Partial Proportional Odds Model Research Articles

Understanding work enjoyment among older workers: The significance of flexible work options and age discrimination in the workplace

ABSTRACTAlthough the number of older workers in the U.S. is increasing, there is a gap in knowledge on whether or not they actually enjoy working. This study, based on a conceptual framework focusing on job resources and demands, explored likely workplace determinants of work enjoyment among older workers aged 50 or over. Using the 2012 wave of the Health and Retirement Study, a partial proportional odds model was used to detect determinants of work enjoyment. Results showed that higher levels of work enjoyment were significantly and negatively associated with the level of perceived retirement pressure and promotion preference for younger workers, and positively associated with moving to less demanding positions. Self-employment showed a noticeable enhancement of work enjoyment. This study highlights the significance of flexible work options and age discrimination in the workplace in understanding work enjoyment later in life.

Modeling and comparing injury severity of at-fault and not at-fault drivers in crashes

This paper examines and compares the effect of selected variables on driver injury severity of, both, at-fault and not at-fault drivers. Data from the Highway Safety Information System (HSIS) for the state of North Carolina was used for analysis and modeling. A partial proportional odds model was developed to examine the effect of each variable on injury severity of at-fault driver and not at-fault driver, and, to examine how each variable affects these two drivers’ injury severity differently. Road characteristics, weather condition, and geometric characteristics were observed to have a similar effect on injury severity in a crash to at-fault and not at-fault drivers. Age of the driver, physical condition, gender, vehicle type, and, the number and type of traffic rule violations were observed to play a significant role in the injury severity of not at-fault drivers when compared to at-fault drivers in the crash. Moreover, motorcyclists and drivers 70 years or older are observed to be the most vulnerable road users.

Understanding Characteristics of Victims and Perpetrators of Anti-LGBT Hate Crimes in the United States.

This article examines the effects of victim and perpetrator characteristics on the seriousness of anti-lesbian, gay, bisexual (LGB) and anti-transgender (T) hate crime incidents, as there are limited and contradictory findings regarding characteristics that contribute to anti-LGB(T) hate crimes. This analysis uses a series of χ2 tests to examine the importance of these characteristics using data from the 2003-2013 National Crime Victimization Survey (NCVS), and a partial proportional odds model for data from the 2012 Uniform Crime Report (UCR) Hate Crime Data to examine differences in seriousness among LGB and T victims. This study finds that victim characteristics may not impact victimization in the ways suggested by previous studies. Additionally, the importance of offender race is largely impacted by whether the data are self-report data or police data.

PIN57 - Prior Perceptions and Subsequent Decisional Conflict Regarding HIV Cure Trial Participation in Bangkok, Thailand

Studies of clinical and research decision making and decision satisfaction often measure participant-reported decision conflict as an outcome. As part of our longitudinal study in an HIV cohort, we examine the relationship between pre-existing perceptions and subsequent decision conflict among those accepting and declining trial participation. We conducted a baseline questionnaire in a cohort of individuals diagnosed with acute HIV in Bangkok, Thailand, prior to their recruitment to a cure trial. Questions included respondents’ perceptions of anticipated benefits and harms of HIV cure trials, and a rating of their intention to participate in a hypothetical trial that includes interruption of antiretroviral therapy. During the following 16-24 months, individuals were offered participation in one of three HIV cure trials. Joiners and decliners completed the Decisional Conflict Scale (DCS) during longitudinal interviews starting at participation offer and concluding shortly after trial completion. We categorized participants into high, medium, or low decisional conflict levels based on their most recent DCS score. Using ordinal logistic regression, we examined the relationship between decisional conflict category and pre-existing perceptions using a partial proportional odds model. Of 250 participants completing the baseline questionnaire, 74 have been offered participation in one of three trials, and consented to interviews. Baseline assessment of potential trial participation benefit to the individual, potential benefit to science, and the participants’ intention to participate in a trial all were correlated with decision conflict score, while potential risk to self was not. When participants’ most recent decision conflict scores were categorized, ordinal regression indicated the strongest association between intention to participate and subsequent decision conflict category. Participant or decliner status was not associated with decision conflict. Our results suggest a strong impact of baseline intention to participate on later decision conflict for trial joiners and decliners, consistent with several other studies.

The impacts of heavy rain on speed and headway Behaviors: An investigation using the SHRP2 naturalistic driving study data

Adverse weather conditions can significantly impact roadways by influencing roadway conditions, vehicle performance and driver behavior. Vehicle user characteristics and behavior can be considered as the most important factors affecting the driving task. The ability to see objects in motion, so called “dynamic visual acuity”, and the proper reaction process, such as headway and speed selection, are imperative factors for safe driving. In this study, data from the SHRP2 naturalistic driving study (NDS) are used to provide better understanding of driver speed and headway selection behaviors in clear and rainy weather conditions. A unique procedure to identify rain-related trips from the massive SHRP2 database was introduced in this study. In addition, roadway information database (RID) and NDS were utilized to compare driver behavior in clear and heavy rain conditions using matching trips. Matching trips were defined as trips with same driver, same vehicle, and same traversed routes. Preliminary descriptive statistics, partial proportional odds model, as well as geographical information system analyses showed significant differences between driver behavior and performance in clear and rainy weather conditions. One interesting finding of this research is that drivers were less likely to drive above the speed limits on road segments with higher posted speed limits. In addition, it was found that the probability of reducing speed more than 5 kph below the speed limits were 23% and 29% higher in light rain and heavy rain, respectively. Not only will the findings of the study help in providing better insights on drivers’ behavior and performance in rainy weather conditions, but it will also serve as a foundation for further studies to investigate driver behavioral factors in other weather conditions using naturalistic driving data.

Abstract WP89: A Functional Outcome Model for a Telestroke-guided Tissue Plasminogen Activator Treatment of Stroke Patients

Background: The success of telestroke systems as a means to extend stroke-care expertise into rural and underserved areas has been reported. Evidence of clinical outcomes associated with the use of tissue-type plasminogen activator (rtPA) in community hospitals has been documented, but evidence of functional outcome studies in telestroke-guided rtPA treatment are lacking. This issue was investigated in this study. Methods: Data from a stroke registry of 307 acute ischemic patients treated with intravenous rtPA in a telestroke network was analyzed. A new functional outcome model was developed to asses rtPA treatment outcome using a combination of visual basic and multivariate logistic models. Differences in functional outcomes between patients that received rtPA and those did not receive rtPA were determined. A partial proportional-odds model was developed for common odd ratio based on cumulative logits for all functional outcome variables. The significance level of all analysis was set to a probability level of p<0.05. Results: In a telestroke-guided rtPA treatment of stroke patients, age was significantly (P<0.0001, OR 0.902) associated with improved functional outcome. Stroke patients with a medical history of obesity (P=0.002, OR 4.016), and a previous history of stroke (P=0.024, OR 2.765) were associated with improved functional outcomes. The odd of an improved functional outcome was higher for stroke patients with systolic blood pressure (P=0.012, OR 3.135) as the initial vital sign. Moreover, patients with initial lower NIH stroke scale scores (P<0.001, OR 0.911) were associated with improved functional outcome. Conclusion: Modified Rankin Scale distribution and other measures have been used to measure clinical outcomes in rtPA-treated patients. The development of a new functional outcome model provides a new approach to improve the efficiency of the telestroke and increase of rtPA for stroke patients.

Correlates of expected eMental Health intervention uptake among Filipino domestic workers in China.

Transnational migrant populations face critical barriers to mental health service utilization that perpetuate mental health disparities globally. Overseas Filipino workers (OFWs) number over 2 million globally and 25% are female domestic workers. Structural barriers prevent equitable access to mental health services for this population. Electronic mental health (eMental Health) intervention is a scalable alternative to face-to-face treatment. The current study sought to identify key correlates of intention to use eMental Health within a community of female Filipino domestic workers living and working in Macao (SAR), China. Respondent-driven sampling implemented at a community field site was used to reach a sample of 1364 female domestic workers. A multivariable adjusted partial proportional-odds (PPO) model was used to assess relevant correlates of intent to use eMental Health. The majority (62.8%) reported being likely to utilize eMental Health. The adjusted PPO model showed that younger age (18-25, 26-35, 36-45 v. over 55), longer time as an OFW, being likely (v. neutral and unlikely) to seek professional services, willingness to pay for services (v. not), belief that mental health services are a priority (v. low priority), having access to Wi-Fi outside the employer's home (v. not), and higher levels of social support were associated with increased odds of intent to use eMental Health. eMental Health is a promising intervention with high potential for uptake among OFWs. The majority of the study population owned a smartphone and were able to connect to the Internet or Wi-Fi. Future work will rigorously evaluate eMental Health programs for use among OFWs.

Precision Medicine for Cardiac Resynchronization: Predicting Quality of Life Benefits for Individual Patients-An Analysis From 5 Clinical Trials.

Clinical trials have established the average benefit of cardiac resynchronization therapy (CRT), but estimating benefit for individual patients remains difficult because of the heterogeneity in treatment response. Accordingly, we created a multivariable model to predict changes in quality of life (QoL) with and without CRT. Patient-level data from 5 randomized trials comparing CRT with no CRT were used to create a prediction model of change in QoL at 3 months using a partial proportional odds model for no change, small, moderate, and large improvement, or deterioration of any magnitude. The C statistics for not worsening or obtaining at least a small, moderate, and large improvement were calculated. Among the 3614 patients, regardless of assigned treatment, 33.3% had a deterioration in QoL, 9.2% had no change, 9.2% had a small improvement, 13.5% had a moderate improvement, and the remaining 34.9% had a large improvement. Patients undergoing CRT were less likely to have a decrement in their QoL (28.2% versus 38.9%; P<0.001) and more likely to have a large QoL improvement (38.7% versus 30.6%; P<0.001). A partial proportional odds model identified baseline QoL, age, and an interaction of CRT with QRS duration as predictors of QoL benefits 3 months after randomization. C statistics of 0.65 for not worsening, 0.68 for at least a small improvement, 0.69 for at least a moderate improvement, and 0.73 for predicting a large improvement were observed. There is marked heterogeneity of treatment benefit of CRT that can be predicted based on baseline QoL, age, and QRS duration.

Application of partial proportional odds model for analyzing pedestrian crash injury severities in Switzerland

ABSTRACTAccording to the Swiss Microcensus, Switzerland is a place where people choose to walk more than 40% of their daily trip time, resulting in a higher pedestrian crash to total crash ratio when compared to many other developed countries. Furthermore, the high number of old and young pedestrians as well as the large number of pedestrian crashes with higher levels of severity makes the pedestrian crash analysis for Switzerland very important. This study aims to identify the important factors that influence the injury severity levels of pedestrian–vehicle crashes in Switzerland. To address this we employ a partial proportional odds model to analyze pedestrian crash injury severities in Switzerland, based on the 5-year (2008–2012) pedestrian safety data. The study also focuses on a separate analysis for two vulnerable groups of pedestrians (young and old pedestrians) to examine the change in the influence of factors over age. The analysis involves an extensive evaluation based on pedestrian and driver characteristics; main reason for the crash; traffic, temporal, vehicle, roadway geometric and environmental characteristics. The study shows that irrespective of age, dark unlighted areas, sight obstructions, midblock crossings and heavy vehicles adversely affect the injury severity levels of pedestrians.

Racial identity and the quality of life among blacks and whites in the U.S.

Social identity theory and research on mental health among racial minority groups suggest that a stronger, more positive racial identity will be related to a higher subjective quality of life. We investigate how ingroup closeness, ingroup evaluation, and ingroup bias are associated with happiness, positive affect about life, and generalized trust for blacks and whites, using partial proportional odds models. Data came from the 1996–2014 General Social Surveys (N = 6553). Ingroup closeness and more favorable ingroup evaluation had mostly positive associations with the quality of life dimensions. Contrary to what social identity theory would predict, ingroup bias was either unrelated or negatively related to them. Racial identity functions somewhat differently for blacks and whites. Ingroup evaluation and ingroup bias were related to greater positive affect about life for blacks but lower positive affect about life for whites.

GEEORD: A SAS macro for analyzing ordinal response variables with repeated measures through proportional odds, partial proportional odds, or non-proportional odds models

Background and objectiveA SAS macro, GEEORD, has been developed for the analysis of ordinal responses with repeated measures through a regression model that flexibly allows the proportional odds assumption to apply (or not) separately for each explanatory variable. Methods and resultsPreviously utilized in an analysis of a longitudinal orthognathic surgery clinical trial by Preisser et al. [1,2], the basis of GEEORD is the generalized estimating equations (GEE) method for cumulative logits models described by Lipsitz et al. [3]. The macro extends the capabilities for modeling correlated ordinal data of GEECAT, a SAS macro that allows the user to model correlated categorical response data [4]. The macro applies to independent ordinal responses as a special case. Applications and conclusionsExamples are provided to demonstrate the convenient application of GEEORD to two different datasets. The macro's features are illustrated in fitting models to ordinal response variables in univariate and repeated measures settings; this includes the capacity to fit the non-proportional odds model, the partial proportional odds model, and the proportional odds model. The macro additionally provides relevant tests of the proportional odds assumption.

Modelling single-vehicle, single-rider motorcycle crash injury severity: an ordinal logistic regression approach

ABSTRACTMotorcycles represent an increasing proportion of traffic fatalities in the United States, accounting for more than 12.7% of the total traffic casualties within 2005–2014. Specifically, in North Carolina, motorcycles comprise less than 1% of vehicles involved in crashes but account for more than 7% of total fatalities, representing a top state in the United States. This study tries to investigate the motorcycle crashes in North Carolina more in depth. In doing so, five years’ (2009–2013) worth of crash data was obtained from the Federal Highway Administration’s Highway Safety Information System database. A partial proportional odds (PPO) logistic regression model was developed to examine the influence of the explanatory variable on the ordered dependent variable, that is, injury severity. Moreover, two other popular ordered-response models, that is, proportional odds and non-proportional odds models, as well as one similar unordered-response model, that is, multinomial logit model, were also developed to evaluate their performances compared to the PPO model. Older riders, DUI riding, not wearing helmets, crashes during summer and weekends, darkness, crashes with fixed objects, reckless riding, and speeding were found to increase the severity of injuries. In contrast, younger riders, winter season, adverse weather condition, and wet surface were associated with lower injury severities. Furthermore, crashes in rural areas, overturn/rollover, and crashes happened while negotiating a curve showed fluctuating effects of injury severity. According to two information criteria calculated for all three developed models fitted to the same data, the PPO model was found to outperform the other models and provide more reliable results. Based on the obtained average direct pseudo-elasticities, this study determines the effect of the various identified variables and develops several safety countermeasures as a resource for policy-makers to prevent or mitigate the severity of motorcycle crashes in North Carolina.

London Measure of Unplanned Pregnancy: guidance for its use as an outcome measure.

BackgroundThe London Measure of Unplanned Pregnancy (LMUP) is a psychometrically validated measure of the degree of intention of a current or recent pregnancy. The LMUP is increasingly being used worldwide, and can be used to evaluate family planning or preconception care programs. However, beyond recommending the use of the full LMUP scale, there is no published guidance on how to use the LMUP as an outcome measure. Ordinal logistic regression has been recommended informally, but studies published to date have all used binary logistic regression and dichotomized the scale at different cut points. There is thus a need for evidence-based guidance to provide a standardized methodology for multivariate analysis and to enable comparison of results. This paper makes recommendations for the regression method for analysis of the LMUP as an outcome measure.Materials and methodsData collected from 4,244 pregnant women in Malawi were used to compare five regression methods: linear, logistic with two cut points, and ordinal logistic with either the full or grouped LMUP score. The recommendations were then tested on the original UK LMUP data.ResultsThere were small but no important differences in the findings across the regression models. Logistic regression resulted in the largest loss of information, and assumptions were violated for the linear and ordinal logistic regression. Consequently, robust standard errors were used for linear regression and a partial proportional odds ordinal logistic regression model attempted. The latter could only be fitted for grouped LMUP score.ConclusionWe recommend the linear regression model with robust standard errors to make full use of the LMUP score when analyzed as an outcome measure. Ordinal logistic regression could be considered, but a partial proportional odds model with grouped LMUP score may be required. Logistic regression is the least-favored option, due to the loss of information. For logistic regression, the cut point for un/planned pregnancy should be between nine and ten. These recommendations will standardize the analysis of LMUP data and enhance comparability of results across studies.

Estimation of Crash Severity on Mountainous Freeways in Chongqing

Mountainous freeways always suffer from accidents due to special terrain, weather conditions, driving environment, and so on. Based on the records of 898 accidents that occurred on mountainous freeways in Chongqing during the past 6 years, the partial proportional odds model is used to identify the factors affecting the accident severity. The time of the accident, season, involvement of trucks, accident characteristics, speeding, maximum driving experience of involved drivers, and weather and road conditions are found to be important for the levels of accident severity. Zero to 6 a.m. and 19 to 24 p.m. are the times prone to serious traffic accidents. The probability of serious traffic accidents in summer and autumn is greater than that in spring and winter. Once a truck is involved in an accident, the consequence is often more severe. Turnover and speeding will result in a grave accident. When there is an experienced driver, the probability of serious traffic accidents is low. The fog is extremely unfavorable weather conditions. The probability of serious accident happening in the downgrade, ramp, curve, bridge, and tunnel sections is greater than the others. The results aim to provide valuable reference for traffic safety on mountainous freeways.

Farmers' Perception about Soil Erosion in Ethiopia

AbstractSoil erosion is a significant problem in the Ethiopian highlands. The objective of this study was to investigate how farmers perceive the severity of soil erosion in the Upper Blue Nile Basin. The study is based on a detailed survey of 300 households and 1,010 plots owned by these households in three watersheds. Descriptive statistics and a partial proportional odds model were applied to analyze factors that affected farmers' perceived soil erosion severity at the plot‐level. Results showed that variables such as plot distance from the residence, plot shape and position on hill slopes affected farmers' perceptions of soil erosion severity, as well as the amount of rainfall during the growing season. Farmer interaction with extension service agents also affected farmers' perception of soil erosion severity. Despite their expected importance, education and number of livestock owned had no effect on the farmers' perception of soil erosion. The results indicate that farmers' perceptions generally match empirical and theoretical findings on soil erosion determinants; thus, farmers should be considered as important partners not only to counter soil erosion, but also to obtain local expertise on soil erosion severity and restoration of degraded land. © 2016 The Authors. Land Degradation and Development published by John Wiley &amp; Sons, Ltd.

Estimates of Disease Course in Inflammatory Bowel Disease Using Administrative Data: A Population-level Study.

We sought develop a predictive model of disease course in inflammatory bowel disease [IBD] using health care utilization measures from administrative health data, and to apply this model to estimate disease course at a population level over time. Study participants were IBD patients who were prospectively followed for a 1-year period between 2009 and 2010 in a Canadian clinic setting to assess their IBD disease course [i.e. remission, mild, moderate, severe]. Clinic data were linked with population-based administrative health data. A multivariable partial proportional odds model tested health care utilization measures that discriminated disease course groups. The model was applied to project the distribution of disease course for the Manitoba IBD population for 1995-2013. There were 407 participants (54.3% females, 64.4% Crohn's disease [CD]) with mean age at diagnosis of 29.8 years [SD 14.9]. Forty-one per cent of participants were clinically in remission, while 14.0% had severe IBD. Mild, moderate or severe disease was associated with three or more gastroenterologist visits (odds ratio [OR] = 3.33, 95% confidence interval [CI]: 2.03-5.54) or three or more general practitioner visits [OR = 2.97, 95% CI: 1.44-6.37] with an IBD diagnosis and ≥1 radiology test [OR = 2.22, 95% CI: 1.31-3.80]. The percentages of the Manitoba IBD population in remission rose steadily from 1995 to 2013 [43.6 to 59.9%], while the percentages of individuals with mild, moderate or severe disease declined. This study demonstrated that health care utilization measures from administrative data can be used to predict disease course in the IBD population.

Drinking, Cigarette Smoking, and Employment Among American College Freshmen at a Four-Year University

ABSTRACTBackground: For American college students, alcohol and cigarette use are important health concerns, and employment concurrent with school attendance is on the rise. Given the lifelong importance of employment and substance use trajectories begun in college, parsing out the relationship between the two is meaningful. Objectives: This study's purpose is to determine whether employment during college is associated with substance use. Methods: Cross-sectional associations between employment (work hours, earnings) and substance use (drinking frequency, drinking quantity, smoking frequency) were estimated using partial proportional odds models in a sample of N = 1457 freshmen attending a large, public 4-year university in 2011, after accounting for demographics, personality, social environment, and parental influences. Results: Working 10 more hours and earning $50 more per week as a freshman had modest positive associations with higher smoking frequency and with moderate drinking frequency and quantity prior to adjustment. After adjustment, work hours remained modestly associated with moderate drinking frequency and quantity. No adjusted associations were found among employment measures and smoking or between weekly earnings and drinking frequency. Different relationships emerged for moderate versus heavy alcohol use frequency and quantity. Conclusions: Both employment and substance use are commonplace among college freshmen. After extensive controls for potential confounders, the relationship between the two appears modest. Employment may yet play a role in college student substance use, but work hours and earnings are likely only small parts of a larger web of influences on drinking and smoking.

On association between diabetes status and stature of individual in Bangladesh: an ordinal regression analysis

Contravening findings are found in the literature about the association between individuals stature and diabetes status. This paper attempts to find evidence regarding the nature and strength of this association by analyzing a nationally representative data of Bangladesh. Under ordinal regression setup, a partial proportional odds model, a variant of the proportional odds model, has been found to fit better compared to the proportional odds model for analyzing diabetes status in Bangladesh. A strong evidence of inverse relationship is found between diabetes and height. Gender, education level, wealth index, place of residence, division and type of work are also found to be significantly associated with diabetes.

Analysis of driver injury severity in wrong-way driving crashes on controlled-access highways

For more than five decades, wrong-way driving (WWD) has been notorious as a traffic safety issue for controlled-access highways. Numerous studies and efforts have tried to identify factors that contribute to WWD occurrences at these sites in order to delineate between WWD and non-WWD crashes. However, none of the studies investigate the effect of various confounding variables on the injury severity being sustained by the at-fault drivers in a WWD crash. This study tries to fill this gap in the existing literature by considering possible variables and taking into account the ordinal nature of injury severity using three different ordered-response models: ordered logit or proportional odds (PO), generalized ordered logit (GOL), and partial proportional odds (PPO) model. The findings of this study reveal that a set of variables, including driver’s age, condition (i.e., intoxication), seatbelt use, time of day, airbag deployment, type of setting, surface condition, lighting condition, and type of crash, has a significant effect on the severity of a WWD crash. Additionally, a comparison was made between the three proposed methods. The results corroborate that the PPO model outperforms the other two models in terms of modeling injury severity using our database. Based on the findings, several countermeasures at the engineering, education, and enforcement levels are recommended.

Modeling severity of single vehicle run-off-road crashes in rural areas: model comparison and selection

Run-off-road (ROR) crashes account for a large proportion of fatalities and serious injuries to vehicle occupants, especially in rural areas. While performing crash severity analysis using discrete choice models (DCMs), researchers may be confused by the following questions: first, should an ordered or unordered model structure be used and secondly, which modeling level is more appropriate, basic or advanced? A model selection framework is developed considering the following factors: (1) model structure — ordered or unordered; (2) intrinsic deficiency of each model; (3) computational burdens; (4) complexity of parameter interpretation; and (5) model fitness. Historical ROR crash data were utilized to illustrate how to choose an appropriate DCM based on the proposed framework. Using statistical tests and comparison of evaluation and validation measurements, both the mixed logit model and the partial proportional odds model yield a reasonable performance. All factors that significantly affect the severity level of a single-vehicle ROR crash were identified as well.