Statistical Inference Based on a General Model of Unobserved Heterogeneity

Statistical Modelling - Lecture Notes in Statistics ◽

10.1007/978-1-4612-0789-4_10 ◽

1995 ◽

pp. 75-82 ◽

Cited By ~ 5

Author(s):

Ekkehart Dietz ◽

Dankmar Böhning

Keyword(s):

Statistical Inference ◽

General Model ◽

Unobserved Heterogeneity

Download Full-text

Essentials of Statistical Inference

10.1017/cbo9780511755392 ◽

2005 ◽

Cited By ~ 85

Author(s):

G. A. Young ◽

R. L. Smith

Keyword(s):

Statistical Inference

Download Full-text

Randomization and Statistical Inference

Communications in Statistics - Simulation and Computation ◽

10.1080/03610917508548430 ◽

1975 ◽

Vol 4 (8) ◽

pp. 723-735

Author(s):

Robert Easterling

Keyword(s):

Statistical Inference

Download Full-text

University Dropout

Journal of Individual Differences ◽

10.1027/1614-0001.29.3.134 ◽

2008 ◽

Vol 29 (3) ◽

pp. 134-147 ◽

Cited By ~ 9

Author(s):

Manuel C. Voelkle ◽

Nicolas Sander

Keyword(s):

Grade Point Average ◽

Structural Equation ◽

Unobserved Heterogeneity ◽

School Grade ◽

Grade Point ◽

High School Grade ◽

Hands On ◽

University Dropout ◽

The Individual ◽

A Prospective Study

University dropout is a politically and economically important factor. While a number of studies address this issue cross-sectionally by analyzing different cohorts, or retrospectively via questionnaires, few of them are truly longitudinal and focus on the individual as the unit of interest. In contrast to these studies, an individual differences perspective is adopted in the present paper. For this purpose, a hands-on introduction to a recently proposed structural equation (SEM) approach to discrete-time survival analysis is provided ( Muthén & Masyn, 2005 ). In a next step, a prospective study with N = 1096 students, observed across four semesters, is introduced. As expected, average university grade proved to be an important predictor of future dropout, while high-school grade-point average (GPA) yielded no incremental predictive validity but was completely mediated by university grade. Accounting for unobserved heterogeneity, three latent classes could be identified with differential predictor-criterion relations, suggesting the need to pay closer attention to the composition of the student population.

Download Full-text