# Our tags for nominal, ordinal, and categorical data

Around a month ago, @gung suggested to make ×113 a synonym of ×1987 because they are "literally synonyms". This has initially gotten some support, until @ttnphns objected on the grounds that ×531 is also "categorical": nominal and ordinal, he says, are two sub-types of categorical data, and therefore we should leave all three tags separate. The whole discussion is here in the comments: https://stats.meta.stackexchange.com/a/4717/28666.

I feel that we need to arrive to some decision here.

In either case we will need to edit the wikis and wiki excerpts to clarify the usage. At the moment some information there is conflicting, e.g. wiki for says

categorical values are considered as abstract entities without any mathematical structure such as an order or a topology [written by @whuber in 2012]

but its wiki excerpt says

Some people consider ordinal scale categorical too [written by @ttnphns in 2013]

• Upvote this comment if you are in favor of gung's approach ([nominal]$\to$[categorical-data]). – amoeba says Reinstate Monica May 16 '17 at 12:50
• Upvote this comment if you are in favor of ttnphns's approach (keep three tags separate). – amoeba says Reinstate Monica May 16 '17 at 12:50
• Upvote this comment if you think both approaches are fine and/or don't care. – amoeba says Reinstate Monica May 16 '17 at 12:51
• I do not see any sin in keeping [nominal], [ordinal], and [categorical-data] three independent tags. As data types, nominal+ordinal <= categorical, where '<=' means that it would be possible perhaps to conceive of other categorical data types, some special/unusial, which do not easily drop into nominal or ordinal. Besides that, I somewhat doubt that we need to add "data" to [ordianal] and [nominal] tags; well, a question might be very abstract one about concepts of ordinal/nominal scales (measurements) rather than about some realization of it in a form of a data (analysis). – ttnphns May 17 '17 at 7:22
• @ttnphns I don't disagree. Regarding renaming, note that [ordinal] has already been renamed into [ordinal-data], after questions about ordinal logistic regression were moved out of it, see here: stats.meta.stackexchange.com/questions/4743. So now we have ordinal-logit ( = ordered-logit) and ordinal-data. I suggested to rename [nominal], because we have categorical-data, binary-data, now also ordinal-data, etc. so I thought it makes sense to have the tag name nominal-data too. IMHO a question about nominal scales can be tagged with [nominal-data] without any big problem. – amoeba says Reinstate Monica May 17 '17 at 8:05
• Amoeba, because you are much more active a meta user/fixer I recognise your priority. Do how you find it worth in the end. My task is to drop here my opinion, among others', for you to collect considerations. – ttnphns May 17 '17 at 9:16
• Are there any guidelines about the tags? How do they actually function in practice? (I know that they are linked to badges and privileges, but beyond that, is there something useful in searching?) – Sextus Empiricus Apr 23 '18 at 11:45
• @MartijnWeterings Sorry I don't quite understand what you are asking. Maybe it's better if you ask this in more details either as a new Meta post or in chat. – amoeba says Reinstate Monica Apr 23 '18 at 12:20
• @amoeba I am considering to join voting on one of the three comments. But, somewhere between the moment of starting this consideration and putting the actual vote I realized that, I do not know what the (practical) function is of these tags. (and now I see this is an old question, with just a new answer) – Sextus Empiricus Apr 23 '18 at 15:40
• @MartijnWeterings Yes, this particular voting is not relevant anymore; the merge has been made. – amoeba says Reinstate Monica Apr 23 '18 at 15:48

## 1 Answer

Meanwhile, @ttnphns has changed his opinion (see here) and following that, @gung approved the pending synonym suggestion.

became a synonym of .