*John G. Orme and Terri Combs-Orme*

- Published in print:
- 2009
- Published Online:
- May 2009
- ISBN:
- 9780195329452
- eISBN:
- 9780199864812
- Item type:
- chapter

- Publisher:
- Oxford University Press
- DOI:
- 10.1093/acprof:oso/9780195329452.003.0004
- Subject:
- Social Work, Research and Evaluation

This chapter discusses ordinal logistic regression (also known as the ordinal logit, ordered polytomous logit, constrained cumulative logit, proportional odds, parallel regression, or grouped ...
More

This chapter discusses ordinal logistic regression (also known as the ordinal logit, ordered polytomous logit, constrained cumulative logit, proportional odds, parallel regression, or grouped continuous model), for modeling relationships between an ordinal dependent variable and multiple independent variables. Ordinal variables have three or more ordered categories, and ordinal logistic regression focuses on cumulative probabilities of the dependent variable and odds and odds ratios based on those cumulative probabilities, estimating a single common odds ratio. The chapter discusses the proportional odds or parallel regression assumption; this is the assumption that the odds ratios for each cumulative level are equal in the population (although they might be different in a sample due to sampling error). The concepts of threshold, sometimes called a cut-point, proportional odds or parallel regression assumption, are also discussed.Less

This chapter discusses ordinal logistic regression (also known as the ordinal logit, ordered polytomous logit, constrained cumulative logit, proportional odds, parallel regression, or grouped continuous model), for modeling relationships between an ordinal dependent variable and multiple independent variables. Ordinal variables have three or more ordered categories, and ordinal logistic regression focuses on cumulative probabilities of the dependent variable and odds and odds ratios based on those cumulative probabilities, estimating a single common odds ratio. The chapter discusses the proportional odds or parallel regression assumption; this is the assumption that the odds ratios for each cumulative level are equal in the population (although they might be different in a sample due to sampling error). The concepts of *threshold*, sometimes called a *cut*-*point*, *proportional odds* or *parallel regression assumption*, are also discussed.

*Christine DeMars*

- Published in print:
- 2010
- Published Online:
- March 2012
- ISBN:
- 9780195377033
- eISBN:
- 9780199847341
- Item type:
- chapter

- Publisher:
- Oxford University Press
- DOI:
- 10.1093/acprof:oso/9780195377033.003.0001
- Subject:
- Psychology, Cognitive Psychology

Item response theory (IRT) models show the relationship between the ability or trait (θ) measured by the instrument and an item response. Both the IRT and classical test theory (CTT) indices and how ...
More

Item response theory (IRT) models show the relationship between the ability or trait (θ) measured by the instrument and an item response. Both the IRT and classical test theory (CTT) indices and how they are related in terms of item parameters, conceptualization of reliability and standard error of measurement, and population invariance are described. The IRT indices are more readily understood in the context of the formal mathematical models for describing the item response probabilities, in which these models are discussed. The IRT indices are introduced only in general terms. The three-parameter logistic (3PL), two-parameter logistic (2PL), and one-parameter logistic (1PL) dichotomous models are presented, along with the Graded Response (GR) and Generalized Partial Credit (GPC) polytomous models.Less

Item response theory (IRT) models show the relationship between the ability or trait (θ) measured by the instrument and an item response. Both the IRT and classical test theory (CTT) indices and how they are related in terms of item parameters, conceptualization of reliability and standard error of measurement, and population invariance are described. The IRT indices are more readily understood in the context of the formal mathematical models for describing the item response probabilities, in which these models are discussed. The IRT indices are introduced only in general terms. The three-parameter logistic (3PL), two-parameter logistic (2PL), and one-parameter logistic (1PL) dichotomous models are presented, along with the Graded Response (GR) and Generalized Partial Credit (GPC) polytomous models.