Citation Context

...EM algorithm, issnot nested within our ordered probit models.sIn our empirical analysis, we employed thesAkaike Information Criterion (AIC) (Aikake 1974) and the Bayesian InformationsCriterion (BIC) (=-=Schwarz 1978-=-) to compare (21) with our alternative models.sThe formerscriterion corresponds to 2N − 2L * , where L *sis the maximized value of the logslikelihood function Lsin (2) orsLsin (18) and Nsis the total...

Citation Context

...n Q β,θ,β t( ),θ t( )( )sof the extended log likelihood in (4) issseparable in the parameters βsand θ . It is well known that the EM procedure convergessto the maximum of the likelihood function (1) (=-=Redner and Walker 1984-=-) (Xu and Jordans1996).s2.3.sPoisson Conditional DistributionssIn what follows, we assume that the conditional distribution of prenatal visits ni ,sgiven that pregnant woman isbelongs to latent class ...

Citation Context

...ns, investigators have assumed that the unknownsprobabilities of class membership are fixed parameters.sIn a few papers, analysts havesmodeled these probabilities as functions of observed covariates (=-=Nagin and Land 1993-=-;sRoeder, Lynch et al. 1999; Greene and Hensher 2003; Clark, Elité et al. 2005; Bagosd'Uva 2006; Clark and Etile 2006; Bago d'Uva and Jones 2008; Greene, Harris et al.s2008).sLatent class models have ...

Citation Context

...s in economics dates back at least to McKenzie’ssanalysis of factor prices in world trade (McKenzie 1955).s(See also Leamer’s use ofs“endowment triangles” in a three-factor general equilibrium model (=-=Leamer 1987-=-).)sFigure 3 contains a pair of triangular plots, the upper panel displaying the values of π i A ,sand the lower panel displaying the corresponding plot of the values of π i D .sIn both of thespanels ...

Citation Context

...I).sThird, we have not pursued a number of potentially important analytical strategiessfor assessing the impact of Seguro Popular on prenatal care.sIn particular, we did notsestimate two-part models (=-=Pohlmeier and Ulrich 1995-=-; Chen, Liu et al. 2003; Sepehri,sSarma et al. 2008).sTo be sure, we found that pregnant women in our first latent classshad a mean of 0.52 prenatal visits (Table II and Figure 1), and our predicted d...

Citation Context

...).sThe asset index,sin particular, was based upon household infrastructure, building materials, and ownershipsof certain durable assets, such as a refrigerator, television, telephone, oven and stoves(=-=McKenzie 2004-=-).sAt the locality level, we included variables reflecting rural location and level ofssocial deprivation.sThe latter indicator, also known as the locality’s index ofssocioeconomic marginality (“índic...

Citation Context

...investigators have used similar individual-, household- and local-levelscovariates to explain prenatal care utilization in developing countries (Wong, Popkin etsal. 1987; Pebley, Goldman et al. 1996; =-=Celik and Hotchkiss 2000-=-; Magadi, Madise et al.s2000; Chen, Liu et al. 2003; Sepehri, Sarma et al. 2008).s4.sRESULTSs4.1.sPrincipal Estimation ResultssTable II shows the estimates derived from our principal specification, as...

Citation Context

...tcomes, particularly in the area of maternal and reproductive health.sEvaluations ofsOportunidades, in particular, have demonstrated an increase in prenatal visits amongspregnant women beneficiaries (=-=Gertler 2000-=-).sDespite these initiatives, half of Mexico’s population remained uninsured at thesstart of 2001.sAt that time, a pregnant woman faced essentially three choices for hersprenatal and obstetric care.sF...

Citation Context

...icient is consistent with our prior work on Seguro Popular (Sosa-Rubí,sGalárraga et al. 2009).sA negative correlation coefficient has been seen in othersobservational studies in developing countries (=-=Waters 1999-=-), although the evidencesoverall remains mixed (Harmon and Nolan 2001; Trujillo 2003; Jowett, Deolalikar et al.s2004).sIn the second column of estimates in Table II, we find that indicators of lowerse...

Citation Context

...ers, analysts havesmodeled these probabilities as functions of observed covariates (Nagin and Land 1993;sRoeder, Lynch et al. 1999; Greene and Hensher 2003; Clark, Elité et al. 2005; Bagosd'Uva 2006; =-=Clark and Etile 2006-=-; Bago d'Uva and Jones 2008; Greene, Harris et al.s2008).sLatent class models have made significant progress in addressing thesheterogeneous effects of health-sector policies, such as those policies a...

Citation Context

... class models have made significant progress in addressing thesheterogeneous effects of health-sector policies, such as those policies aimed at improvingsaccess to medical care (Deb and Trivedi 1997; =-=Deb and Holmes 2000-=-; Jimenez-Martin,sLabeaga et al. 2002; Conway and Deb 2005).sThese models have demonstrated how suchsunobserved or partially observed factors as health status and attitudes toward health risksscan det...

Citation Context

...) π1 + π 2 + π 3 = 1{ } , we can plot them within a triangular planarsregion.sThe use of such triangular plots in economics dates back at least to McKenzie’ssanalysis of factor prices in world trade (=-=McKenzie 1955-=-).s(See also Leamer’s use ofs“endowment triangles” in a three-factor general equilibrium model (Leamer 1987).)sFigure 3 contains a pair of triangular plots, the upper panel displaying the values of π ...

Citation Context

... thesheterogeneous effects of health-sector policies, such as those policies aimed at improvingsaccess to medical care (Deb and Trivedi 1997; Deb and Holmes 2000; Jimenez-Martin,sLabeaga et al. 2002; =-=Conway and Deb 2005-=-).sThese models have demonstrated how suchsunobserved or partially observed factors as health status and attitudes toward health risksscan determine consumers’ demand responses.sHowever, such models h...

Citation Context

...t al. 2009).sA negative correlation coefficient has been seen in othersobservational studies in developing countries (Waters 1999), although the evidencesoverall remains mixed (Harmon and Nolan 2001; =-=Trujillo 2003-=-; Jowett, Deolalikar et al.s2004).sIn the second column of estimates in Table II, we find that indicators of lowerseducational attainment, lower family wealth, and participation in the Oportunidadessp...

Citation Context

...2 ,α 3( ) . This model, which can also be estimated by the EM algorithm, issnot nested within our ordered probit models.sIn our empirical analysis, we employed thesAkaike Information Criterion (AIC) (=-=Aikake 1974-=-) and the Bayesian InformationsCriterion (BIC) (Schwarz 1978) to compare (21) with our alternative models.sThe formerscriterion corresponds to 2N − 2L * , where L *sis the maximized value of the logsl...

Citation Context

... consumers’ demand responses.sHowever, such models have not squarelysconfronted the serious problem that health-care coverage is endogenous in nonexperimental databases (Cameron, Trivedi et al. 1988; =-=Atella and Deb 2008-=-).sEven inscontexts where the government exogenously introduces a change in eligibility criteria orsHarris & Sosa-Rubí May 18, 2009 Page 4sinsurance coverage for specific types of health care, the ind...

Citation Context

... dwellings with a dirt floor, with overcrowding, withoutsHarris & Sosa-Rubí May 18, 2009 Page 21srunning water, sewer drainage, or electricity, and the proportion of the population understhe poverty (=-=CONAPO 2005-=-).sWe also included a locality-level variable for the percentage penetration ofsSeguro Popular among the eligible population (Instituto Nacional de Estadística ysGeografía (INEGI) 2005).sSP penetratio...

Citation Context

...(Sosa-Rubí,sGalárraga et al. 2009).sA negative correlation coefficient has been seen in othersobservational studies in developing countries (Waters 1999), although the evidencesoverall remains mixed (=-=Harmon and Nolan 2001-=-; Trujillo 2003; Jowett, Deolalikar et al.s2004).sIn the second column of estimates in Table II, we find that indicators of lowerseducational attainment, lower family wealth, and participation in the ...

Citation Context

...ptures thesheterogeneity that is implicit in the negative binomial distribution.sFinally, our solesmeasure of prenatal care was the number of visits.sWe did not address the timing of thesfirst visit (=-=Harris 1982-=-; Wong, Popkin et al. 1987; Magadi, Madise et al. 2000), the typesof provider (Pebley, Goldman et al. 1996), or the content of care (Wong, Popkin et al.s1987) in the present study.sHarris & Sosa-Rubí ...

Citation Context

...ation, medication spending, or health outcomes (King,sGakidou et al. 2009).sThe latter negative findings may have been due to the relativelysshort, 10-month duration of the experimental intervention (=-=Victora and Peters 2009-=-).s3.2.sThe 2006 ENSANUT Survey DatasWe analyzed data from the 2006 National Health and Nutrition Survey (EncuestasNacional de Salud y Nutrición, or “ENSANUT”), a nationally representative crosssectio...