Teacher turnover is a well-studied phenomenon, particularly in highly urbanized locales, but not well researched in a state as geographically and demographically diverse as Pennsylvania, which is a composition of two major metropolitan areas combined with smaller urban centers and expansive rural regions. Those retention studies that do exist have been mainly exclusive to the Philadelphia region, with limited research devoted to the remainder of the state. This lack of a comprehensive empirical approach that compares turnover in three distinct settings limits a nuanced understanding of the issue and, in turn, can lead to incomplete policy considerations. This study utilizes Pennsylvania Department of Education data from 2012-2017, which describes the entire public-school workforce in all local education agencies (LEAs), to study how compensation and auxiliary spending (per student spending sans instructional costs) influence teacher turnover using multiple, parallel Cox Proportional Hazards survival models. Findings suggest that despite a “one size fits all” approach to public school funding policy popular amongst politicians on both sides of the political aisle, the effects of a monetary increase in reducing the likelihood of turnover varies considerably when accounting for the region, Title I status, experience and subject matter. The study highlights how the lack of monetary investment can lead teachers to seek employment elsewhere since low pay functions as a strong demotivator. Additionally, the results suggest that while a pay raise may arrest turnover risk, it is a poor long-term motivator or cause of job satisfaction. The study concludes by offering state and LEA leaders with policy recommendations that may improve both retention and job satisfaction. To date, this is the only study in the current literature that explores teacher turnover extensively in the nation’s fifth most populous state. / Urban Education

Submitted by admin tede (tede@pucgoias.edu.br) on 2016-08-18T13:03:27Z No. of bitstreams: 1 Jonas Garcia de Almeida.pdf: 3226956 bytes, checksum: 9d13dc0a531d222433b4267892b08ac2 (MD5) / Made available in DSpace on 2016-08-18T13:03:27Z (GMT). No. of bitstreams: 1 Jonas Garcia de Almeida.pdf: 3226956 bytes, checksum: 9d13dc0a531d222433b4267892b08ac2 (MD5) Previous issue date: 2015-03-26 / For human identification purposes in Forence area, the United States created the database (DB) of DNA Index System Combined (CODIS), containing 13 loci: CSF1PO, FGA, TH01, TPOX, VWA, D3S1358, D5S818, D7S820, D8S1179, D13S317, D16S539, D18S51 and D21S11. Then the DBs of the markers of the Y-STR X-STRs haplotypes were created. Comparison of the obtained data was conducted from allele frequencies (AF), paternity index (PI) by marker and paternity probabilities (PPs) obtained from nine BD population, one of those national (BR), using 241 cases containing the PP 99.99 using 13 loci STRs (CODIS) and 2 more STRs markers of Penta D- and E, granted by the BDs of Núcleo de Pesquisas Replicon, of the Pontifícia Universidade Católica de Goiás (PUC - GO) and LaGene, Laboratório Central de Saúde Pública (SES - GO). 241 cases were analyzed in Trio (alleged father, son and progenitor) and Duo (alleged father and son) situations, using 13 STRs markers (CODIS) by Bayes theorem and the Likelihood Ratio (LR). The results were submitted to the Kruskal-Wallis, chi-square, Spearman correlation and Fishers exact tests and SPSS software (2010). Static analysis of RV cases containing the Trio resulted in 1,148 results PPs 99.99% and 21 results PPs <99.99%, and in a position to Duo 1686 results with PPs 99.99% and 483 results PPs <99.99%. The analysis of the IP average of each marker showed the D21S11 of STRs markers and FGA with the highest power of inclusion and TH01 and D3S1358 with the lowest power of inclusion. The PPs did not show significantly different, containing mostly positive correlation, of moderate to strong, between 8 BDs compared to the BR population databases. This study demonstrated the statistics interference that each allele frequencies DB can have in PP calculations using only 13 loci of genetic, thus making it more significant in cases Duo situation. According to the information available in databases of gene frequencies of the different geographical regions of Brazil, it became possible to conclude that the allele frequencies obtained and the IPs per marker and PPs obtained, suggest strong similarities to those found in the national database. / Para fins de identificação humana na área forense os Estados Unidos criou o Banco de Dados (BD) do Sistema de Índice de DNA Combinado (CODIS), contendo 13 loci: CSF1PO, FGA, TH01, TPOX, VWA, D3S1358, D5S818, D7S820, D8S1179, D13S317, D16S539, D18S51 e D21S11. Foi realizado no presente estudo, a comparação dos dados obtidos das frequências alélicas (FA), índices de paternidade (IP) por marcador e probabilidades de paternidade (PPs) obtidas a partir de 9 BD populacionais, sendo 1 nacional (BR), utilizando 241 casos contendo a PP 99,99 com o uso de 13 loci STRs (CODIS) e mais 2 marcadores STRs do Penta-D e E, cedidos pelos BDs do Núcleo de Pesquisa Replicon, da Pontifícia Universidade Católica de Goiás (PUC-GO) e do LaGene, Laboratório Central de Saúde Pública (SES-GO). Os 241 casos foram analisados nas situações de Trio (suposto pai, filho e genitora) e Duo (suposto pai e filho) usando 13 marcadores STRs (CODIS) através do teorema de Bayes e pela Razão de Verossimilhança (RV). Os resultados foram submetidos aos testes de Kruskal-Wallis, Qui-quadrado, Correlação de Spearman e Exato de Fisher usando o software SPSS (2010). As análises estáticas da RV dos casos contendo o Trio resultaram em 1.148 resultados com PPs 99,99% e 21 resultados com PPs < 99,99%, e em situação de Duo 1.686 resultados com PPs 99,99% e 483 resultados com PPs < 99,99%. As análises das médias dos IPs de cada marcador demonstraram os marcadores STRs do D21S11 e FGA com os maiores poder de inclusão e TH01 juntamente com o D3S1358 os menores poderes de inclusão. As PPs não apresentaram diferenças estatísticas significativas contendo em sua maioria correlações positivas de moderadas a fortes entre os 8 BDs comparados ao BDs populacional do BR. Este estudo demonstrou as interferências estatísticas que cada BDs de frequências alélicas pode exercer nos cálculos de PP quando se utiliza apenas 13 loci para confirmação de vínculo genético, tornando-se assim mais significativas nos casos em situação de Duo. De acordo as informações disponíveis nos BDs de frequências alélicas das diferentes regiões geográficas do Brasil, tornou-se possível concluir, que as frequências alélicas obtidas, bem como os IPs por marcador e as PPs obtidas, sugerem fortes similaridades às encontradas no banco de dados nacional.

The advances in molecular biology have increased the production of enormous amount of genetic information in a small period of time. This capacity of data production motivated the researchers to increase the rhythm of their researches. This necessity demands the use of efficient softwares in order to manage these data. Besides this, it also demands the development of good softwares in order to assist the researchers in the task of analyzing the data and giving them a biological meaning in a brief space of time. This work proposes a software model that will support the study of Forensic DNA, whose main repository is the autossomic DNA. This software intends to support the researchers in the identification of condemned persons or persons that are suspected of a crime. It also intends to assist the researchers in the study of paternity and the search for disappeared persons. The results of this work will be applied in the Forensic DNA Laboratory of UFAL. The software modeled here has four modules study of paternity , criminal , disappeared people and the bank of populational frequencies . The modules were modeled independently from each other, considering the specifications related to the analysis of genetic links. The software was developed using the JAVA programming language together with PostgreSQL database. Both are free software and have an excellent relationship between cost and benefit usage / Fundação de Amparo a Pesquisa do Estado de Alagoas / Os avanços da biologia molecular vêm favorecendo a geração de uma enorme quantidade de informações genéticas em um tempo cada vez menor. Essa capacidade de geração de dados permite que os pesquisadores acelerem o ritmo de suas pesquisas, exigindo a utilização de ferramentas eficientes para o gerenciamento desses dados. Outra necessidade está relacionada com o desenvolvimento de ferramentas computacionais com capacidade de auxiliar na tarefa de análisar e dar um significado biológico a estes dados em um breve espaço de tempo para os pesquisadores. Este trabalho propõe a modelagem de um ambiente de apoio à análise e ao estudo do DNA Forense, cujo principal repositório seja o DNA autossômico. Este ambiente visa dar suporte a identificação de pessoas condenadas ou suspeitas de ter realizado algum tipo de crime contra a sociedade, bem como auxiliar no estudo de paternidade e na busca de pessoas desaparecidas. Este ambiente irá atender ao Laboratório de DNA Forense, da UFAL, que vêm realizando estas atividades. O modelo do ambiente aqui proposto, possui quatro módulos, estudo de paternidade , criminal , desaparecido e o banco de freqüência das populações . Os módulos foram modelados de forma que funcionem independentemente, atendendo as especificações inerentes à análise sobre vínculo genético. O sistema foi desenvolvido na linguagem de programação JAVA com banco de dados PostgreSQL. Ambas as ferramentas possuem característica de software aberto e uma relação custo/benefício excelentes