On ordinal prediction problems

Lawrence S. Mayer, I. J. Good

Research output: Contribution to journalArticle

1 Citation (Scopus)

Abstract

One limitation in building empirically testable models in sociology is that many familiar statistical techniques such as least-squares regression analysis require interval-level measurements while sociologists often have only ordinal-level measurements. Fortunately there do exist statistical techniques that use ordinal measurements. In this article we consider several types of prediction procedures which use ordinal data, as input, including individual ordinal prediction procedures and pairwise ordinal prediction procedures. The latter were studied by Wilson (1971) who claimed that ordinal variables could neither be used to build empirically testable models nor to state substantive propositions rigorously. But his claims are weakened because (1) he states, but fails to prove, that a particular loss function is the only one that can be used in pairwise ordinal prediction procedures, (2) he ignores alternative types of ordinal prediction procedures, and (3) his main mathematical theorem is in error. We consider his arguments, salvage his theorem, and display a similar theorem for individual ordinal prediction procedures. We argue that, for the three reasons mentioned, Wilson’s results do not show that it is unprofitable to use ordinal variables in prediction procedures. Finally we consider a generalized individual ordinal prediction procedure in which one ordinal variable is only useless for predicting a second ordinal variable if the two variables are statistically independent.

Original languageEnglish (US)
Pages (from-to)543-549
Number of pages7
JournalSocial Forces
Volume52
Issue number4
DOIs
StatePublished - 1974
Externally publishedYes

Fingerprint

Prediction
sociologist
regression analysis
sociology
Sociologists
Sociology
Regression Analysis
Salvage

ASJC Scopus subject areas

  • Sociology and Political Science
  • Anthropology
  • History

Cite this

Mayer, L. S., & Good, I. J. (1974). On ordinal prediction problems. Social Forces, 52(4), 543-549. https://doi.org/10.1093/sf/52.4.543

On ordinal prediction problems. / Mayer, Lawrence S.; Good, I. J.

In: Social Forces, Vol. 52, No. 4, 1974, p. 543-549.

Research output: Contribution to journalArticle

Mayer, LS & Good, IJ 1974, 'On ordinal prediction problems', Social Forces, vol. 52, no. 4, pp. 543-549. https://doi.org/10.1093/sf/52.4.543
Mayer, Lawrence S. ; Good, I. J. / On ordinal prediction problems. In: Social Forces. 1974 ; Vol. 52, No. 4. pp. 543-549.
@article{5ebbd0698b164fb9ae99cf7356df7d7b,
title = "On ordinal prediction problems",
abstract = "One limitation in building empirically testable models in sociology is that many familiar statistical techniques such as least-squares regression analysis require interval-level measurements while sociologists often have only ordinal-level measurements. Fortunately there do exist statistical techniques that use ordinal measurements. In this article we consider several types of prediction procedures which use ordinal data, as input, including individual ordinal prediction procedures and pairwise ordinal prediction procedures. The latter were studied by Wilson (1971) who claimed that ordinal variables could neither be used to build empirically testable models nor to state substantive propositions rigorously. But his claims are weakened because (1) he states, but fails to prove, that a particular loss function is the only one that can be used in pairwise ordinal prediction procedures, (2) he ignores alternative types of ordinal prediction procedures, and (3) his main mathematical theorem is in error. We consider his arguments, salvage his theorem, and display a similar theorem for individual ordinal prediction procedures. We argue that, for the three reasons mentioned, Wilson’s results do not show that it is unprofitable to use ordinal variables in prediction procedures. Finally we consider a generalized individual ordinal prediction procedure in which one ordinal variable is only useless for predicting a second ordinal variable if the two variables are statistically independent.",
author = "Mayer, {Lawrence S.} and Good, {I. J.}",
year = "1974",
doi = "10.1093/sf/52.4.543",
language = "English (US)",
volume = "52",
pages = "543--549",
journal = "Social Forces",
issn = "0037-7732",
publisher = "University of North Carolina Press",
number = "4",

}

TY - JOUR

T1 - On ordinal prediction problems

AU - Mayer, Lawrence S.

AU - Good, I. J.

PY - 1974

Y1 - 1974

N2 - One limitation in building empirically testable models in sociology is that many familiar statistical techniques such as least-squares regression analysis require interval-level measurements while sociologists often have only ordinal-level measurements. Fortunately there do exist statistical techniques that use ordinal measurements. In this article we consider several types of prediction procedures which use ordinal data, as input, including individual ordinal prediction procedures and pairwise ordinal prediction procedures. The latter were studied by Wilson (1971) who claimed that ordinal variables could neither be used to build empirically testable models nor to state substantive propositions rigorously. But his claims are weakened because (1) he states, but fails to prove, that a particular loss function is the only one that can be used in pairwise ordinal prediction procedures, (2) he ignores alternative types of ordinal prediction procedures, and (3) his main mathematical theorem is in error. We consider his arguments, salvage his theorem, and display a similar theorem for individual ordinal prediction procedures. We argue that, for the three reasons mentioned, Wilson’s results do not show that it is unprofitable to use ordinal variables in prediction procedures. Finally we consider a generalized individual ordinal prediction procedure in which one ordinal variable is only useless for predicting a second ordinal variable if the two variables are statistically independent.

AB - One limitation in building empirically testable models in sociology is that many familiar statistical techniques such as least-squares regression analysis require interval-level measurements while sociologists often have only ordinal-level measurements. Fortunately there do exist statistical techniques that use ordinal measurements. In this article we consider several types of prediction procedures which use ordinal data, as input, including individual ordinal prediction procedures and pairwise ordinal prediction procedures. The latter were studied by Wilson (1971) who claimed that ordinal variables could neither be used to build empirically testable models nor to state substantive propositions rigorously. But his claims are weakened because (1) he states, but fails to prove, that a particular loss function is the only one that can be used in pairwise ordinal prediction procedures, (2) he ignores alternative types of ordinal prediction procedures, and (3) his main mathematical theorem is in error. We consider his arguments, salvage his theorem, and display a similar theorem for individual ordinal prediction procedures. We argue that, for the three reasons mentioned, Wilson’s results do not show that it is unprofitable to use ordinal variables in prediction procedures. Finally we consider a generalized individual ordinal prediction procedure in which one ordinal variable is only useless for predicting a second ordinal variable if the two variables are statistically independent.

UR - http://www.scopus.com/inward/record.url?scp=84959796145&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=84959796145&partnerID=8YFLogxK

U2 - 10.1093/sf/52.4.543

DO - 10.1093/sf/52.4.543

M3 - Article

AN - SCOPUS:84959796145

VL - 52

SP - 543

EP - 549

JO - Social Forces

JF - Social Forces

SN - 0037-7732

IS - 4

ER -