Two-stage Logistic Regression Model Research Articles

Does Fossil Fuel Financing Affect Banks’ ESG Ratings?

The study was conducted on a sample of 60 of the world’s biggest banks financing the largest fossil fuel entities. The aim is to identify the determinants of ESG ratings of these banks and to determine how relevant their actual credit and investment exposure is to this assessment. The indirect objective is also an examination of whether coal power financing affects ESG ratings. Two logistic regression models have been explored: one dedicated to the identification of high ESG risk banks and the second to predict low ESG risk, which thereafter were combined into one final model. The results indicate that an increase in the Sustainable Development Index (SDI) translates into a decline in the odds of being assigned to the high-risk ESG group relative to the probability of being qualified to the low- or medium-risk ESG group. This study is the first to analyse the impact of actual exposures of the world’s largest banks to the fossil fuels sector on their ESG ratings. The value added is the use of a unique database, the focus on actual rather than declared effects of banks’ policies, and the use of a two-stage logistic regression model construction. The proved relationships are important and of practical relevance to bank managers, regulators, and ESG rating providers. Since the research is conducted on the basis of ESG provided only by one rating agency verification of conclusions with the use of ratings of other agencies, confronting benefits from financing the fossil fuels sector with losses resulting from an increase in the cost of obtaining financing are only selected directions for further research.

Construction of Credit Evaluation Index System for Two-Stage Bayesian Discrimination: An Empirical Analysis of Small Chinese Enterprises

In China, small enterprises have a direct role in economic growth, but they have difficulty in financing development. To address this problem, this paper creates a small business credit evaluation index using a two-stage Bayesian discriminant model. In the first stage, customers are distinguished by whether they are in default, and in the second stage, customers with continuing default are divided into those with a high default loss rate and those with a low default loss rate. The literature to date has identified a credit index only for the first stage; the credit evaluation index proposed here is based on two stages, which is more sensitive. Then, we conduct an empirical analysis using credit data on 3,111 small enterprises in China with a two-stage nonparametric Bayesian discriminant model and a parametric discriminant model, and then, we test the two indicator systems with discriminant accuracy and an ROC curve; the discriminant accuracy of the established index system is 77.95% and 70.95%, respectively, and their prediction accuracy is 0.902 and 0.866, respectively; they show that the constructed indicator system is robust and effective. Finally, we conduct a comparative analysis of discriminant accuracy in three models, finding that the two-stage nonparametric model is optimal, the two-stage logistic regression model is suboptimal, and the two-stage parametric model is poor.

Search Engine Consumer Journeys: Exploring and Segmenting Click-Through Behaviors

The consumer online search journey is viewed as one of the key determinants toward consumer purchase decisions. Using a unique disaggregated query-level data set collected from a leading Chinese search engine, we employ a latent-class, two-stage logistic regression model and identify the existence of three distinct consumer segments in the hospitality industry. Each group represents a unique stage of the consumer experience along a typical search journey, leading to a set of click-through behaviors. The consumer search and click-through insights garnered from this study provide hotel marketers with deeper understandings and actionable guidelines to the development of their search advertising campaign and (re)targeting strategies. The segmentation model and the empirical insights also demonstrate the possibility that a segment-specific ad assortment approach could improve advertising efficiency, enrich consumer’s search experience, and eventually create a win-win situation that benefit all entities involved in the hospitality-related, search advertising domain.

Gender, financial literacy, and preretirement planning in the UK

ABSTRACTThis study uses a two-stage logistic regression model to explore the determinants of retirement planning behavior. A survey of 516 UK women and men of prestate retirement age was used to establish interrelationships between gender, retirement planning, and financial literacy, taking account of attitudinal and expectational variables. Findings confirm lower levels of planning among women, but contrary to previous studies, this study reveals that financial literacy is not significantly related to planning. Furthermore, when attitudinal and expectational variables are incorporated as independent variables, gender becomes statistically insignificant as a determinant of planning.

A mobile platform for automated screening of asthma and chronic obstructive pulmonary disease.

Chronic Obstructive Pulmonary Disease (COPD) and asthma each represent a large proportion of the global disease burden; COPD is the third leading cause of death worldwide and asthma is one of the most prevalent chronic diseases, afflicting over 300 million people. Much of this burden is concentrated in the developing world, where patients lack access to physicians trained in the diagnosis of pulmonary disease. As a result, these patients experience high rates of underdiagnosis and misdiagnosis. To address this need, we present a mobile platform capable of screening for Asthma and COPD. Our solution is based on a mobile smart phone and consists of an electronic stethoscope, a peak flow meter application, and a patient questionnaire. This data is combined with a machine learning algorithm to identify patients with asthma and COPD. To test and validate the design, we collected data from 119 healthy and sick participants using our custom mobile application and ran the analysis on a PC computer. For comparison, all subjects were examined by an experienced pulmonologist using a full pulmonary testing laboratory. Employing a two-stage logistic regression model, our algorithms were first able to identify patients with either asthma or COPD from the general population, yielding an ROC curve with an AUC of 0.95. Then, after identifying these patients, our algorithm was able to distinguish between patients with asthma and patients with COPD, yielding an ROC curve with AUC of 0.97. This work represents an important milestone towards creating a self-contained mobile phone-based platform that can be used for screening and diagnosis of pulmonary disease in many parts of the world.

A test for detecting etiologic heterogeneity in epidemiological studies

Current statistical methods for analyzing epidemiological data with disease subtype information allow us to acquire knowledge not only for risk factor-disease subtype association but also, on a more profound account, heterogeneity in these associations by multiple disease characteristics (so-called etiologic heterogeneity of the disease). Current interest, particularly in cancer epidemiology, lies in obtaining a valid p-value for testing the hypothesis whether a particular cancer is etiologically heterogeneous. We consider the two-stage logistic regression model along with pseudo-conditional likelihood estimation method and design a testing strategy based on Rao's score test. An extensive Monte Carlo simulation study is carried out, false discovery rate and statistical power of the suggested test are investigated. Simulation results indicate that applying the proposed testing strategy, even a small degree of true etiologic heterogeneity can be recovered with a large statistical power from the sampled data. The strategy is then applied on a breast cancer data set to illustrate its use in practice where there are multiple risk factors and multiple disease characteristics of simultaneous concern.

Suing Internet Firms to Police Online Misconduct: An Empirical Study of Strategic Litigation by Secondary Stakeholders

Internet intermediary platforms are online services that provide a means for information to be hosted, shared, and transmitted between third parties. As such, they also enable the distribution of content that is ‘objectionable’ from a societal point of view. Consequently, a growing population of individuals and organizations has filed civil suits against Internet platforms as a consequence of online speech or misconduct by the third-party users of these services. Such actors comprise a platform’s “secondary stakeholders,” meaning that they neither own stock nor are part of the firm’s supply chain or customer base. Secondary stakeholders may be particularly motivated to act when they feel they are direct victims of a negative consumption externality that occurs as a result of a firm’s economic activities and thus are ‘owed’ a response. Although these 'secondary' stakeholders do not hold sway over the firm via formal or implied contractual obligations, regulatory power, or market mechanisms, the use of intermediary liability litigation affords them the ability to exercise sometimes powerful demands. I situate legal action targeting Internet intermediary platforms in the context of stakeholder theory and argue that specific attributes of the stakeholders and of their requests in intermediary liability suits may confer legitimacy and urgency which results in increased salience, or likelihood of positive responses, from judges. While the singular locus of examination in much prior work is the salience decision, I argue for the importance of addressing factors that shape stakeholder demands in order to better explicate the relationship between stakeholders, their requests, and litigation outcomes (salience). I develop a two stage framework in which I first examine the relationship between Stakeholder Attributes and Request Development and then the relationship between Request Development and Salience (or litigation outcomes). In order to test my hypotheses, I built a hand-collected database of 295 objectionable content lawsuits filed against Internet intermediary platforms in the United States between 1995 and 2014. I conduct an empirical analysis using a two-stage logistic regression model using the following measures developed over a two-year period. The type of plaintiff is coded as individual, firm, or government. Acts of online misconduct were classified on a four-point scale ranging from 1 (most likely tort) to 4 (most likely crime). Intermediary platforms were assigned to six categories which were ranked on a 6-point ordinal item measuring the category's distance (in an engineering context) from the Internet 'backbone,' or how 'downstream' it is, as a proxy for its location in the ecosystem. The farther downstream a category is, the less central it is to the network and the more visible it is to the end user. The size of the stakeholder’s request is measured in terms of ‘Duties of Care,’ a 4-point ordinal item representing the type of remedy requested by a plaintiff (information disclosure, content blocking, content filtering, service discontinuation) and the corresponding burden to the intermediary. Finally, litigation outcomes are coded (liable or not liable) for each intermediary platform. My primary findings are: 1) The more likely an act of online misconduct is to be classified as a crime rather than as a tort, the greater will be the size of the stakeholder’s request. 2) The more visible an intermediary is to users in terms of its location in the network architecture, the greater the size of the stakeholder’s request. 3) While the criminality of misconduct and the visibility of intermediary targets contribute positively to the likelihood of positive litigation outcomes for the stakeholder, the size of the stakeholder’s request is negatively correlated with the likelihood of a positive litigation outcome.

Gainful employment and risk of mortality after spinal cord injury: effects beyond that of demographic, injury and socioeconomic factors

To evaluate the association of three levels of gainful employment with the risk of mortality after traumatic spinal cord injury (SCI) while controlling for known predictors of mortality status (including education and income). Prospective cohort study. A total of 20 federally funded SCI Model Systems of care in the United States. Participants included 7955 adults with traumatic SCI. Preliminary assessments were conducted between 1995 and 2006. Mortality status was determined by the Social Security Death Index (1308 deaths). A two-stage logistic regression model was used to estimate the chance of dying in any given year. Life expectancy was calculated under different economic assumptions. Compared with those who were working 30+ h per week, the odds of mortality was 1.37 for those who worked 1-29 h and 1.67 for those who were unemployed. The addition of gainful employment only modestly reduced the effects of household income and education, both of which remained significant. For instance, the odds of mortality for household income (referent $75 000+) decreased from 1.50 to 1.38 for $25 000-$75 000 and from 2.10 to 1.82 for < $25 000. Life expectancy varied widely depending on socioeconomic characteristics more than doubling under certain assumptions. Substantial variation in mortality is attributable to employment, above and beyond the effects of previously established demographic, injury and socioeconomic predictors. Although some excess mortality may be the inevitable consequence of SCI, risk is substantially increased with poor socioeconomic characteristics.

Two-Stage Logistic Regression for Cancer Classi cation and Prediction from Copy-Numbe Changes in cDNA Microarray-Based Comparative Genomic Hybridization

cDNA microarray-based comparative genomic hybridization(CGH) data includes low-intensity spots and thus a statistical strategy is needed to detect subtle differences between different cancer classes. In this study, genes displaying a high frequency of alteration in one of the different classes were selected among the pre-selected genes that show relatively large variations between genes compared to total variations. Utilizing copy-number changes of the selected genes, this study suggests a statistical approach to predict patients' classes with increased performance by pre-classifying patients with similar genetic alteration scores. Two-stage logistic regression model(TLRM) was suggested to pre-classify homogeneous patients and predict patients' classes for cancer prediction; a decision tree(DT) was combined with logistic regression on the set of informative genes. TLRM was constructed in cDNA microarray-based CGH data from the Cancer Metastasis Research Center(CMRC) at Yonsei University; it predicted the patients' clinical diagnoses with perfect matches (except for one patient among the high-risk and low-risk classified patients where the performance of predictions is critical due to the high sensitivity and specificity requirements for clinical treatments. Accuracy validated by leave-one-out cross-validation(LOOCV) was 83.3% while other classification methods of CART and DT performed as comparisons showed worse performances than TLRM.

Two-stage logistic regression model

In this article, a logistic regression model combined with decision tree for dealing with a significant interaction effect among the explanatory variables is suggested. Decision tree is applied for investigating the interaction among explanatory variables and grouping subjects based on χ 2 value for optimal split. Each group of subjects which is named cluster is determined by optimal split for the interacting explanatory variables. The suggested model incorporates this cluster as an explanatory variable for including significant interaction in the logistic regression model. This model shows better performances in assessment of predictive model than the logistic regression model or decision tree: better ranked classes, increased correct classification rate and R 2, improved Kolmogorov–Smirnov (K–S) statistic, and a better lift. National pension data are applied to this model, and as an application of the suggested model, strategies for reducing financial risks in managing and planning for pension financing are illustrated.

Comparing four estimation methods for uninsurance in Florida

Although the high percentage of the uninsured is an important public policy issue, discrepancies in both state and national estimates of the numbers of uninsured are reported. There is a critical need to address the methodological problem of the estimation. This study compares four advanced estimation methods for uninsurance by using Florida Health Insurance Survey data as an example. The four predictive models are decision tree, neural network, general logistic regression and two-stage logistic regression. The two-stage logistic regression model is found to be the best model for imputing missing data on health insurance. Risk factors to uninsurance are identified. Corresponding policy implications are discussed.

A two-stage logistic regression model for analyzing inter-rater agreement

Studies of agreement commonly occur in psychiatric research. For example, researchers are often interested in the agreement among radiologists in their review of brain scans of elderly patients with dementia or in the agreement among multiple informant reports of psychopathology in children. In this paper, we consider the agreement between two raters when rating a dichotomous outcome (e.g., presence or absence of psychopathology). In particular, we consider logistic regression models that allow agreement to depend on both rater- and subject-level covariates. Logistic regression has been proposed as a simple method for identifying covariates that are predictive of agreement (Coughlin et al., 1992). However, this approach is problematic since it does not take account of agreement due to chance alone. As a result, a spurious association between the probability (or odds) of agreement and a covariate could arise due entirely to chance agreement. That is, if the prevalence of the dichotomous outcome varies among subgroups of the population, then covariates that identify the subgroups may appear to be predictive of agreement. In this paper we propose a modification to the standard logistic regression model in order to take proper account of chance agreement. An attractive feature of the proposed method is that it can be easily implemented using existing statistical software for logistic regression. The proposed method is motivated by data from the Connecticut Child Study (Zahner et al., 1992) on the agreement among parent and teacher reports of psychopathology in children. In this study, parents and teachers provide dichotomous assessments of a child's psychopathology and it is of interest to examine whether agreement among the parent and teacher reports is related to the age and gender of the child and to the time elapsed between parent and teacher assessments of the child.

Determinants of bank growth choice

We study the determinants of bank growth in a two-stage logistic regression model. We first compare banks that branch, Bank Acquire, or Product Expand with banks that do not grow externally. Banks that are federally chartered, in states with higher income growth, and with higher labor prices are less likely to grow externally. Larger banks are more likely to grow externally. In the second stage, we study determinants of growth activity for banks that expand products, branch, or acquire other banks. Depending on the time period, bank structure, regulatory environment, performance, and balance sheet characteristics determine bank growth choices.

Statistical models for predicting a beneficial response to interferon-alpha in patients with chronic hepatitis B.

Therapy with interferon-alpha has been reported to induce remissions in 35% of patients with chronic hepatitis B. The ability to identify patients likely to respond would be helpful in making recommendations for treatment. In this statistical analysis we included 82 patients with chronic hepatitis B who received interferon-alpha in clinical trials at the National Institutes of Health between 1984 and 1991. A response was defined as the loss of hepatitis B virus (HBV) DNA and hepatitis B e antigen (HBeAg) within 1 year of therapy. Multiple clinical parameters measured at pretreatment (month 0) and after the first month (month 1) of therapy were selected by stepwise regression to support the development of the prognostic models: the two-stage logistic regression model and a neural network that utilized higher-order non-linear interactions between variables. Among the 82 patients, 24 (29%) were responders. The two-stage logistic model using pretreatment variables: sex, hepatic fibrosis and alanine aminotransferase (ALT) levels correctly identified 61% of responders and 76% of non-responders. When HBV DNA at month 1 along with sex, initial ALT and fibrosis was included, the resultant model correctly identified 69% of responders and 77% of non-responders. The neural network, by incorporating interactions between variables, correctly identified 77% and 86% of responders, and 87% and 92% of non-responders, using pretreatment factors alone and the combination of pretreatment and month 1 factors respectively. Hence, the neural network was more accurate than the simple logistic regression model in predicting a response to interferon-alpha in chronic hepatitis B. The universality of these models needs to be further verified.

Health effects of acid aerosols on North American children: respiratory symptoms.

We examined the respiratory health effects of exposure to acidic air pollution among 13,369 white children 8 to 12 years old from 24 communities in the United States and Canada between 1988 and 1991. Each child's parent or guardian completed a questionnaire. Air quality and meteorology were measured in each community for a 1-year period. We used a two-stage logistic regression model to analyze the data, adjusting for the potential confounding effects of sex, history of allergies, parental asthma, parental education, and current smoking in the home. Children living in the community with the highest levels of particle strong acidity were significantly more likely [odds ratio (OR) = 1.66; 95% confidence interval (CI) 1.11-2.48] to report at least one episode of bronchitis in the past year compared to children living in the least-polluted community. Fine particulate sulfate was also associated with higher reporting of bronchitis (OR = 1.65; 95% CI 1.12-2.42). No other respiratory symptoms were significantly higher in association with any of the air pollutants of interest. No sensitive subgroups were identified. Reported bronchitis, but neither asthma, wheeze, cough, nor phlegm, were associated with levels of particle strong acidity for these children living in a nonurban environment.