1,721,058 research outputs found
Using Structural Equation and Item Response Models to Assess Relationship between Latent Traits
No full textWe deepen the two main approaches to the problem of measurement error in social sciences, the Structural Equation Models (SEM) and the Item Response Theory Models (IRM), comparing two different estimation procedures.
The One-step procedure (related to SEM) requires that researcher specifies a complete model of both measurement aspects (single link between the latent variable and its indicators) and structural aspects (links between different latent variables), with the model parameters estimated simultaneously. In the Two-step procedure (related to IRM), we first estimate the measures (one for each construct), then we will assess, through a regression model, the relationships between these measures and the latent variables that they represent.
Our aim is to define a Two-step method that, using information obtained in the first step about the measurement error, presents low levels of bias and loss of efficiency, as close as possible to that of One-step method
Estimation Procedures for latent Variable Models with psychological Traits
No full textThe starting point for this thesis is a concrete problem: to measure, using statistical models, aspects of subjective perceptions and assessments and to understand their dependencies. The objective is to study the statistical properties of some estimators of the parameters of regression models with variables affected by measurement errors. These models are widely used in surveys based on questionnaires developed to detect subjective assessments and perceptions with Likert-type scales. It is a highly debated topic, as many of the relevant aspects in this field are not directly observable and therefore the variables used to estimate them are affected by measurement errors. The models with measurement errors were very thorough in literature. In this work we will developed two of the most used approaches that the authors have with this topic. Obviously, according to the approach chosen,
different models were proposed to estimate the relationships between variables affected by measurement error. After exposing the main features of these models, the thesis focuses on providing an original contribution to comparative analysis of the two presented approaches
Formative and reflective models: state of the art
No full textAlthough the dispute between formative models and reflective models is not exactly recent, it is still alive in current literature, largely in the context of structural equation model. There are many aspects of SEM that should be considered in deciding the right approach. This work is intended to be a brief presentation of the state of the art for SEM based on covariance matrices. I outline the different positions on five particular issues: causality, selection of observed measures, internal consistency, identifiability and measurement error
Formative and reflective models to determine latent construct.
No full textIn numerous contexts, experts have to handle ordinal data and many methods have been proposed to treat this type of data. All possible approaches can be grouped into two main strands: formative and reflective models. The area of application is crucial in choosing which of two approaches to develop. We present an brief overview of the proposed models, trying to define the peculiarities of each aspect and their essential characteristics which must be verified during the analysis
Modelli quantitativi per l’analisi della biodiversità negli agroecosistemi
No full textNell’ultimo decennio, agenzie e istituzioni internazionali hanno emanato direttive e attivato linee di ricerca che sanciscono il ruolo chiave dal capitale naturale nella definizione delle strategie che promuovono l’agricoltura sostenibile. Esempi recenti di questi nuovi indirizzi di policy sono l’adozione della ‘Strategy on Mainstreaming Biodiversity across Agricultural Sectors’ , da parte della FAO, e della ‘Strategia sulla biodiversità per il 2030’ , da parte della Unione Europea, entrambe nel 2020. Da questi documenti si evidenzia la necessità di supportare l’adozione di pratiche agricole sostenibili a tutela, valorizzazione e ripristino della biodiversità, come elemento chiave del capitale naturale, ed emerge la necessità di disporre di strumenti che consentano di analizzare il ruolo che la biodiversità, nella duplice componente strutturale e funzionale, svolge nei processi di genesi e rigenerazione dei servizi ecosistemici.
In questa tesi si propone un framework quantitativo per l’analisi della biodiversità, definita come una rete di elementi. Ciascun elemento è uno specifico taxon microbico, animale o vegetale e può essere descritto tramite differenti attributi (ad esempio presenza/assenza, abbondanza o tratti funzionali). Le proprietà delle reti e la descrizione degli elementi è basata su misure quantitative. Il framework consente di analizzare differenti livelli di indagine (in termini di risoluzione spaziale), diverse dimensioni e componenti della biodiversità ed è basato su un approccio generativo (ossia consenta di indagare la relazione tra tratti e servizi ecosistemici). In termini di risoluzione spaziale, il framework considera tre livelli: i) l’unità ambientale, ossia la singola unità spaziale determinata dalla comunità vegetale predominante, ii) il livello della singola azienda agraria, composta da unità ambientali-produttive contigue, iii) il paesaggio (o landscape), un insieme eterogeneo di unità ambientali, sia produttive che non produttive. In termini di dimensioni, in primo luogo vengono distinte la dimensione ipogea e quella epigea. Successivamente, per ciascuna di queste dimensioni sono distinte la componente microbica, dei metazoi (con particolare riferimento ad artropodi e nematodi) e la componente dei vegetali (parte radicale e parte epigea).
Nel capitolo introduttivo viene presentato in dettaglio il framework, quale risposta scientifica all’esigenza di sviluppare strumenti quantitativi per analizzare la biodiversità negli agroecosistemi. Il secondo capitolo si concentra sull’analisi dell’unità spaziale, in particolare vengono indagati i singoli taxa ed i modelli quantitativi che consentono di studiare il legame tra caratteristiche del taxon e determinanti ambientali. Il caso studio sviluppato riguarda modelli di habitat suitability per Popillia japonica. Nel terzo capitolo viene indagato il livello aziendale, in particolare sono presentati i modelli che consentono la valutazione dell’impatto dei determinanti ambientali su alcuni componenti dellla biodiversità. Sono presentati dei casi studio di analisi della biodiversità degli artropodi nell’agroecosistema vigneto. Il quarto capitolo è dedicato alla revisione delle proposte metodologiche di strumenti quantitativi a supporto del framework, con il duplice focus sulle analisi a livello landscape e l’implementazione dell’approccio generativo. In questo capitolo sono esplorati principalmente modelli multidimensionali e multilivello. Nella sezione conclusiva viene proposta una sintesi delle linee di ricerca e delle innovazioni sviluppate e sono tracciate le prospettive di ricerca future.Over the last decade, major international agencies and institutions have established the key role played by natural capital in the definition of strategies promoting sustainable agriculture. Recent examples are the adoption of the 'Strategy on Mainstreaming Biodiversity across Agricultural Sectors' by the Food and Agriculture Organization of the United Nations (FAO) and the 'Biodiversity Strategy for 2030' by the European Union, both in 2020. These documents highlight the need to support the adoption of sustainable agricultural practices for the protection, enhancement and restoration of biodiversity, as a key element of natural capital, and the need to have quantitative tools to analyse the role that biodiversity plays in the processes of genesis and regeneration of ecosystem services.
This thesis proposes a quantitative framework for the analysis of biodiversity. In the framework, biodiversity is defined as a network of elements. Each element is a specific microbial, animal or plant taxon and can be described by different attributes (e.g. presence/absence, abundance or functional traits evaluations). The description of these elements, as well as the properties of the networks, are based on quantitative measures. The framework allows to analyse different levels of investigation in terms of spatial resolution, different dimensions and components of biodiversity and is based on a generative approach (i.e. it allows to investigate the relationship between traits and ecosystem services). In terms of spatial resolution, the framework considers three levels: i) the environmental unit, i.e. the single spatial unit determined by the predominant plant community, ii) the farm level, composed of contiguous productive environmental units, iii) the landscape, a heterogeneous set of environmental units, both productive and non-productive. In terms of dimensions, the hypogeal and epigeal dimensions are first distinguished. Subsequently, for each of these dimensions, the microbial component of metazoans (with particular reference to arthropods and nematodes) and the component of plants (root and epigeal part) are distinguished.
The introduction deeply describes the proposed framework, as a scientific response to the need of quantitative tools to analyse biodiversity in agroecosystems. The second chapter focuses on the analysis at the environmental unit level. In particular, quantitative models to study the link between taxon characteristics and environmental determinants are investigated. The case study concerns models of habitat suitability for Popillia japonica. The third chapter investigates the farm level, presenting the models allowing the analysis of the impact of environmental determinants on some biodiversity components. Case studies refer to the analysis of biodiversity of arthropods in the vineyard agroecosystem. The fourth chapter is a review of quantitative tools to support the framework, with the dual focus on the analysis at the landscape level and the implementation of the generative approach. Multidimensional and multilevel models are mainly explored in this chapter. In the final section, a synthesis of the lines of research and innovations developed is proposed and future research perspectives are outlined
Big Data to Monitor Big Social Events: Analysing the mobile phone signals in the Brescia Smart City
No full textA micro approach to cognitive skills’ growth in a university context
No full textThis paper focuses on the measurement of human capital, specifically on the growth of cognitive skills (CS) during the higher education at university. CS have already been evaluated in a macroeconomic perspective inside the neoclassical growth models, but not still in a micro perspective. However the measure of educational quality and learning process is still an issue not fully addressed. The micro approach allows the researcher to focus on the evaluation of CS acquisition process. Based on these measurements, different types of CS accumulation can be identified. We will investigate CS through nonlinear latent growth modeling, and we will apply this methodology to administrative data of an Italian university
Using Surrogate Models and Variable Importance to better Understand Random Forests Regression Fitting
Get PDFInterpretability mechanisms helping users in better understanding machine learning models are crucial for Artificial Intelligence acceptance. In this manuscript, our experience in interpretation of random forest regression via surrogate models, i.e. models trying to replicate in an interpretable framework an original fitting difficult to understand, is reported. It is shown how, beyond classical R2 analysis, adequacy of surrogate models can be assessed via variable importance analysis
Treating ordinal data: a comparison between rating scale and structural equation models
No full textThe aim of this study is to apply rating scale model and structural equation model to the same polytomous data in order to highlight the differences and similarities between the two models. For this purpose a simulation study is developed. Moreover, we present a real case regarding the analysis of the quality of work in an Italian municipality
On the use of Item Response Models in the SEM perspective.
No full textFor the analysis of complex models for latent constructs measured with several items, the Structural Equation Models (SEM) are being widely disseminated. In this study, our intent is to show how to include in a SEM framework an Item Response Model (IRM), in order to preserve the important characteristics that distinguish this type of approach, such as the possibility of calculate the measurement scales and the reduction of the complexity of the model. We compare these results with a standard SEM
- …
