Logistic Regression, Categorical Predictors, and Goodness-of-Fit: It Depends on Who You Ask

Simonoff, Jeffrey S.

doi:10.1080/00031305.1998.10480529

Cited by 16 publications

(9 citation statements)

References 2 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…It takes the form (SimonoOE 1998) of: G 2 5 2 2 [(maximized log ± likelihood for hypothesized model)2 (maximized log ± likelihood for satur ± ated model)]. With large samples, G 2 follows a chi ± square distribution, with degrees of freedom equaling the diOEerence between the number of parameters in the saturated model and the number in the hypothesized model (Agresti 1996 ;SimonoOE 1998). 8. We thank the editor and the anonymous reviewer s for making this suggestion .…”

Section: Seementioning

confidence: 98%

Urbanization and Community Attachment in Rural Areas

Theodori¹,

Luloff²

2000

Society & Natural Resources

View full text Add to dashboard Cite

Section: Seementioning

confidence: 98%

Urbanization and Community Attachment in Rural Areas

Theodori¹,

Luloff²

2000

Society & Natural Resources

View full text Add to dashboard Cite

“…G 2 is a likelihood-ratio, goodness-of-fit statistic that tests the improvement in fit between two logistic regression models (Simonoff, 1998). The G 2 indicated that the introduction of the freshman seminar format variable did not improve the explanatory power of Set 2 over Set 1 concerning new students' retention rates.…”

Section: Resultsmentioning

confidence: 99%

The Impact of Clustering First Year Seminars with English Composition Courses on New Students' Retention Rates

Crissman

2001

Journal of College Student Retention: Research, Theory & Pr

View full text Add to dashboard Cite

Research has shown that two effective programs that positively impact upon new students' retention are first year seminars and clustering two or more academic courses together. This article discusses the results of one institution's approach to increasing student retention. The study analyzed the impact of clustering a first year seminar with an English composition course on new students' retention rates at a small, independent college in the northeast. The study, using multivariate regression models, showed no statistical difference in retention rates between students taking a clustered first year seminar and students taking a nonclustered first year seminar. This article discusses reasons that may account for the inconsistency with the literature, and closes with implications of the study and future research ideas. 137

show abstract

“…Logistic regressions cannot be computed with empty cells, and the commonly accepted solution is to exclude these categories from the analysis and note any patterns of empty cells when reporting results (Simonoff, 1998).…”

Section: Design and Analysesmentioning

confidence: 99%

“…Because change in risk status was a categorical variable, it was necessary to examine the data set before the second analysis to identify and exclude empty cells (i.e., categories in which no participants received multiple ODRs). Logistic regressions cannot be computed with empty cells, and the commonly accepted solution is to exclude these categories from the analysis and note any patterns of empty cells when reporting results (Simonoff, 1998).…”

Section: Design and Analysesmentioning

confidence: 99%

Kindergarten Reading Skill Level and Change as Risk Factors for Chronic Problem Behavior

McIntosh

Sadler

Brown

2011

Journal of Positive Behavior Interventions

View full text Add to dashboard Cite

In this study, the authors explored the effect of prereading skills at the start of kindergarten and change in skills during kindergarten on response to Tier I (universal) Schoolwide Positive Behavior Support in Grade 5. A longitudinal data set of 473 students, including Dynamic Indicators of Basic Early Literacy Skills measures at the start, middle, and end of kindergarten and office discipline referrals in Grade 5, was used to determine whether reading skills at school entry or change in reading skills over the course of kindergarten were more predictive of chronic problem behavior in Grade 5. Results of logistic regression analyses indicated that low initial phonological awareness predicted problem behavior, but including skill growth in the model resulted in significantly improved and more accurate prediction. Results are discussed in terms of early screening and intervention and reducing risk for problem behavior through quality Tier I reading instruction in kindergarten.

show abstract

Logistic Regression, Categorical Predictors, and Goodness-of-Fit: It Depends on Who You Ask

Cited by 16 publications

References 2 publications

Urbanization and Community Attachment in Rural Areas

Urbanization and Community Attachment in Rural Areas

The Impact of Clustering First Year Seminars with English Composition Courses on New Students' Retention Rates

Kindergarten Reading Skill Level and Change as Risk Factors for Chronic Problem Behavior

Contact Info

Product

Resources

About