Latent Class Cluster Approach Research Articles

Driver Psychology Latent Classes as Predictors of Traffic Incident Occurrence in Naturalistic Driving Study Data

Crash risk depends on several factors, driver factors being the most significant of all. Previous studies have tended to use roadway and driver demographic information to explain crash risk, overlooking driver psychological characteristics, which are also important for crash risk estimation. Crash data from police reports are available only for reportable crashes and do not detail driver characteristics or comparable information on driving activities. Naturalistic driving studies (NDSs) offer unique opportunities to obtain information about driver attributes, behavior, or other precrash factors for predicting crash occurrence. This study estimated NDS event-oriented models to evaluate the interaction between driver attributes and roadway environmental factors for predicting safety critical events. A latent class clustering approach was used to uncover categories of drivers by psychological, perceptual, and cognitive characteristics, and by driving experience. The results revealed four driver types: risk-taker, careful-impaired, careful-unimpaired, and distractible. These types were incorporated in mixed-effects binary logistic models, with roadway, traffic, and environmental variables to estimate and predict crash risk. The models that included driver factors more successfully predicted crash risk than those without. Risk-takers showed the highest probability of being in crashes. However, careful-impaired drivers—those whose impairments made it difficult to identify the location of another vehicle, visualize missing information, who had difficulties with visual–spatial perception and executive functioning—posed a higher crash risk in roadway conditions such as snow, lack of lane markings, and certain traffic operating conditions. The results point to novel avenues for educational and behavioral interventions to improve road safety.

Perceptions of the impacts of urban sprawl among urban and peri-urban dwellers of Hyderabad, India: a Latent class clustering analysis

Like many other developing countries, urban sprawl is a growing phenomenon in India, which poses socio-economic and environmental challenges that worryingly affect urban sustainability. In this study, a latent class clustering approach was used to investigate perceptions of urban sprawl among 622 urban and peri-urban dwellers in Hyderabad. The empirical results clustered the respondents into three distinct classes based on their perceptions of urban sprawl impacts: ‘undecided respondents’, ‘negative perceivers’, and ‘opportunity perceivers’. The majority of respondents were undecided with no strong views towards the impacts of urban sprawl, which may increase their vulnerability and hinder effective adaptation to the adverse economic, social and environmental effects of urban sprawl. This also provokes concerns about the effectiveness of government interventions to build public awareness of urban development and its impacts on the city. With regard to the role of demographic and socio-economic characteristics in shaping the perception of the respondents, the results revealed that social caste plays a determining role in forming dwellers’ perception. In particular, members of marginalised social castes were more likely to form positive perceptions of the impacts of urban sprawl as urban expansion generates better and stable income that improve their social status. In addition, individuals with higher levels of education were more likely to form negative or positive perceptions, implying that efforts to raise social capital could be a useful means for mitigating the impacts of urban sprawl. Finally, membership in community development organisations was a key factor in dictating membership of the negative perceivers’ class. Overall, our findings suggest that an appropriate policy framework and specific programmes are needed for enhancing dwellers’ perception towards the impacts of urban sprawl, which can enhance the design, acceptance, and implementation of a more sustainable governance of urbanisation and contribute to achieving urban sustainability in developing countries.

EVALUASI KINERJA METODE CLUSTER ENSEMBLE DAN LATENT CLASS CLUSTERING PADA PEUBAH CAMPURAN

Most of the traditional clustering algorithms are designed to focus either on numeric data or on categorical data. The collected data in the real-world often contain both numeric and categorical attributes. It is difficult for applying traditional clustering algorithms directly to these kinds of data. So, the paper aims to show the best method based on the cluster ensemble and latent class clustering approach for mixed data. Cluster ensemble is a method to combine different clustering results from two sub-datasets: the categorical and numerical variables. Then, clustering algorithms are designed for numerical and categorical datasets that are employed to produce corresponding clusters. On the other side, latent class clustering is a model-based clustering used for any type of data. The numbers of clusters base on the estimation of the probability model used. The best clustering method recommends LCC, which provides higher accuracy and the smallest standard deviation ratio. However, both LCC and cluster ensemble methods produce evaluation values that are not much different as the application method used potential village data in Bengkulu Province for clustering.

Understanding Filipino Rice Farmer Preference Heterogeneity for Varietal Trait Improvements: A Latent Class Analysis.

Using an experimental methodology based on investment games, we examine whether smallholder rice farmers from Nueva Ecija, Philippines have heterogeneous preferences for improvements in 10 rice varietal traits. We use a latent class cluster approach to identify different segments of rice producing households and their distinct preferences for trait improvements. These clusters were characterised post hoc using household, farm, and marketing characteristics. On average, farmers invested the most in rice varietal trait improvements that offered opportunities to reduce losses caused by lodging, insects and diseases. We found four classes of farmers with distinct preferences for improvements in variety traits. The clusters were significantly different in terms of household and farm characteristics. These findings can guide breeding research in the development of varieties that have the traits farmers identified for improvement, and that will address the unique needs of distinct farmer segments.

Exploring the severity of bicycle–vehicle crashes using latent class clustering approach in India

IntroductionBicyclists are vulnerable users in the shared asset like roadways. However, people still prefer to use bicycles for environmental, societal, and health benefits. In India, the bicycle plays a role in supporting the mobility to more people at lower cost and are often associated with the urban poor. Bicyclists represents one of the road user categories with highest risk of injuries and fatalities. According to the report by the Ministry of Road Transport and Highways (Accidents, 2017) in India, there is a sharp increase in the number of fatal victims for bicyclists in 2017 over 2016. The number of cyclists killed jumped from 2,585 in 2016 to 3,559 in 2017, a 37.7% increase. Method: Few studies have only investigated the crash risk perceived by the bicyclists while interacting with other road users. The present paper investigates the injury severity of bicyclists in bicycle-vehicle crashes that occurred in the state of Tamilnadu, India during the nine year period (2009–2017). The analyses demonstrate that dividing bicycle-vehicle collision data into five clusters helps in reducing the systematic heterogeneity present in the data and identify the hidden relationship between the injury severity levels of bicyclists and cyclists demographics, vehicle, environmental, temporal cause for the crashes. Results: Latent Class Clustering (LCC) approach was used in the present study as a preliminary tool for the segmentation of 9,978 crashes. Later, logistic regression analysis was used to identify the factors that influence bicycle crash severity for the whole dataset as well as for the clusters that were obtained from the LCC model. Results of this study show that combined use of both techniques reveals further information that wouldn’t be obtained without prior segmentation of the data. Few variables such as season, weather conditions, and light conditions were significant for certain clusters that were hidden in the whole dataset. This study can help domain experts or traffic safety researchers to segment traffic crashes and develop targeted countermeasures to mitigate injury severity.

Influence of senior housing types on cognitive decline and nursing home admission among lower-income older adults

Objectives: Focusing on unique ageing populations in subsidized senior housing for lower-income older adults, this study contributes to literature on housing and aging; provides initial understanding of existing housing types; and explores the extent to which living in different housing types may influence changes in cognitive function and likelihood of nursing home admission.Method: Data came from seven waves (2002–2014) of the Health and Retirement Study. A latent-class clustering approach was used to identify senior-housing types among lower-income older people;Results: We identified four discernible housing types among lower-income older adults: (1) High physical & Low service, (2) Low physical & Low service, (3) High physical & High service, and (4) Medium physical & High service. Individuals in Medium physical & High service and High physical & Low service types were likely to have higher cognitive-function levels at baseline (B = 0.58, p < .001; 0.58, p < .001) and slower rates of decline over time (B = 0.42, p < .001; B = 0.32, p < .01). Older adults in High physical & High service are significantly less likely to be admitted to a nursing home (OR = 0.55, p < .00).Conclusion: The mismatch between health needs and lack of service and support suggests that current residents in each housing type relocate, based on knowledge of subsidized housing or availability. Future studies should examine possible mismatches between health needs and housing environment.

Investigating injury severities of motorcycle riders: A two-step method integrating latent class cluster analysis and random parameters logit model

Due to the wide existence of heterogeneous nature in traffic safety data, traditional methods used to investigate motorcyclist rider injury severity always lead to masking of some underlying relationships which may be critical for the formulation of efficient safety countermeasures. Instead of applying one single model to the whole dataset or focusing on pre-defined crash types as done in previous studies, the present study proposes a two-step method integrating latent class cluster analysis and random parameters logit model to explore contributing factors influencing the injury levels of motorcyclists. A latent class cluster approach is first used to segment the motorcycle crashes into relatively homogeneous clusters. A mixed logit model is then elaborately developed for each cluster to identify its unique influential factors. The analysis was based on the police-reported crash dataset (2015–2017) of Hunan province, China. The goodness-of-fit indicators and the Receiver Operating Characteristic curves show that the proposed method is more accurate when modeling the riders’ injury severities. The heterogeneity found in each homogeneous subgroup supports the application of the random parameters logit model in the study. More importantly, the results demonstrate that segmenting motorcycle crashes into relatively homogeneous clusters as a preliminary step helps to uncover some important influencing factors hidden in the whole-data model. The proposed method is proved to have great potential for accounting for the source of heterogeneity. The injury risk factors identified in specific cases provide more reliable information for traffic engineers and policymakers to improve motorcycle traffic safety.

Modelling severity of pedestrian-injury in pedestrian-vehicle crashes with latent class clustering and partial proportional odds model: A case study of North Carolina

There are more than 2000 pedestrians reported to be involved in traffic crashes with vehicles in North Carolina every year. 10%–20% of them are killed or severely injured. Research studies need to be conducted in order to identify the contributing factors and develop countermeasures to improve safety for pedestrians. However, due to the heterogeneity inherent in crash data, which arises from unobservable factors that are not reported by law enforcement agencies and/or cannot be collected from state crash records, it is not easy to identify and evaluate factors that affect the injury severity of pedestrians in such crashes. By taking advantage of the latent class clustering (LCC), this research firstly applies the LCC approach to identify the latent classes and classify the crashes with different distribution characteristics of contributing factors to the pedestrian-vehicle crashes. By considering the inherent ordered nature of the traffic crash severity data, a partial proportional odds (PPO) model is then developed and utilized to explore the major factors that significantly affect the pedestrian injury severities resulting from pedestrian-vehicle crashes for each latent class previously obtained in the LCC. This study uses police reported pedestrian crash data collected from 2007 to 2014 in North Carolina, containing a variety of features of motorist, pedestrian, environmental, roadway characteristics. Parameter estimates and associated marginal effects are mainly used to interpret the models and evaluate the significance of each independent variable. Lastly, policy recommendations are made and future research directions are also given.

A Latent Class Approach for Classifying the Problem and Disordered Gamers in a Group of Adolescence.

Gaming disorder is not yet recognized as a formal psychiatric disorder, and consensus is still lacking in the field concerning the definition of gaming disorder and what methods should be used to measure it. In order to deal with methodological challenges related to previously suggested approaches, the aim of the present study was to develop an alternative assessment procedure for gaming disorder using a latent class cluster approach, and to compare the criterion validity of this procedure with existing assessment procedures. A representative sample of 3,000 adolescents (n = 1,500 female) aged 17.5 years was drawn from the National Registry, and 2,055 participants responded (yielding a response rate of 70.3%). The Gaming Addiction Scale for Adolescents was used to measure gaming disorder and measures of loneliness, anxiety, depression, and aggression were used to test criterion validity. A model with five Latent Class Clusters represented the best fit [BIC(LL) = 21,253,7; L2 = 3,881,204; df = 1,978; Class. Err. = 0.1239]. The five different groups were labeled never symptoms (46.2%), rarely symptoms (22.3%), occasionally symptoms (23.5%), problem gamers (6.9%), and disordered gamers (1.2%). The groups displayed different probabilities of responses (never/rarely/sometimes/often/very often) to the seven Gaming Addiction Scale items. Regarding criterion validity, MANOVA revealed a significant overall main effect of latent classes [F(20, 6359) = 13.50, p < 0.001; Wilks Lambda = 0.871]. All dependent variables (loneliness, depression, anxiety, verbal, and physical aggression) reached statistical significance when results from the dependent variables were considered separately. Comparing the present approach with previous suggested classifications of gaming addiction offered by Lemmens et al. and Charlton and Danforth, the present approach showed greater specificity in terms of number of classes identified. We conclude that the Latent Class approach identifying five different groups of gamers offers a more refined view on addiction compared to previous assessment procedures.

Exploring differential health effects of work stress: a latent class cluster approach.

BackgroundWhile evidence highlights the detrimental health consequences of adverse working conditions, effect sizes vary by the stressor examined. In this study, we aimed to explore the differential effects various constellations of job demands have on prevalent symptom clusters.MethodsWe analysed self-reported data from a nationwide Austrian survey (N = 16,466), based on a cross-sectional design. By means of latent class analysis, a set of items was used to assess the burden from several job demands as well as the frequency of occurrence of mental and physical symptoms in order to identify stress profiles and symptom clusters, respectively.ResultsAnalysis revealed four subgroups that each demonstrated a typological response pattern regarding job demands and health symptoms, respectively. The revealed stress profiles were found to be strongly related to the symptom clusters, while the effects differed considerably depending on the types of demands experienced.ConclusionThe current study presents an alternative method of examining the stress-health link by using a combined person- and variable-centred approach. The findings suggest a hierarchy in stress exposure with the most pronounced health consequences found for a synchronous burden from physical, psychosocial and organizational demands.

Fatal and serious road crashes involving young New Zealand drivers: a latent class clustering approach

The over-representation of young drivers in road crashes remains an important concern worldwide. Cluster analysis has been applied to young driver sub-groups, but its application by analysing crash occurrence is just emerging. We present a classification analysis that advances the field through a holistic overview of crash patterns useful for designing youth-targeted road safety programmes. We compiled a database of 8644 New Zealand crashes from 2002 to 2011 involving at least one 15–24-year-old driver and a fatal or serious injury for at least one road user. We considered crash location, infrastructure characteristics, environmental conditions, demographic characteristics, driving behaviour, and pre-crash manoeuvres. The analysis yielded 15 and 8 latent classes of, respectively, single-vehicle and multi-vehicle crashes, and average posterior probabilities measured the odds of correct classification that revealed how the identified clusters contain mostly crashes of a particular class and all the crashes of that class. The results raised three major safety concerns for young drivers that should be addressed: (1) reckless driving and traffic law violations; (2) inattention, error, and hazard perception problems; and (3) interaction with road geometry and lighting conditions, especially on high-speed open roads and state highways.

Cyclist–Motorist Crash Patterns in Denmark: A Latent Class Clustering Approach

Objective: The current study aimed at uncovering patterns of cyclist–motorist crashes in Denmark and investigating their prevalence and severity. The importance of implementing clustering techniques for providing a holistic overview of vulnerable road users’ crash patterns derives from the need to prioritize safety issues and to devise efficient preventive measures. Method: The current study focused on cyclist–motorist crashes that occurred in Denmark during the period between 2007 and 2011. To uncover crash patterns, the current analysis applied latent class clustering, an unsupervised probabilistic clustering approach that relies on the statistical concept of likelihood and allows partial overlap across clusters. Results: The analysis yielded 13 distinguishable cyclist–motorist latent classes. Specific crash patterns for urban and rural areas were revealed. Prevalent features that allowed differentiating the latent classes were speed limit, infrastructure type, road surface conditions, number of lanes, motorized vehicle precrash maneuvers, the availability of a cycle lane, cyclist intoxication, and helmet wearing behavior. After the latent class clustering, the distribution of cyclists’ injury severity within each cluster was analyzed. Conclusions: The latent class clustering approach provided a comprehensive and clear map of cyclist–motorist crash patterns. The results are useful for prioritizing and resolving safety issues in urban areas, where there is a significant share of cyclists potentially involved in multiple hazardous situations or where extensive bicycle sharing programs are planned.

Heterogeneity in climate change risk perception amongst dairy farmers: A latent class clustering analysis

Livestock production has been criticized for its effect on greenhouse gas emissions and policy makers are now supporting actions to reduce these impact. Voluntary adoption of these actions will be precluded by the farmer perception of the risks from a changing climate. We employ a latent class clustering approach to understand the heterogeneity within a sample of dairy farmers, based on 8 statements related to climate change risk.The majority of farmers are found to be 'confused moderates' who have no strong opinion towards the possible future impacts of climate change. Two further classes emerged, namely 'deniers' and 'risk perceivers'. We find that higher education levels have an influence on increasing risk perception, as does the intention to pass the farm onto another family member. Membership of agri-environmental schemes does not preclude awareness or acceptance of climate change risk, principally due to the lack of emphasis on greenhouse gas emissions within these schemes. In addition, use of social networks seems to be a significant factor in raising the profile of risk perception within farmer decision-making.We conclude that advisors and those engaging with the farming community must accommodate climate messages in their communication strategies. Furthermore, emphasis on the greenhouse gas benefits from adoption of agri-environmental schemes would seem to be an efficient vehicle for raising the risk profile of climate change and influence future uptake of Government and industry supported actions.

SMEs' internationalisation patterns: descriptives, dynamics and determinants

PurposeThe purpose of this paper is to identify the internationalisation patterns of small‐ and medium‐sized enterprises (SMEs) quantitatively, to describe SMEs as they follow different patterns over time and to discuss the determinants of these patterns through empirical study.Design/methodology/approachThe authors conducted a questionnaire survey among mature German SMEs (n=674). To identify internationalisation patterns, a latent class clustering approach was applied. Because of the large sample, a multinomial logistic regression analysis could be used to analyse the factors influencing these patterns.FindingsThe authors empirically find three internationalisation patterns: traditionals, born globals and born‐again globals. Comparing modern SMEs with the same SMEs from ten years ago, it was found that firms may change their patterns. Moreover, the patterns are determined by international orientation, growth orientation, communication capability, intelligence generation capability and marketing‐mix standardisation.Research limitations/implicationsCombining elements of the Uppsala model (countries and operation modes) and born global research (time lag and foreign sales ratio), three internationalisation patterns of established international SMEs from traditional sectors were identified empirically. Because of the multidimensional nature of internationalisation, the patterns may change over time. Different firm‐level factors determine the internationalisation patterns.Originality/valueInstead of applying “arbitrary” thresholds, the paper provides a quantitative approach to identifying internationalisation patterns. These patterns confirm the three main internationalisation pathways discussed in the international marketing literature. The paper further advances the field by describing the patterns, showing evidence that the patterns may cross over time and providing information on the factors that influence the patterns.

Impacts of fast food and the food retail environment on overweight and obesity in China: a multilevel latent class cluster approach

To simultaneously identify consumer segments based on individual-level consumption and community-level food retail environment data and to investigate whether the segments are associated with BMI and dietary knowledge in China. A multilevel latent class cluster model was applied to identify consumer segments based not only on their individual preferences for fast food, salty snack foods, and soft drinks and sugared fruit drinks, but also on the food retail environment at the community level. The data came from the China Health and Nutrition Survey (CHNS) conducted in 2006 and two questionnaires for adults and communities were used. A total sample of 9788 adults living in 218 communities participated in the CHNS. We successfully identified four consumer segments. These four segments were embedded in two types of food retail environment: the saturated food retail environment and the deprived food retail environment. A three-factor solution was found for consumers' dietary knowledge. The four consumer segments were highly associated with consumers' dietary knowledge and a number of sociodemographic variables. The widespread discussion about the relationships between fast-food consumption and overweight/obesity is irrelevant for Chinese segments that do not have access to fast food. Factors that are most associated with segments with a higher BMI are consumers' (incorrect) dietary knowledge, the food retail environment and sociodemographics. The results provide valuable insight for policy interventions on reducing overweight/obesity in China. This study also indicates that despite the breathtaking changes in modern China, the impact of 'obesogenic' environments should not be assessed too strictly from a 'Western' perspective.

Social discourses of healthy eating. A market segmentation approach

This paper proposes a framework of discourses regarding consumers’ healthy eating as a useful conceptual scheme for market segmentation purposes. The objectives are: (a) to identify the appropriate number of health-related segments based on the underlying discursive subject positions of the framework, (b) to validate and further describe the segments based on their socio-demographic characteristics and attitudes towards healthy eating, and (c) to explore differences across segments in types of associations with food and health, as well as perceptions of food healthfulness.316 Danish consumers participated in a survey that included measures of the underlying subject positions of the proposed framework, followed by a word association task that aimed to explore types of associations with food and health, and perceptions of food healthfulness. A latent class clustering approach revealed three consumer segments: the Common, the Idealists and the Pragmatists. Based on the addressed objectives, differences across the segments are described and implications of findings are discussed.

Measuring exposure to bullying at work: The validity and advantages of the latent class cluster approach

This paper addresses the construct and predictive validity of two methods for classifying respondents as victims of workplace bullying. Although bullying is conceived as a complex phenomenon, the dominant method used in bullying surveys, the operational classification method, only distinguishes two groups: victims versus non-victims. Hence, the complex nature of workplace bullying may not be accounted for. Therefore a latent class cluster approach is suggested to model the data, which was obtained by using the Negative Acts Questionnaire (NAQ) administered to employees in Belgium (n=6,175). Latent class modelling is a method of analysis that does not appear to have been used in occupational health psychology before. In this study, six latent classes emerged: “not bullied,” “limited work criticism,” “limited negative encounters,” “sometimes bullied,” “work related bullied,” and “victims.” The results show that compared to the traditional operational classification method, the latent class cluster approach shows higher construct and higher predictive validity with respect to self-assessments and indicators of strain and well-being at work. The consequences of these results for theory, future research, and practice are discussed.