Whether those in a socioeconomic class have greater or lesser access to resources is of no relevance to whether a data set is ordinal or nominal in character. Whether data are nominal, ordinal, or interval/ratio data is determined by the nature of the data themselves. Nominal data are categories, as stated, ordinal data has order but no distance, while interval data is characterized by order and distance. The only difference between interval and ratio data is that a ratio data set has a real zero point. Interval data generally has a conventional zero point.
The locus classicus is S S Stevens, although his definitions are complex and involve mathematical transformations. Some introductory statistics books provide simpler definitions than Stevens does of these different data scale types, as Stevens calls them. I have defined them in the simplest way possible. How any given data scale type is used is determined by the theoretical context in which it is embedded.
A number of statistical texts differentiate among different statistical procedures in terms of whether they are appropriate to a given data scale type. For instance, Pearson's r assumes an interval data set, while Spearman's rho assumes an ordinal data set. Each of these two statistics is appropriate to a different data scale type. However, not all statistics texts arrange their statistical procedures in this way.
Whether ordinal data is treated as being nominal in character depends on the way it is being used. Effectively, you are not taking account of information that is contained in the data. But, depending on the theoretical context being used, that may not matter.
If the ONS contends that derived SEC data can't be regarded as ordinal but must be treated as nominal for analytical purposes, that depends on the nature of the derivation, or transformation, the data has undergone. Not all transformations alter the data's scale type. And that has to be what they are claiming for what they say to make any sense. What they should say is that if a transformation of a data set alters its scale type, that is, loses information that is contained in the original data, then one must use a data scale type analysis that does not depend on what has been lost. So, if ordered data, ordinal data, is transformed in such a way that the order is lost, then one must use a statistic where order is irrelevant, one designed for nominal data.
A good question to ask them is why they are transforming the data they have in this way, since there is no iron clad rule that says it must be done like this. I would suspect they have a ready answer.
Larry
[log in to unmask] type="cite">
Hello,
I have been looking at the documentation for the Office for National Statistics Socioeconomic Classification (
Rebasing the NS-SEC on SOC2010NS-SEC). On page 13, para 7.2 it advises that derived socioeconomic classification cannot be regarded as an ordinal scale but should be treated as categorical for purposes of analysis.
I was wondering if anyone could explain why that is. From a conceptual point of view, socioeconomic classification indicates access to resources. I have always assumed that socioeconomic status that is "higher" indicates access to more and more varied resources which is the reason why it is a good predictor of good health or education outcomes amongst others.
I have been looking for some literature on this (or an opposite view) but have so far been unsuccessful.
I wonder if anyone could help.
Thank you. ****************************************************** Please note that if you press the 'Reply' button your message will go only to the sender of this message. If you want to reply to the whole list, use your mailer's 'Reply-to-All' button to send your message automatically to
[log in to unmask]. Disclaimer: The messages sent to this list are the views of the sender and cannot be assumed to be representative of the range of views held by subscribers to the Radical Statistics Group. To find out more about Radical Statistics and its aims and activities and read current and past issues of our newsletter you are invited to visit our web site
http://www.radstats.org.uk/. *******************************************************