Logit Link Function Research Articles

Identification and Characterization of a Compound That Protects Cardiac Tissue from Human Ether-à-go-go-related Gene (hERG)-related Drug-induced Arrhythmias

The human Ether-à-go-go-related gene (hERG)-encoded K(+) current, I(Kr) is essential for cardiac repolarization but is also a source of cardiotoxicity because unintended hERG inhibition by diverse pharmaceuticals can cause arrhythmias and sudden cardiac death. We hypothesized that a small molecule that diminishes I(Kr) block by a known hERG antagonist would constitute a first step toward preventing hERG-related arrhythmias and facilitating drug discovery. Using a high-throughput assay, we screened a library of compounds for agents that increase the IC(70) of dofetilide, a well characterized hERG blocker. One compound, VU0405601, with the desired activity was further characterized. In isolated, Langendorff-perfused rabbit hearts, optical mapping revealed that dofetilide-induced arrhythmias were reduced after pretreatment with VU0405601. Patch clamp analysis in stable hERG-HEK cells showed effects on current amplitude, inactivation, and deactivation. VU0405601 increased the IC(50) of dofetilide from 38.7 to 76.3 nM. VU0405601 mitigates the effects of hERG blockers from the extracellular aspect primarily by reducing inactivation, whereas most clinically relevant hERG inhibitors act at an inner pore site. Structure-activity relationships surrounding VU0405601 identified a 3-pyridiyl and a naphthyridine ring system as key structural components important for preventing hERG inhibition by multiple inhibitors. These findings indicate that small molecules can be designed to reduce the sensitivity of hERG to inhibitors.

Power Analysis for Trials With Discrete-Time Survival Endpoints

Studies on event occurrence aim to investigate if and when subjects experience a particular event. The timing of events may be measured continuously using thin precise units or discretely using time periods. The latter metric of time is often used in social science research and the generalized linear model (GLM) is an appropriate model for data analysis. While the design of trials with continuous-time survival endpoints has been extensively studied, hardly any guidelines are available for trials with discrete-time survival endpoints. This article studies the relationship between sample size and power to detect a treatment effect in a trial with two treatment conditions. The authors use the exponential and the Weibull survival functions to represent constant and varying hazard rates. Furthermore, logit and complementary log–log link functions are used. For constant hazard rates, the power depends on the event proportions at the end of the trial in both treatment arms and on the number of time periods. For varying hazard rates, the power also depends on the shape of the survival functions and different power levels are observed in each time period in case the logit link function is used. For any survival function, power decreases if attrition is present. The authors provide R code to perform the power calculations.

Predicting Emergency Department Inpatient Admissions to Improve Same‐day Patient Flow

The objectives were to evaluate three models that use information gathered during triage to predict, in real time, the number of emergency department (ED) patients who subsequently will be admitted to a hospital inpatient unit (IU) and to introduce a new methodology for implementing these predictions in the hospital setting. Three simple methods were compared for predicting hospital admission at ED triage: expert opinion, naïve Bayes conditional probability, and a generalized linear regression model with a logit link function (logit-linear). Two months of data were gathered from the Boston VA Healthcare System's 13-bed ED, which receives approximately 1,100 patients per month. Triage nurses were asked to estimate the likelihood that each of 767 triaged patients from that 2-month period would be admitted after their ED treatment, by placing them into one of six categories ranging from low to high likelihood. Logit-linear regression and naïve Bayes models also were developed using retrospective data and used to estimate admission probabilities for each patient who entered the ED within a 2-month time frame, during triage hours (1,160 patients). Predictors considered included patient age, primary complaint, provider, designation (ED or fast track), arrival mode, and urgency level (emergency severity index assigned at triage). Of the three methods considered, logit-linear regression performed the best in predicting total bed need, with a receiver operating characteristic (ROC) area under the curve (AUC) of 0.887, an R(2) of 0.58, an average estimation error of 0.19 beds per day, and on average roughly 3.5 hours before peak demand occurred. Significant predictors were patient age, primary complaint, bed type designation, and arrival mode (p < 0.0001 for all factors). The naïve Bayesian model had similar positive predictive value, with an AUC of 0.841 and an R(2) of 0.58, but with average difference in total bed need of approximately 2.08 per day. Triage nurse expert opinion also had some predictive capability, with an R(2) of 0.52 and an average difference in total bed need of 1.87 per day. Simple probability models can reasonably predict ED-to-IU patient volumes based on basic data gathered at triage. This predictive information could be used for improved real-time bed management, patient flow, and discharge processes. Both statistical models were reasonably accurate, using only a minimal number of readily available independent variables.

Risk factors for Salmonella spp in Portuguese breeding pigs using a multilevel analysis

Salmonella is the second most frequent cause of foodborne illness in the European Union (EU), so EU enforced legislation to achieve a reduction in Salmonella prevalence in the swine sector. To set the reduction target each country carried out a baseline survey to estimate Salmonella prevalence. The aim of our study was to identify risk factors for the presence of Salmonella in breeding pigs based on the data of the Baseline Study for Salmonella in Breeding Pigs in Portugal. In total, 1670 pen fecal samples from 167 herds were tested by culture and 170 samples tested positive. Along with the collection of the samples a survey was applied to collect information about the herd management and potential risk factors. Multilevel analysis was applied to the data using generalized linear mixed models and a logit link function. The outcome variable was the presence/absence of Salmonella in the pen fecal samples. The first level was assigned to the pen fecal samples and the second level to the herds. The results showed significant associations between Salmonella occurrence and the factors (p<0.05): maternity pens versus mating pens (OR=0.39, 95%CI: 0.24–0.63), feed from external or mixed source versus home source (OR=2.81, 95%CI: 1.19–6.61), more than 10 animals per pen versus 10 animals per pen (OR=2.02, 95%CI: 1.19–3.43), North Region versus Alentejo Region (OR=3.86, 95%CI: 1.08–13.75), rodents control (OR=0.23, 95%CI: 0.090–0.59), more than 90% of boars homebred or no boars versus more than 90% of boars from an external source (OR=0.54, 95%CI: 0.3–0.97), semen from another herd versus semen from insemination centers (OR=4.47, 95%CI: 1.38–14.43) and herds with a size of 170 or more sows (OR=1.82, 95%CI: 1.04–3.19). This study offers very relevant information for both the Portuguese veterinary authorities and the pig farmers currently developing control programmes for Salmonella. This is the first study providing evidence for semen and boars source as risk factors for Salmonella in breeding pigs.

P2‐283: Sensitivity of cognition measures in a longitudinal study of healthy elderly, amnestic MCI and mild‐to‐moderate Alzheimer's disease

semantic) and Language (comprehension, writing, repetition, naming, and reading). Distinguishing between Fluency and Language is important because distinct aspects of language appear to decline independently, as reflected in MCI; e.g., recent evidence of early onset of decline in fluency (both phonemic and semantic) but not in vocabulary (Singh-Manoux, et al 2012). Additionally, recent research indicates association of MCI and a subtle decline in syntactic and semantic factors involved in complex sentence formation, components not tested by ACE-R (Sherman et al, 2011, Lust et al 2011; see also Kemper et al 1993, Almor et al 1999). Methods: In this study, we investigate early language changes by comparing results in three populations (amnestic MCI; n 1⁄4 13, ages 65-80, healthy aging; n 1⁄4 14, ages 65-80, and young controls; n 1⁄4 10, ages 20-29) on the ACE-R (overall scores, fluency and language subscores) to results on psycholinguistic tests of syntactic and semantic processing of complex sentences (elicited imitation). Statistical analysis was by a logistic-linear mixed model with binomial error assumption and a logit link function. Results: Results show that while the overall ACE-R scores and ‘fluency’ subscore differ significantly between the young controls and the two aging groups, they do not differ significantly between the two aging groups (healthy and MCI). The ACE-R Language subscore showed no difference across the groups. Psycholinguistic tests of linguistic performance in complex sentences did show a significant difference between Healthy Aging and MCI populations. Regressions of linguistic performance in complex sentences on both ACE-R language and fluency subscores were significant in the Healthy Aging group, and regressions on fluency (but not language) subscores were significant in the MCI group. Conclusions: These findings suggest that attention to psycholinguistic factors may increase our sensitivity to early changes in the language domain in MCI.

Modelos lineares generalizados mistos na avaliação genética da prenhez precoce na raça Nelore

Utilizaram-se os modelos lineares generalizados com as funções de ligação probit e logit na avaliação da prenhez precoce, e observaram-se os efeitos na variabilidade genética e na seleção de reprodutores quando diferentes idades são adotadas na definição dessa característica. A prenhez precoce foi estudada aos 15 (PP15) e 21 meses (PP21). Correlações entre os valores genéticos preditos e a porcentagem de touros em comum, considerando 10% dos touros com maiores valores genéticos (TOP10), entre a classificação dos modelos com função de ligação logit e probit e em cada modelo entre PP15 e PP21, foram calculadas. As herdabilidades para PP15 e PP21 foram próximas entre os modelos, exceto para PP15 utilizando a função de ligação probit. Aquelas entre os modelos e a TOP/10 foram altas. Os critérios de Akaike e Bayesiano apresentaram-se semelhantes entre os modelos. As correlações entre PP15 e PP21 e a TOP10, considerando o mesmo modelo, foram de média magnitude.

Hiding from the Moonlight: Luminosity and Temperature Affect Activity of Asian Nocturnal Primates in a Highly Seasonal Forest

The effect of moonlight and temperature on activity of slow lorises was previously little known and this knowledge might be useful for understanding many aspects of their behavioural ecology, and developing strategies to monitor and protect populations. In this study we aimed to determine if the activity of the pygmy loris (Nycticebus pygmaeus) is affected by ambient temperature and/or moonlight in a mixed deciduous forest. We radio-collared five females and five males in the Seima Protection Forest, Cambodia, in February to May, 2008 and January to March, 2009 and recorded their behaviour at 5 minutes intervals, totalling 2736 observations. We classified each observation as either inactive (sleeping or alert) or active behaviour (travel, feeding, grooming, or others). Moon luminosity (bright/dark) and ambient temperature were recorded for each observation. The response variable, activity, was binary (active or inactive), and a logit link function was used. Ambient temperature alone did not significantly affect mean activity. Although mean activity was significantly affected by moonlight, the interaction between moonlight and temperature was also significant: on bright nights, studied animals were increasingly more active with higher temperature; and on dark nights they were consistently active regardless of temperature. The most plausible explanation is that on bright cold nights the combined risk of being seen and attacked by predators and heat loss outweigh the benefit of active behaviours.

Link Functions in Multi‐Locus Genetic Models: Implications for Testing, Prediction, and Interpretation

“Complex” diseases are, by definition, influenced by multiple causes, both genetic and environmental, and statistical work on the joint action of multiple risk factors has, for more than 40 years, been dominated by the generalized linear model (GLM). In genetics, models for dichotomous traits have traditionally been approached via the model of an underlying, normally distributed, liability. This corresponds to the GLM with binomial errors and a probit link function. Elsewhere in epidemiology, however, the logistic regression model, a GLM with logit link function, has been the tool of choice, largely because of its convenient properties in case-control studies. The choice of link function has usually been dictated by mathematical convenience, but it has some important implications in (a) the choice of association test statistic in the presence of existing strong risk factors, (b) the ability to predict disease from genotype given its heritability, and (c) the definition, and interpretation of epistasis (or epistacy). These issues are reviewed, and a new association test proposed. Genet. Epidemiol. 36:409–418, 2012. © 2012 Wiley Periodicals, Inc.

Evaluating strawberry breeding selections for postharvest fruit decay

Fruit from the annual replicated yield assessments for the USDA-ARS strawberry (Fragaria ×ananassa Duchesne ex Rozier) breeding program at Beltsville, MD, 2007 and 2008, were evaluated for postharvest decay development after storage at 5 °C. Generalized linear genotype × year mixed model analysis of covariance with overdispersed binomial distribution, logit link function, and Laplace optimization method was used to determine the relationships between postharvest proportion decay and storage period (SP), season’s harvest number (HN), and genotype-specific harvest number through the season; conditioned on the confounding of harvest with SP. To determine if limiting analyses to replicates (plot × harvest combinations) with a minimum number of berries would increase statistical precision, concurrent analyses were conducted using data from replicates with more than: zero, three, and six berries. Limiting data to genotypes for which more than three or more than six berries could be evaluated, resulted in loss of data and reduced statistical power. Within the range of SP observed, 13–19 days, consistency among genotypes of the linear relationship between SP and logit postharvest proportion decay, unless strongly influenced by a confounded harvest effect, indicated all SP in the range tested provided equivalent ability to discern genotypes. The significant genotype × HN interaction effect indicated maximizing the number of harvests was preferable to limiting to a few “informative” harvests. The significant genotype × year interaction effect indicated the decision to release a genotype as a cultivar will demand multiple years of evaluation. Making determinations each year to discard genotypes or use them as parents for postharvest quality was best facilitated by the simple process of calculating arithmetic genotypic means for each harvest date separately, charting the results, and using the charts to visually assess genotypic performance trends across harvests.

Genetic parameter estimates for growth traits and prolificacy in Raeini Cashmere goats

The main objectives of this study were to estimate genetic and phenotypic parameters for growth traits and prolificacy in the Raeini Cashmere goat. Traits included, birth weight (BWT), weaning weight (WWT), 6-month weight (6WT), 9-month weight (9WT), 12-month weight (12WT), average daily gain from birth to weaning (ADG1), average daily gain from weaning to 6WT (ADG2), average daily gain from 6WT to 12WT (ADG3), survival rate (SR), litter size at birth (LSB) and litter size at weaning (LSW) and total litter weight at birth (LWB). Data were collected over a period of 28 years (1982-2009) at the experimental breeding station of Raeini goat, southeast of Iran. Genetic parameters were estimated with univariate models using restricted maximum likelihood (REML) procedures. In addition to an animal model, sire and threshold models, using a logit link function, were used for analyses of SR. Age of dam, birth of type, sex and of kidding had significant influence (p < 0.05 or 0.01) all the traits. Direct heritability estimates were low for prolificacy traits (0.04 ± 0.01 for LSB, 0.09 ± 0.02 for LSW, 0.16 ± 0.02 for LWB and 0.05 ± 0.02 for SR) and average daily gain (0.12 ± 0.03 for ADG1, 0.08 ± 0.02 for ADG2, and 0.07 ± 0.03 for ADG3) to moderate for production traits (0.22 ± 0.02 for BWT, 0.25 ± 0.02 for WWT, 0.29 ± 0.04 for 6WT, 0.30 ± 0.02 for 9WT, 0.32 ± 0.05 for 12WT). The estimates for the maternal additive genetic variance ratios were lower than direct heritability for BWT (0.17 ± 0.03) and WWT (0.07 ± 0.02).

Generalized linear mixed models for the genetic evaluation of binary reproductive traits: a simulation study

The objective of this study was to evaluate the use of probit and logit link functions for the genetic evaluation of early pregnancy using simulated data. The following simulation/analysis structures were constructed: logit/logit, logit/probit, probit/logit, and probit/probit. The percentages of precocious females were 5, 10, 15, 20, 25 and 30% and were adjusted based on a change in the mean of the latent variable. The parametric heritability (h²) was 0.40. Simulation and genetic evaluation were implemented in the R software. Heritability estimates (ĥ²) were compared with h² using the mean squared error. Pearson correlations between predicted and true breeding values and the percentage of coincidence between true and predicted ranking, considering the 10% of bulls with the highest breeding values (TOP10) were calculated. The mean ĥ² values were under- and overestimated for all percentages of precocious females when logit/probit and probit/logit models used. In addition, the mean squared errors of these models were high when compared with those obtained with the probit/probit and logit/logit models. Considering ĥ², probit/probit and logit/logit were also superior to logit/probit and probit/logit, providing values close to the parametric heritability. Logit/probit and probit/logit presented low Pearson correlations, whereas the correlations obtained with probit/probit and logit/logit ranged from moderate to high. With respect to the TOP10 bulls, logit/probit and probit/logit presented much lower percentages than probit/probit and logit/logit. The genetic parameter estimates and predictions of breeding values of the animals obtained with the logit/logit and probit/probit models were similar. In contrast, the results obtained with probit/logit and logit/probit were not satisfactory. There is need to compare the estimation and prediction ability of logit and probit link functions.

Bayesian and REML analysis of twinning and fertility in Thoroughbred horses

Horses have low fertility, which affects profitability of breeding. Additional economic loss is generated by various complications accompanying twin pregnancies which are particularly prevalent in Thoroughbred horses. Therefore it is important to identify environmental and genetic factors influencing this trait. The analysis was based on studbooks from 1994 to 2004 provided by the Czech Jockey Club in Prague. The dataset contained 7703 records of 2161 Thoroughbred mares born between 1975 and 1996. Variance components for twinning rate and fertility were estimated using the ASREML and MCMCglmm/R packages, with a threshold animal model and a threshold sire model with binary distribution and a LOGIT link function. Estimates of additive genetic variance for fertility were much higher for the Bayesian methods than for REML. This resulted in higher heritabilities, though the credible interval for the Bayesian estimates was wide. Fertility traits, such as twinning rate should be modelled with generalised linear mixed-effects models (GLMMs). The GLMMs are much more suited for traits having a binary distribution than linear models. For twining, a trait with low frequency, sire model provided more precise estimates than animal model.

Changes in condom use during a microbicide clinical trial in Pune, India

Researchers have raised concerns that microbicide use during clinical trials would displace condom use. We sought to understand whether condom use changed for participants in a microbicide clinical trial in Pune, India, to understand whether condom shifts were a legitimate concern. We hypothesize that women participating in a microbicide clinical trial in Pune, India, were more likely, on average, to report condom use at follow-up. We further hypothesize that men, whose female partners participated in a microbicide clinical trial were more likely, on average, to report condom use at follow-up. The outcome measure for reported condom use was a dichotomous variable to indicate whether or not the participant had used a male or female condom with a sexual partner since 2 months before enrollment or since the last survey, depending on the visit. Data are from semi-structured interviews at baseline, 2 months, 4 months, and 6 months with HPTN 059 clinical trial participants (100 women and 57 male partners). We used generalized estimating equations with a logit link function, exchangeable correlation, and a binomial family to model condom use. The odds of condom use for clinical trial women increased from baseline to 6 months by a factor of 3.7 (95% CI: 1.84–7.63) and the change in odds of condom use for clinical trial men from baseline to 6 months increased by a factor of 2.58 (95% CI: 1.37–4.85). We found concerns about microbicide use displacing condom use were not merited in this study population. The percent of participants reporting condom use declined from 4 to 6 months, suggesting that increases in condom use may only be during active study participation. Information about clinical trial factors that enabled these men and women to enact this important HIV prevention behavior is needed to develop interventions.

Association of claw disorders with claw horn colour in norwegian red cattle - a cross-sectional study of 2607 cows from 112 herds

Claw disorders cause problems in dairy cattle all over the world. Nutrition, feeding, environment, claw trimming routines, hormonal changes related to calving and genetics are among the factors which influence the pathogenesis. The colour of the claw horn (pigmentation) has been suggested to play a role. The aim of this study was to investigate if there were any associations between the colour of the sole horn and claw disorders detected at claw trimming. Altogether, 2607 cows on 112 farms were claw trimmed once and the colour (dark, mixed or light) of the right lateral hind claw and hind claw disorders were recorded by 13 trained claw trimmers. The data were analysed using logistic regression models with logit link function, binomial distribution and herd and claw trimmer as repeated effects, with herd nested within claw trimmer. Haemorrhages of the sole (HS) and white line (HWL) were more frequently found in light than in dark claws (OR = 2.61 and 2.34, respectively). Both HS (OR = 1.43) and corkscrewed claws (OR = 1.84) were slightly more prevalent among cows which had claws with mixed colour versus dark claws. There were no significant associations of other claw disorders with claw horn colour.

Logit models for the analysis of two-way categorical data

This article discusses the application of logit models for the analyses of 2-way categorical observations. The models described are generalized linear models using the logit link function. One of the models is the Rasch model (Rasch, 1960). The objective is to test hypotheses of marginal and conditional independence between explanatory quantities and multiple binary responses. By means of a numerical example, the relevance of such hypotheses for educational research is demonstrated. The aim is, in particular, to extend the applicability of the Rasch model to test hypotheses of conditional independence.

Genetic parameter estimates for pre-weaning performance and reproduction traits in Markhoz goats

Genetic parameters for growth, mortality and reproductive performances of Markhoz goats were estimated from data collected during 1993–2010 at Markhoz goat Performance Testing Station in Sanandaj, Iran. For kid performance traits 3763 records were available for birth weight (BW), 2931 for weaning weight (WW), average daily gain (ADG) and Kleiber ratio (KR) (approximated as ADW/WW0.75) and 3032 for pre-weaning mortality (PWM). For doe reproductive performance traits there were 2920 records available for litter size at birth (LSB), litter size at weaning (LSW), total litter weight at birth (TLWB) and litter mean weight per kid born (LMWKB), and 2182 for total litter weight at weaned (TLWW) and litter mean weight per kid weaned (LMWKW). Genetic parameters were estimated with univariate and bivariate models using restricted maximum likelihood (REML) procedures. Random effects were explored by fitting additive direct genetic effects, maternal additive genetic effects, maternal permanent environmental effects, the covariance between direct and maternal genetic effects, and common litter effects in different models for pre-weaning traits of kids. Also, in addition to an animal model, sire and threshold models, using a logit link function, were used for analyses of PWM. Models for LSB, LSW, TLWB, TLWW, LMWKB, and LMWKW included direct additive genetic effects, permanent environmental effects due to the animal as well as service sire effects. Estimated direct heritabilities were moderate for pre-weaning traits (0.22 for BW, 0.16 for WW, 0.21 for ADG, and 0.27 for KR and 0.29 for PWM), and low for reproduction traits (0.01 for LSB, 0.01 for LSW, 0.02 for TLWB, 0.03 for TLWW, 0.07 for LMWKB, and 0.06 for LMWKW). The estimates for the maternal additive genetic variance ratios were lower than direct heritability for BW (0.07) and KR (0.04). The estimate for the maternal permanent environmental variance ratios (c2) varied from 0.01 for KR to 0.07 for WW and ADG. The magnitude of common litter variance ratios (l2) was more substantial for BW (0.46) than the PWM (0.19) and KR (0.16). The estimate for the permanent environmental variance due to the animal (c2) ranged from 0.03 for LMWKB to 0.07 for TLWB and LMWKW, whereas service sire effects (s2) ranged from 0.02 to 0.04. The correlation between direct and maternal genetic effects were negative and high for BW (−0.51) and KR (−0.62). The genetic correlations between pre-weaning growth traits were positive and moderate to strong, as were genetic correlations between reproductive traits. Between BW and PWM the correlation was −0.35. Phenotypic and environmental correlations for all traits were generally lower than genetic correlations.

Evidence for genetic variance in resistance to tuberculosis in Great Britain and Irish Holstein-Friesian populations

BackgroundHere, we jointly summarise scientific evidence for genetic variation in resistance to infection with Mycobacterium bovis, the primary agent of bovine tuberculosis (TB), provided by two recent and separate studies of Holstein-Friesian dairy cow populations in Great Britain (GB) and Ireland.MethodsThe studies quantified genetic variation within archived data from field and abattoir surveillance control programmes within each country. These data included results from the single intradermal comparative tuberculin test (SICTT), abattoir inspection for TB lesions and laboratory confirmation of disease status. Threshold animal models were used to estimate variance components for responsiveness to the SICTT and abattoir confirmed M. bovis infection. The link functions between the observed 0/1 scale and the liability scale were the complementary log-log in the GB, and logit link function in the Irish population.Results and discussionThe estimated heritability of susceptibility to TB, as judged by responsiveness to the SICTT, was 0.16 (0.012) and 0.14 (0.025) in the GB and Irish populations, respectively. For abattoir or laboratory confirmation of infection, estimates were 0.18 (0.044) and 0.18 (0.041) from the GB and the Irish populations, respectively.ConclusionsEstimates were all significantly different from zero and indicate that exploitable variation exists among GB and Irish Holstein Friesian dairy cows for resistance to TB. Epidemiological analysis suggests that factors such as variation in exposure or imperfect sensitivity and specificity would have resulted in underestimation of the true values.

Median infectious dose (ID 50) of porcine reproductive and respiratory syndrome virus isolate MN-184 via aerosol exposure

The median infectious dose (ID 50) of porcine reproductive and respiratory syndrome (PRRS) virus isolate MN-184 was determined for aerosol exposure. In 7 replicates, 3-week-old pigs ( n = 58) respired 10 l of airborne PRRS virus from a dynamic aerosol toroid (DAT) maintained at −4 °C. Thereafter, pigs were housed in isolation and monitored for evidence of infection. Infection occurred at virus concentrations too low to quantify by microinfectivity assays. Therefore, exposure dose was determined using two indirect methods (“calculated” and “theoretical”). “Calculated” virus dose was derived from the concentration of rhodamine B monitored over the exposure sequence. “Theoretical” virus dose was based on the continuous stirred-tank reactor model. The ID 50 estimate was modeled on the proportion of pigs that became infected using the probit and logit link functions for both “calculated” and “theoretical” exposure doses. Based on “calculated” doses, the probit and logit ID 50 estimates were 1 × 10 −0.13 TCID 50 and 1 × 10 −0.14 TCID 50, respectively. Based on “theoretical” doses, the probit and logit ID 50 were 1 × 10 0.26 TCID 50 and 1 × 10 0.24 TCID 50, respectively. For each point estimate, the 95% confidence interval included the other three point estimates. The results indicated that MN-184 was far more infectious than PRRS virus isolate VR-2332, the only other PRRS virus isolate for which ID 50 has been estimated for airborne exposure. Since aerosol ID 50 estimates are available for only these two isolates, it is uncertain whether one or both of these isolates represent the normal range of PRRS virus infectivity by this route.

A generalized linear mixed model for longitudinal binary data with a marginal logit link function

Longitudinal studies of a binary outcome are common in the health, social, and behavioral sciences. In general, a feature of random effects logistic regression models for longitudinal binary data is that the marginal functional form, when integrated over the distribution of the random effects, is no longer of logistic form. Recently, Wang and Louis (2003) proposed a random intercept model in the clustered binary data setting where the marginal model has a logistic form. An acknowledged limitation of their model is that it allows only a single random effect that varies from cluster to cluster. In this paper, we propose a modification of their model to handle longitudinal data, allowing separate, but correlated, random intercepts at each measurement occasion. The proposed model allows for a flexible correlation structure among the random intercepts, where the correlations can be interpreted in terms of Kendall's τ. For example, the marginal correlations among the repeated binary outcomes can decline with increasing time separation, while the model retains the property of having matching conditional and marginal logit link functions. Finally, the proposed method is used to analyze data from a longitudinal study designed to monitor cardiac abnormalities in children born to HIV-infected women.

Bayesian Analysis of Individual Choice Behavior With Aggregate Data

The statistical analysis of individual choice behavior can be hampered by aggregate information. When a choice response variable is fully observed, individual choice behavior is typically studied with generalized linear models with the logit or probit link function. The individual choice response variable is, however, often available in an aggregate form because of data confidentiality, privacy protection, and a limited database quota. Then the generalized linear models are no longer directly applicable to such aggregate data. In this article, we confront such an aggregate data problem by using the method of data augmentation, where latent individual choice responses are augmented while satisfying a constraint on their aggregate sum. To do so, we devise the novel choice-wise sampling algorithm for a generalized linear model with aggregate binary responses. The proposed algorithm is applied to the 2006 Pennsylvania gubernatorial election data, where only aggregate votes cast for each candidate are available because of the privacy protection of voters. This article has supplementary material online.