INTRODUÇÃO: Problemas de saúde mental são responsáveis por uma morbidade significativa em todo o mundo, por sua frequência e pela associação com comorbidades físicas, níveis de incapacitação e prejuízo na qualidade de vida de portadores e cuidadores. A ocorrência de transtornos mentais comuns (TMC) é influenciada por fatores biológicos, sociais, econômicos e demográficos. O contexto social pode ter papel importante na etiologia dos transtornos mentais e na sua prevalência. OBJETIVOS: Investigar fatores de risco que contribuem para a prevalência, incidência e prognóstico de TMC em população de baixa renda da cidade de São Paulo, considerando distintos níveis contextuais: indivíduo, domicílio e setor censitário. MÉTODO: O presente estudo utilizou dados da investigação longitudinal de base populacional \"São Paulo Ageing & Health Study\" (SPAH). Os indivíduos selecionados eram residentes em domicílios em que houvesse pelo menos dois participantes do estudo com avaliação para presença de TMC, identificada pelo instrumento Self Reporting Questionaire (SRQ-20). Foram avaliadas as associações independentes entre TMC e características sociodemográficas e dos domicílios dos participantes, através de modelos de regressão logística multinível, tendo como desfechos a prevalência de TMC na inclusão e a presença de TMC em avaliação de dois anos de seguimento. RESULTADOS: Foram incluídos 2.366 indivíduos no estudo transversal, realizado no período de 2003 a 2005. A prevalência de TMC nesta amostra foi de 43%. As características individuais sexo, idade, escolaridade e ocupação estiveram associadas à prevalência de TMC. As características de domicílios aglomeração, morar com crianças, bens materiais, saneamento básico e renda familiar também se associaram à prevalência de TMC. Modelos de regressão logística multinível mostraram que parte da variância na prevalência de TMC foi associada ao nível do domicílio, com associações entre aglomeração, renda familiar e prevalência de TMC, mesmo após controle para características individuais. No estudo longitudinal foram incluídos 1.733 indivíduos, reavaliados no período de 2005 a 2007. A prevalência de TMC na avaliação de seguimento foi de 33%, sendo que 8% não apresentavam TMC na inclusão. As características individuais sexo, idade e ocupação, e a característica de domicílio renda familiar estiveram associadas à presença de TMC na avaliação de seguimento. Modelos de regressão logística multinível para os dados longitudinais mostraram que a maior parte da variância na presença de TMC foi associada ao nível do indivíduo, com associações entre sexo, faixa etária, escolaridade, ocupação e TMC, mesmo após controle para características do domicílio. O nível de domicílio também contribuiu de forma independente para a variância relacionada à presença de TMC no seguimento, com associação de efeito fixo para renda familiar, mesmo após controle para características individuais. CONCLUSÃO: Os resultados mostraram que características individuais contribuem para a maior parte da variância na prevalência, incidência e prognóstico de TMC, mas há uma associação independente com o nível domicílio, que não é explicada completamente pela renda familiar. Esses resultados indicam que características do ambiente onde as pessoas vivem contribuem para sua saúde mental, sugerindo que pesquisas futuras se concentrem nas características psicossociais de domicílios e vizinhanças para estudo de contexto social e TMC / INTRODUCTION: Mental health problems are responsible for significant morbidity worldwide, due to its high frequency and association with physical comorbidities, levels of disability and impact in quality of life of patients and caregivers. The occurrence of common mental disorders (CMD) is influenced by biological, social, economic and demographic factors. The social context may play an important role in the etiology of mental disorders and their prevalence. OBJECTIVES: To investigate risk factors associated with the prevalence, incidence and prognosis of CMD in a low income population from the city of São Paulo, considering different contextual levels: individual, household, and census tract. METHOD: The present study used data from the population-based prospective investigation \"São Paulo Ageing & Health Study\" (SPAH). The individuals selected were living in households in which there were at least two study participants with assessments for presence of TMC, identified by the instrument Self Reporting Questionnaire (SRQ-20). We assessed independent associations between CMD and the sociodemographic and households characteristics of the participants, through multilevel logistic regression models, having as outcome variables the prevalence of CMD at inclusion and the presence of TMC at the 2-year follow-up assessment. RESULTS: We included 2.366 individuals in the cross-sectional study, carried out in the period from 2003 to 2005. The prevalence of CMD in this sample was 43%. Individual characteristics sex, age, education and occupation were associated with the prevalence of CMD. The household characteristics crowding, living with children, possessions, basic sanitation and family income were also associated with the prevalence of CMD. Multilevel logistic regression models showed that part of the variance in the prevalence of CMD was associated with the household level, with associations between crowding, family income and prevalence of CMD, even after controlling for individual characteristics. In the longitudinal study, 1.733 individuals were reassessed in the period of 2005-2007. The prevalence of CMD at 2 years was 33%, with 8% in individuals without TMC at inclusion. Individual characteristics sex, age and occupation and household characteristic family income were associated with presence of CMD at follow-up. Multilevel logistic regression models for longitudinal data showed that most of the variance associated with presence of CMD was associated with the level of the individual, with associations between sex, age, education, occupation and CMD, even after controlling for household characteristics. The household level also contributed independently to the variance related to the presence of TMC at follow-up, with associated fixed effect for family income, even after controlling for individual characteristics. CONCLUSION: The results showed that individual characteristics contribute to most of the variance in the prevalence, incidence and prognosis of TMC, but there is an independent association with the household level, which is not fully explained by family income. These results indicate that characteristics of the environment where people live contribute to their mental health, suggesting that future research focus on the psychosocial characteristics of households and neighborhoods to study the social and TMC

The focus of this thesis is the factorial survey approach as a method for studying professional judgements in social work. The factorial survey approach, which was first introduced in the social sciences around the beginning of the 1980s, constitutes an advanced method for measuring human judgements of social objects. At the general level, this quasi-experimental approach involves presenting respondents with fictive descriptions of social objects (vignettes), in which selected characteristics describing the objects to be judged are simultaneously manipulated. This thesis consists of four studies: In Study I, I explore the general use of the factorial survey approach in sociology between 1982 and 2006. Study II and Study III consist of factorial survey applications in the field of professional judgement in Swedish substance misuse treatment, as organized by the social services. To be more specific, the aims of these papers are to disentangle predictors of social work practitioners’ choices of inpatient or outpatient substance misuse treatment (Study II), and of social work practitioners’ judgements about eligibility for compulsory care (Study III). Finally, in Study IV, I present a conceptual and an analytical framework for the application of the factorial survey approach to the study of professional judgements in social work.

Cardiovascular diseases (CVDs) are the leading cause of death and disability around the world. The purpose of this thesis is to investigate the impact of socio-environmental determinants of CVDs at the neighbourhood scale in order to inform actionable interventions, which may lead to large-scale reductions in preventable CVDs. Drawing on 2411 surveys carried out in Toronto, Canada, this thesis employs multilevel models to estimate the magnitude of socio-environmental influences on the risk of CVD while adjusting for individual-level risk factors. To advance current research methodology, strategies and innovations were developed to 1) improve the characterization of neighbourhoods by empirically testing a full range of socio-environmental influences; 2) account for non-residential exposures by including a combined analysis of work and home contexts; 3) account for variations in the duration of exposure through the use of time-weighted models; 4) deal with problem of spatial data aggregation by developing and testing a novel method of neighbourhood zone design, and 5) account for the spatial scales of different socio-environmental determinants by modeling at multiple scales. The thesis demonstrated that land use decisions are inextricably public health decisions. It found that living in neighbourhoods with inadequate access to food stores and areas for physical activity, burdened by violent crimes and fast food restaurants, and over-dependent on automobiles (leading to air pollution), with a high level of noise may significantly increase the risk of CVDs, over and above individual-level risks. The thesis also found that working in neighbourhoods that are socio-economically disadvantaged or have high-traffic may significantly increase CVD risk. The thesis developed and demonstrated novel methods to reduce the measurement error of neighbourhood exposures through 1) the use of “amoeba buffers” to improve neighbourhood zone design to better reflect participants’ local neighbourhoods and 2) the use of duration of exposure weights to adjust for individual differences in the time spent across different contexts. Finally, it found that the significance of socio-environmental factors depends on the scale of data aggregation; thus, investigation of multiple scales may be required to identify the relevant scale that matches the specific contextual factor in future research.

Cardiovascular diseases (CVDs) are the leading cause of death and disability around the world. The purpose of this thesis is to investigate the impact of socio-environmental determinants of CVDs at the neighbourhood scale in order to inform actionable interventions, which may lead to large-scale reductions in preventable CVDs. Drawing on 2411 surveys carried out in Toronto, Canada, this thesis employs multilevel models to estimate the magnitude of socio-environmental influences on the risk of CVD while adjusting for individual-level risk factors. To advance current research methodology, strategies and innovations were developed to 1) improve the characterization of neighbourhoods by empirically testing a full range of socio-environmental influences; 2) account for non-residential exposures by including a combined analysis of work and home contexts; 3) account for variations in the duration of exposure through the use of time-weighted models; 4) deal with problem of spatial data aggregation by developing and testing a novel method of neighbourhood zone design, and 5) account for the spatial scales of different socio-environmental determinants by modeling at multiple scales. The thesis demonstrated that land use decisions are inextricably public health decisions. It found that living in neighbourhoods with inadequate access to food stores and areas for physical activity, burdened by violent crimes and fast food restaurants, and over-dependent on automobiles (leading to air pollution), with a high level of noise may significantly increase the risk of CVDs, over and above individual-level risks. The thesis also found that working in neighbourhoods that are socio-economically disadvantaged or have high-traffic may significantly increase CVD risk. The thesis developed and demonstrated novel methods to reduce the measurement error of neighbourhood exposures through 1) the use of “amoeba buffers” to improve neighbourhood zone design to better reflect participants’ local neighbourhoods and 2) the use of duration of exposure weights to adjust for individual differences in the time spent across different contexts. Finally, it found that the significance of socio-environmental factors depends on the scale of data aggregation; thus, investigation of multiple scales may be required to identify the relevant scale that matches the specific contextual factor in future research.

Konsumenten stehen beim Kauf von Lebensmitteln einer großen Anzahl an Produkten gegenüber. Eine hohe Dichte an Informationen bezüglich der Zutaten, Herkunft oder des Produktionsverfahrens sind verfügbar. Dadurch können Konsumenten nicht nur über Inhalts- und Nährstoffe der Produkte entscheiden, sondern auch über das System, in dem Nahrungsmittel produziert werden. Somit beeinflussen Entscheidungen rund um die Lebensmittelwahl nicht nur das individuelle Wohl, sondern auch das Gesellschaftliche. Primäres Ziel der Dissertation ist es, (i) einen Beitrag zum Wissen und der Erklärung von Verbraucherpräferenzen zu leisten, und (ii) Implikationen und Empfehlungen für Regierungen, Politiken und insbesondere die Wissenschaft abzuleiten. Am Beispiel von Schulmilch und der Tierhaltung werden Präferenzen für Produkte, Produktattribute und für Produktionssysteme der Landwirtschaft untersucht. Sowohl das Entscheidungsverhalten als auch Einstellungen von Verbrauchern werden betrachtet. Das Entscheidungsverhalten von Kindern und Jugendlichen bei Schulmilch ist deswegen von Interesse, da sich Ernährungsgewohnheiten in der Kindheit ausbilden bzw. diese erlernt werden. Die Einflussfaktoren auf die Nachfrage nach Schulmilch werden im ersten Beitrag untersucht. Das Nachfragemodell, welches auf der Methode der Multilevelanalyse beruht, berücksichtigt sowohl individuelle Merkmale der Kinder als auch Merkmale der Familie und des Schulkontextes. Dabei werden vier Level bzw. Ebenen unterschieden: die Preis-, Schüler-, Klassen- und Schulebene. Die Ergebnisse zeigen, dass der Produktpreis (d.h. die Verfügbarkeit einer Subvention) für die Nachfrage nach Schulmilch von Bedeutung ist. Der Anteil der Schüler, die bei einer stärkeren Subventionierung der Preise zusätzlich Schulmilch bestellen, ist jedoch relativ gering. Zudem haben individuelle Faktoren, wie Konsumgewohnheiten zuhause, sozio-demografische Merkmale oder Einstellungen gegenüber Milch einen signifikanten Einfluss. Die Ergebnisse liefern Hinweise auf Sozialisationseffekte und Rollenvorbilder (role modelling), da Einstellungen und Verhalten der Eltern und Lehrer die Entscheidungen der Kinder signifikant beeinflussen. Die Analyse der Präferenzen für Schulmilch in Abhängigkeit des Geschlechtes steht im zweiten Beitrag im Vordergrund. Die Ergebnisse der Multilevelanalyse bestätigen, dass identische und unterschiedliche Faktoren die Schulmilchnachfrage von Jungen und Mädchen bestimmen. Der dritte Beitrag analysiert, ob die am Markt existierenden Schulmilchprodukte (noch) den Präferenzen von jungen Erwachsenen entsprechen. Ergebnisse des Nested Logit-Modells zeigen, dass eine breitere Produktpalette präferiert wird. Weiterhin sind sich die Jugendlichen über den Fett- und Zuckergehalt und die Auswirkungen auf Gesundheit und Körpergewicht bewusst. Wie erwartet, hat der Preis einen signifikanten Einfluss auf die Entscheidung der Jugendlichen. In Bezug auf die Tierhaltung sind das Wissen und Verständnis der Verbraucherpräferenzen ebenfalls von großem Interesse, da die heutige Landwirtschaft und Lebensmittelproduktion häufig sehr kritisch in der Öffentlichkeit und den Medien diskutiert wird. Im vierten Beitrag werden daher am Beispiel der modernen Schweinehaltung die gesellschaftlichen Einstellungen für landwirtschaftliche Produktionssysteme untersucht. Der Hauptkritikpunkt der Diskutanten ist der zur Verfügung stehende Platz pro Tier. Obwohl die Wahrnehmungen und Einstellungen gegenüber der modernen Schweinehaltung generell negativ bzw. kritisch sind, ergibt sich ein differenziertes Bild in der deutschen Bevölkerung. Drei Gruppen mit sehr unterschiedlichen Einstellungen werden bestimmt. Insgesamt zeigen die Ergebnisse der Dissertation, dass durch die Kombination von methodischen Ansätzen detaillierte Erkenntnisse von Verbraucherpräferenzen gewonnen werden können. Bei der Analyse des Entscheidungsverhaltens werden auf Basis von revealed und stated preference-Daten eine Vielzahl an Einflussfaktoren, die das Nachfrageverhalten nach Schulmilch beeinflussen, identifiziert. Zudem werden durch die Kombination von qualitativen mit quantitativen Methoden die in der Gesellschaft vorherrschenden Einstellungen gegenüber der modernen Schweinehaltung detailliert erfasst. Letztendlich werden Konsumenten auch nicht nur in ihrer Rolle als Entscheider oder Käufer betrachtet, sondern auch in ihrer Rolle als Bürger oder als Teil der Gesellschaft.

[Truncated abstract] Collective efficacy (CE) has been consistently shown to contribute to team performance by improving motivation, perseverance, group goals, and expectancy in teams (Bandura, 2000; Feltz & Lirgg, 2001). Having high confidence in one's team is proposed to increase the expectancy for success, thereby increasing motivation, persistence and effort, which in turn, increases overall performance (Bandura, 1997). A series of studies were conducted to examine predictors of self and collective efficacy in semi-professional soccer. In addition, within two of these studies, the effect of pivotal hypothetical injuries on players' and coaches' perceptions of CE were also examined. In the first and second studies, hierarchical linear modeling (HLM) was utilized in order to examine simultaneous player and team level predictors of efficacy perceptions. Data were obtained from semi-professional male soccer players (N = 139, mean age = 23, SD = 4.43 yrs) from nine teams in Western Australia. Participants completed the Group Environment Questionnaire (GEQ), the revised Perceived Motivational Climate in Sport Questionnaire-2 (PMCSQ-2) as well as measures of self-talk, self-efficacy (SE), and CE created for this study. HLM analyses indicated CE was significantly and positively associated with SE and significantly and negatively associated with perceptions of a performance-oriented motivational climate. Significant positive associations of SE were positive self-talk and average number of minutes played each game. Negative correlates of SE were a performance-oriented motivational climate and number of years played. The results illustrate the important cross-level influences of player and team level variables on SE and CE perceptions. Overall, the findings provide support for the propositions of the influence of sources of efficacy information and broaden the existing work on efficacy and motivational climate in sport. ... Future research on team processes following injury that may moderate the injury efficacy relationship may include, the effect of team leaders in an attempt to motivate the team and the verbal and non-verbal strategies of coaches. The present studies contribute to the existing body of knowledge concerning efficacy theory. Specifically, within this series of studies, individual and team level predictors of SE and CE among elite sport participants were examined. A performance-oriented motivational climate was negatively associated with CE perceptions, while players with elevated levels of SE had teams with elevated levels of CE. SE was positively associated with positive self-talk, and negatively correlated with a performance-oriented motivational climate. Further, the effects of player injury on CE perceptions also represent a valuable contribution to efficacy theory. Prior to the current studies, no researchers have studied the influence of athlete absence due to injury on CE.

Fear of crime is a social problem on its own, partly independent of crime as actual crime does not fully explain as to why some individuals are more afraid for crime than others. Structuralist perspectives have offered some explanations, largely neglecting the potential importance of institutional perspectives. This thesis aims to study fear of crime from an institutional perspective using the theoretical construct quality of government which seeks to offer a way to measure well-functioning institutions. It is assumed that quality of government has an impact in different ways; through trust and victimisation. To study these research questions, data were drawn from two sources, the European Social Survey (ESS) and Quality of Government EU Regional Data. The sample contained 85,794 individuals nested in 152 regions which were situated in 18 European countries. The empirical analysis consisted of random intercept multilevel modelling. It was found that the quality of government was correlated with fear of crime, where higher levels of the former tended to result in lower levels in the latter. Trust were found to mediate this correlation whereas victimisation interacted with quality of government.

Magister Philosophiae - MPhil / The South African schools act, (number 5, 1996), asserts that all learners have a right to access both basic and quality education without discrimination of any sort. Since the implementation of the Millennium Development Goals there has been a drive by the Department of Education to ensure that all learners have access to basic education by 2015. However what remains a challenge after almost 20 years of democracy is the poor quality of education and this is clear from the results of international assessment studies. Results from studies like the Trends in International Mathematics and Science Study and Southern and East Africa Consortium for Monitoring Educational Quality, show that South African children perform well below international averages. In this study learner Mathematics achievement scores taken from the Trends in International Mathematics and Science Study 2011 cycle will serve as a proxy for educational quality. Using multilevel analysis the current study aims to use a 2-level Hierarchical Linear Model to firstly; determine the learner and family background factors associated with education quality. Secondly; factors at the school level will be identified and proven to be associated with education quality. Variables selected for the study was based on Creamer’s theory of school effectiveness which looked at school, classroom level inputs as well as learner background variables to explain student level achievement. The results show that at the learner’s level the most significant factors were the age of the leaner, in the sense that grade age appropriate learners obtained higher scores than overage learners. Learner’s perception of mathematics is extremely important and has a positive effect on mathematics performance. In the current study mathematics perception refers to learners valuing and liking mathematics as well learner confidence in learning mathematics. Learners who said they were bullied as school generally scored lower than learners who were not bullied. At the school level the most significant factors were teacher working conditions, teachers’ specialisation in mathematics, school socio-economic status, and general infrastructure. Interesting to note at the school level is when socioeconomic status was included in the model as a single variable the score difference between low socio-economic status and high socio-economic status schools was almost 46 points. However when the factors mentioned above were added to the model the difference in scores dropped by almost half.

Submitted by LINO MEJIA (estebanllamosam@gmail.com) on 2017-06-22T20:39:35Z No. of bitstreams: 1 MasterThesis_Lino_Mejia_FGV_Ficha.pdf: 4647982 bytes, checksum: 29d2e5446d8872f2ad94a5fac6a16c72 (MD5) / Approved for entry into archive by Maria Tereza Fernandes Conselmo (maria.conselmo@fgv.br) on 2017-06-23T21:17:44Z (GMT) No. of bitstreams: 1 MasterThesis_Lino_Mejia_FGV_Ficha.pdf: 4647982 bytes, checksum: 29d2e5446d8872f2ad94a5fac6a16c72 (MD5) / Made available in DSpace on 2017-06-23T21:47:22Z (GMT). No. of bitstreams: 1 MasterThesis_Lino_Mejia_FGV_Ficha.pdf: 4647982 bytes, checksum: 29d2e5446d8872f2ad94a5fac6a16c72 (MD5) Previous issue date: 2017-05-24 / Academics have performed several studies on variance decomposition of several types of performance in context of developed economies. Nevertheless, empirical research of variance decomposition in emerging economies is more limited than that of developed economies. This study analyzes the variance decomposition of Peruvian enterprises in the period between 2010 and 2014, employing multidimensional performance measures, and thus two objectives are present. Firstly; determine the financial performance variance composition and compare this with developed economies. Secondly; determine the variance composition of productivity and compare it with the variance structure of financial performance. This empirical research is based on an original longitudinal dataset which contains financial, production and labour market information at firm level and it uses the Hierarchical Lineal Model (HLM) which provides conceptual and robust statistics for analyzing the nested structure of the data. Regarding the first objective, the empirical results demonstrate that the industry effect account for 34% and the firm effect account for 66% of the variance explained. This differs with previous studies such as Makino et al. (2004) who presented an industry effect of 18% of the variance explained in a developed economy. Regarding the second objective, three measures of performance are used to capture the multidimensional nature of performance. The dimensions are financial (measured as ROA and ROS) and operational (measured as productivity). The findings show that operational productivity model explain 65% of their variability; whereas the ROA and ROS explain 55% and 61% respectively. This study also confirms that the variance structure impacts the performance of each aggregate sector separately. For manufacturing, the result show that the percentage of industry effect is stronger than that of other aggregate sectors. For service, the industry effect (when it uses profitability measures) show similar values with some studies in the Latin American context. Finally; for wholesale & retail trade, the results confirm that the industry effect is lower than the other aggregate sector, and shows values around 4%, and therefore, firm effect ranges between 40% and 70% of the total variance. / Académicos han realizado varios estudios sobre la descomposición de varianza utilizando diversas clases de desempeño y principalmente en contexto de economías desarrolladas. Sin embargo, investigaciones empíricas sobre descomposición de la varianza en economías emergentes es más limitada que en contexto de economías desarrolladas. El presente estudio analiza la descomposición de la varianza en una muestra de empresas peruanas durante el período 2010-2014, empleando medidas de desempeño multidimensionales y, por lo tanto, dos objetivos de investigación son presentados. Primero; determinar la composición de la variación del rendimiento financiero y comparar estos resultados con los estudios en economías desarrolladas. Segundo; determinar la composición de la varianza de la productividad y compararla con la estructura de la varianza del rendimiento financiero. Esta investigación empírica se sustenta en una exclusiva base de datos en panel, los cuales contienen información financiera, de producción y de mercado de trabajo a nivel de empresa y además utiliza el Modelo Lineal Jerárquico (MLJ) el cual proporciona un método conceptual e robusto para analizar datos jerárquicos. En cuanto al primer objetivo, los resultados empíricos demuestran que el efecto industria representa el 34% y el efecto firma representa el 66% de la varianza explicada. Estos resultados difieren con estudios previos como Makino et al. (2004), quienes encontraron un efecto industria del 18% de la varianza explicada en un contexto de economía desarrollada. Con respecto al segundo objetivo, se utilizan tres medidas de rendimiento para captar la naturaleza multidimensional del desempeño. Las dimensiones son financieras (medidas como ROA y ROS) y operacional (medidas como productividad). Los resultados muestran que el modelo de productividad operacional explica el 65% de su variabilidad; Mientras que el ROA y el ROS explican el 55% y el 61%, respectivamente. Este estudio también confirma que la estructura de la varianza depende del desempeño de cada sector agregado. En el sector de manufactura, el resultado muestra que el porcentaje del efecto industria es más fuerte que el de otros sectores agregados. Para el sector servicio, el efecto industria (cuando utiliza la rentabilidad) muestran valores similares en estudios en el contexto latinoamericano. Finalmente; para el sector de comercio, los resultados confirman que el efecto industria es menor en comparación a los otros sectores, y muestra valores alrededor del 4%, y por lo tanto, el efecto firma oscila entre 40% y 70% del total de la varianza. / Acadêmicos têm realizado vários estudos sobre a decomposição da variância de vários tipos de desempenho no contexto de economias desenvolvidas. Não obstante, a investigação empírica da decomposição da variância nas economias emergentes é mais limitada do que a das economias desenvolvidas. Este estudo analisa a decomposição da variância das empresas peruanas no período entre 2010 e 2014, empregando medidas de desempenho multidimensionais, e, portanto, dois objetivos se fazem presentes. Primeiramente; determinar a composição da variância do desempenho financeiro e compará-la com as das economias desenvolvidas. Em segundo lugar; determinar a composição da variância da produtividade e compará-la com a estrutura de variância do desempenho financeiro. Esta pesquisa empírica é baseada em um conjunto de dados longitudinal original, que contém informação financeira, de produção e do mercado de trabalho a nível de empresa e usa o Modelo Hierárquico Linear (MHL), que oferece estatísticas robustas para analisar a estrutura hierárquica dos dados. Em relação ao primeiro objetivo, os resultados empíricos demonstram que o efeito setor responde por 34% e o efeito firma corresponde por 66% da variância explicada. Isso difere de estudos anteriores, como o de Makino et al. (2004) , que apresentou um efeito setor de 18% da variância explicada numa economia desenvolvida. Quanto ao segundo objetivo, três medidas de desempenho são usadas para capturar a natureza multidimensional do desempenho. As dimensões são financeiras (medidas como ROA, ROS, crescimento de ativos e receitas) e operacional (medida como produtividade). Os resultados mostram que o modelo de produtividade operacional explica 65% da sua variabilidade; enquanto que o ROA e ROS explicam 55% e 61% respectivamente. Este estudo também confirma que a estrutura da variância muda para cada setor agregado separadamente. Para manufatura, o resultado mostra que a percentagem do efeito setor é maior do que em outros setores agregados. Para serviços, o efeito setor (quando usada medidas de rentabilidade) mostra valores semelhantes com alguns estudos realizados no contexto latino-americano. Finalmente; para comércio atacadista e varejista, os resultados confirmam que o efeito setor é menor do que o outro setor agregado, e mostra valores em torno de 4%, e, portanto, o efeito firma varia entre 40% e 70% da variância total.

Introdução - O estudo de eventos de saúde deve levar em conta que as características dos indivíduos de uma determinada localidade não constituem simples somatórios das medidas de cada um dos sujeitos e há que se considerar um modelo explicativo baseado em níveis de organização e na estrutura de dependência entre o nível individual e o nível de contexto onde esses sujeitos estão inseridos. Assim, a análise dos nascimentos e da mortalidade neonatal e fetal pode incorporar diferentes variáveis associadas ao contexto onde se expressam considerando a complexidade e as particularidades dessas ocorrências numa população e num espaço tão diverso. Metodologia - Foi realizado estudo transversal dos nascimentos, óbitos neonatais (<28 dias de vida) e óbitos fetais de mães residentes e ocorridos no município de São Paulo. Os endereços de residência materna foram geocodificados e foi calculada a distância entre as residências e o hospital de ocorrência. Além disso, cada indivíduo foi caracterizado com informações socioeconômicas do Censo Demográfico de 2010, segundo área de Ponderação. Os setores censitários de residência foram classificados segundo Índice Paulista de Vulnerabilidade Social IPVS. Os hospitais foram classificados em SUS e não SUS e para os Nascidos Vivos (e óbitos neonatais) também foram classificados segundo referência para atendimento de risco gestacional. Foram obtidos aglomerados de Nascidos Vivos (NV) através da técnica de varredura espacial. Através de análise multinível foi verificado o efeito do contexto socioeconômico na mortalidade neonatal e fetal. Resultados - Verificou-se que os aglomerados tanto SUS como não SUS não são homogêneos entres si, com diferenças em relação à idade das mães, escolaridade, número de consultas pré-natal e prematuridade. A distância média teórica percorrida pelas mães até o hospital foi 51,8% menor nos aglomerados SUS que nos não SUS. A menor distância nos nascimentos SUS indica a regionalização da assistência ao parto no município de São Paulo. Os resultados mostraram que há um aumento da taxa de mortalidade neonatal com o aumento da vulnerabilidade social. Houve um efeito contextual da vulnerabilidade social e observa-se que apenas as variáveis individuais que representam as características da gestação, recém-nascido e assistência pré-natal mostraram-se associadas à mortalidade neonatal. O efeito contextual da vulnerabilidade social nas variáveis individuais que representam as características da gestação, feto e escolaridade materna mostrou-se associadas à mortalidade fetal. Na modelagem multinível não foi observada variabilidade importante da mortalidade fetal entre os níveis. Conclusões - A detecção de aglomerados e sua caracterização socioeconômica das áreas contribuem para o entendimento do padrão de nascimentos e nas intervenções de saúde pública, proporcionando melhoria no atendimento das necessidades de acesso ao pré-natal e parto de forma mais eficiente. Os resultados em relação à mortalidade neonatal e fetal revelam que as desigualdades sociais estão presentes na cadeia causal desses dois desfechos e o que contribui com a compreensão dos fatores de risco para a mortalidade neonatal e fetal, principalmente no que diz respeito à participação da vulnerabilidade social na mortalidade e explicita a distância entre a residência materna e o hospital como um indicador socioeconômico / Introduction - The study of health events should take into account that the characteristics of the individuals of a given locality are not simple sums of the measures of each one of the subjects and it is necessary to consider an explanatory model based on levels of organization and the structure of dependence between the Individual level and the context level where these subjects are inserted. Thus, the analysis of neonatal and fetal births and mortality can incorporate different variables associated to the context considering the complexity and the peculiarities of these occurrences in a population and in such a diverse space. Methodology - A cross-sectional study of births, neonatal deaths (<28 days of life) and fetal deaths of resident mothers occurred in the city of. The maternal residence addresses were geocoded to calculate the distance between the residences and the hospital. In addition, each individual was characterized with socioeconomic information from the Demographic Census of 2010, according to the weighting areas. The census tracts of residence were classified according to Index of Social Vulnerability - IPVS. Hospitals were classified in SUS and non-SUS and for live births (and neonatal deaths) were also classified according to reference for gestational risk care. The clusters of live births (LB) were obtained through the spatial sweep technique. The effect of the socioeconomic context on neonatal and fetal mortality was verified by multilevel analysis. Results - It was verified that the clusters both SUS and non-SUS are not homogeneous between them, with differences in relation to the mothers\' age, schooling, number of prenatal consultations and prematurity. The mean theoretical distance traveled by the mothers to the hospital was 51.8% lower in the SUS clusters than in the non-SUS. The shorter distance in SUS births indicates the regionalization of childbirth care in the city of São Paulo. The results showed that there is an increase in the neonatal mortality rate with increased social vulnerability. There was a contextual effect of social vulnerability and it was observed that only the individual variables that represent the characteristics of gestation, newborn and prenatal care were shown to be associated with neonatal mortality. The contextual effect of social vulnerability on the individual variables that represent the characteristics of gestation, fetus and maternal schooling has been shown to be associated with fetal mortality. In the multilevel modeling whose context was the level of vulnerability of the place of maternal residence, no significant variability of fetal mortality between the levels was observed. Conclusion - The detection of clusters and their socioeconomic characterization of the areas contribute to the understanding of the birth pattern and the public health interventions, providing an improvement in the attendance of prenatal access and delivery needs in a more efficient way. The results in relation to neonatal and fetal mortality reveal that social inequalities are present in the causal chain of these two outcomes and that contributes to the understanding of the risk factors for neonatal and fetal mortality, especially with regard to the participation of social vulnerability In mortality and explicit the distance between the maternal residence and the hospital as a socioeconomic indicator