### Citations

905 |
Spectral Analysis and Time Series.
- Priestley
- 1994
(Show Context)
Citation Context ... between simulations, and an estimate which accounts for the autocorrelations by including a weighted sum of them up to some fixed lag (see Geyer 1992); specifically, the Parzen window function (e.g. =-=Priestley, 1981-=-, p. 443) was used for the weighting. The effective sample sizes were then estimated by dividing the number of simulations by the ratio of these two standard errors. The effective number of simulation... |

381 |
Statistical methods in cancer research. Vol. I. The analysis of case-control studies,
- BRESLOW, DAY
- 1980
(Show Context)
Citation Context ...Y22 is a constant given S = S1 = (N,Y2+,Y+2) ′, the distribution of ˆ ∆ is determined by the distribution of Y22 given both margins of the table. This is an extended hypergeometric distribution (e.g. =-=Breslow and Day 1980-=-, sec. 4.3; Cox and Snell 1989, sec. 2.3). In all but very small samples, it is well approximated by a normal distribution with mean µ22 and variance σ2 y = ( ∑ µ −1 ij )−1. Since ˆπ22 converges to π∗... |

337 | Measures of association for cross-classifications. - Goodman, Kruskal - 1954 |

311 |
An Introduction to Categorical Data Analysis, 2nd ed.;
- Agresti
- 2007
(Show Context)
Citation Context ...be a more important criterion than formal goodness of fit. In the context of models for categorical data a commonly used statistic is the so-called index of dissimilarity or dissimilarity index (e.g. =-=Agresti, 2002-=-, pp. 329–330), which aims to quantify lack of model fit by estimating the smallest fraction of the population under study that would need to be re-classified in order to make the fitted model exactly... |

300 | The Constant Flux: A Study of Class Mobility in Industrial Societies. - Erikson, Goldthorpe - 1992 |

260 |
Theory of statistical estimation.
- Fisher
- 1925
(Show Context)
Citation Context ...more convenient way to describe the distribution of the cell counts Y is to treat them as independent Poisson random variables with means µ = (µ1,...,µK) ′ = Nπ, subject to the constraint ∑ i µi = N (=-=Fisher, 1925-=-; Birch, 1963). Here the two formulations will be used interchangeably. Even when the sample size is actually random and the Poisson thus the appropriate distribution, a standardization argument given... |

216 |
Applied Latent Class Analysis
- Hagenaars, McCutcheon
- 2002
(Show Context)
Citation Context ...and β 3 are parameter vectors; a general expression which also covers RC(m) models is a straightforward extension. Another broad class of models for categorical data are latent class models (see e.g. =-=Hagenaars and McCutcheon 2002-=-). Our general variance results hold also for them, but specific expressions are not given here. The discussion of the previous section suggests that the properties of ˆ ∆ as an estimator of ∆ for a l... |

215 |
An Introduction to Probability Theory and Its Applications (2nd ed)”,
- Feller
- 1957
(Show Context)
Citation Context ... 2 2CRiSM Paper No. 09-26, www.warwick.ac.uk/go/crism which is consistently estimated by ˆ ∆. This is the total variation distance between the two discrete distributions with probabilities π and π∗ (=-=Feller 1971-=-; the definition there is for 2∆). It can also be interpreted as ∆ = supA⊂K |π(A) − π∗ (A)| where π(A) = ∑ i∈A πi, π∗ (A) = ∑ i∈A π∗ i and K = {1,2,...,K}. The population value ∆ depends on the true p... |

177 | A methodological analysis of segregation indexes,” - Duncan, Duncan - 1955 |

165 |
Inference and Asymptotics.
- Barndorff-Nielsen, Cox
- 1994
(Show Context)
Citation Context ...rization ψ = (η′ 1,θ ′ 2) ′. The relationships between the parametrizations are one-to-one when all µi are positive. Furthermore, η1 and θ2 are variation-independent and orthogonal (see, for example, =-=Barndorff-Nielsen and Cox 1994-=-). The most common class of models for contingency tables are loglinear models of the form log µi = X ′ 1iθ1, (5) obtained by setting θ2 = 0 in (4). The model is assumed to be hierarchical, so that fo... |

125 |
Practical Markov chain Monte Carlo (with discussion).
- Geyer
- 1992
(Show Context)
Citation Context ...ion mean of ˆ ∆, a naive estimate which assumes independence between simulations, and an estimate which accounts for the autocorrelations by including a weighted sum of them up to some fixed lag (see =-=Geyer 1992-=-); specifically, the Parzen window function (e.g. Priestley, 1981, p. 443) was used for the weighting. The effective sample sizes were then estimated by dividing the number of simulations by the ratio... |

86 | Intergenerational class mobility in three Western European societies. - Erikson, Goldthorpe, et al. - 1979 |

63 |
Maximum likelihood in three-way contingency tables
- Birch
- 1963
(Show Context)
Citation Context ...t way to describe the distribution of the cell counts Y is to treat them as independent Poisson random variables with means µ = (µ1,...,µK) ′ = Nπ, subject to the constraint ∑ i µi = N (Fisher, 1925; =-=Birch, 1963-=-). Here the two formulations will be used interchangeably. Even when the sample size is actually random and the Poisson thus the appropriate distribution, a standardization argument given in Section 3... |

60 |
The analysis of cross-classified data having ordered and/or unordered 20 Biometrics,
- Goodman
- 1985
(Show Context)
Citation Context ...(5). In particular, they apply also to log-multiplicative models which aim to simplify interactions using bilinear structures: examples are the well-known Goodman RC and RC(m) association models (see =-=Goodman, 1985-=-) and the uniform difference (unidiff) models of Erikson and Goldthorpe (1992) and Xie (1992). The RC and unidiff models are both of the form log µi = X ′ 1iθ1 + (1 + U ′ 1iβ 2)(U ′ 2iβ 3) (9) where U... |

46 | Segregation and Diversity Measures in Population Distribution. - White - 1986 |

41 |
Analysis of Binary Data, 2nd Edition.
- Cox, Snell
- 1989
(Show Context)
Citation Context ... (N,Y2+,Y+2) ′, the distribution of ˆ ∆ is determined by the distribution of Y22 given both margins of the table. This is an extended hypergeometric distribution (e.g. Breslow and Day 1980, sec. 4.3; =-=Cox and Snell 1989-=-, sec. 2.3). In all but very small samples, it is well approximated by a normal distribution with mean µ22 and variance σ2 y = ( ∑ µ −1 ij )−1. Since ˆπ22 converges to π∗ 22, the conditional distribut... |

33 | Further Considerations on the Methodological Analysis of Segregation Indices. - Cortese, Cohen, et al. - 1976 |

31 | Measures of Association for Cross Classifications, II: Further Discussions and References - Goodman, Kruskal - 1959 |

23 | Monte carlo exact conditional tests for log-linear and logistic models - Forster, McDonald, et al. - 1996 |

23 |
Models for contingency tables with known margins when target and sampled populations differ
- Little, Wu
- 1991
(Show Context)
Citation Context ...isfies the conditions X ′ 0p0 = S0 (7) W ′ logp0 = W ′ logp, (8) where S0 is a given set of non-zero marginal proportions and log denotes the element-wise logarithmic operator (see Imrey et al. 1981; =-=Little and Wu 1991-=-). The matrix X0 is a design matrix corresponding to the margins included in S0; here we take X0 = X1 so that NS0 contains the same margins as S1, but this will be relaxed in the next Section. The mat... |

15 | Analysis of dispersion of multinomial responses. - Haberman - 1982 |

11 |
Categorical data analysis: some reflections on the log linear model and logistic regression. Part II: data analysis.
- Imrey, Koch, et al.
- 1982
(Show Context)
Citation Context ... proportions p0 satisfies the conditions X ′ 0p0 = S0 (7) W ′ logp0 = W ′ logp, (8) where S0 is a given set of non-zero marginal proportions and log denotes the element-wise logarithmic operator (see =-=Imrey et al. 1981-=-; Little and Wu 1991). The matrix X0 is a design matrix corresponding to the margins included in S0; here we take X0 = X1 so that NS0 contains the same margins as S1, but this will be relaxed in the n... |

11 |
Sexual harassment as a gendered expression of power.
- Uggen, Blackstone
- 2004
(Show Context)
Citation Context ...issimilarity for model assessment, especially in social science where the computed statistic is sometimes referred to simply as ‘Delta’ (e.g. Erola and Moisio 2007; Jonsson et al. 2009; Pfeffer 2008; =-=Uggen and Blackstone 2004-=-; Wells et al. 2003). The Delta statistic is used as a supplement to, rather than a replacement for, model-selection criteria such as those based on the log likelihood. The present work explores the d... |

10 |
Persistent Inequality in Educational Attainment and Its Institutional Context.’’ European Sociological Review 24:543–65.
- Pfeffer
- 2008
(Show Context)
Citation Context ...the index of dissimilarity for model assessment, especially in social science where the computed statistic is sometimes referred to simply as ‘Delta’ (e.g. Erola and Moisio 2007; Jonsson et al. 2009; =-=Pfeffer 2008-=-; Uggen and Blackstone 2004; Wells et al. 2003). The Delta statistic is used as a supplement to, rather than a replacement for, model-selection criteria such as those based on the log likelihood. The ... |

8 |
Social mobility over three generations in
- Erola, Moisio
- 2007
(Show Context)
Citation Context ...aling properties have led to routine use of the index of dissimilarity for model assessment, especially in social science where the computed statistic is sometimes referred to simply as ‘Delta’ (e.g. =-=Erola and Moisio 2007-=-; Jonsson et al. 2009; Pfeffer 2008; Uggen and Blackstone 2004; Wells et al. 2003). The Delta statistic is used as a supplement to, rather than a replacement for, model-selection criteria such as thos... |

7 | Ox: An Object-Oriented Matrix Language (4th edition - Doornik - 2001 |

7 | The measurement of ecological segregation - Jahn, Schmid, et al. - 1947 |

7 | Sampling Distributions of Segregation Indexes - Ransom |

7 | What happens after the high school years among young persons with disabilities? Social Forces
- Wells, Sandefur, et al.
- 2003
(Show Context)
Citation Context ...ssment, especially in social science where the computed statistic is sometimes referred to simply as ‘Delta’ (e.g. Erola and Moisio 2007; Jonsson et al. 2009; Pfeffer 2008; Uggen and Blackstone 2004; =-=Wells et al. 2003-=-). The Delta statistic is used as a supplement to, rather than a replacement for, model-selection criteria such as those based on the log likelihood. The present work explores the definition and estim... |

2 | Di una misura della dissomiliglianza tra due gruppi di quantitá e delle sue applicazioni allo studio delle relazioni statistiche. Atti del Reale Instituto Veneto di Scienze, Lettere ed Arti (Series 8 - Gini - 1914 |

2 | Approximations to the mean and variance of the index of dissimilarity in 2×C tables under a random allocation model - Inman, Bradley - 1991 |

2 | How useful are approximations to mean and variance of the index of dissimilarity? - Mulekar, Knutson, et al. - 2008 |

1 | Micro-class mobility: Social reproduction in four countries - Carlo, Pollak, et al. - 2009 |