do with the measurement scale.
A possible probability model forsuch a variable is the beta-binomial distribution.
https://en.wikipedia.org/wiki/Beta-binomial_distribution
Thank you Jochen Wilhelm, indeed, the beta-binomial is an excellent proposal as a probability function for this variable.
- Badges
- Science topic
- Similar topics
- Mathematical Sciences
- Statistics
- Statistical Software
- SPSS