Rating scales involving more than two response categories are a popular response format in measurement in education, health and business sciences. Their primary purpose lies in the increase of information and thus measurement precision. For these objectives to be met, the response scale has to provide valid scores with higher numbers reflecting more of the property to be measured. Thus, the response scale is closely linked to construct validity since any kind of malfunctioning would jeopardize measurement. While tests of fit are not necessarily sensitive to violations of the assumed order of response categories, the order of empirical threshold estimates provides insight into the functionality of the scale. The Rasch model and, specifically, the so-called Rasch-Andrich thresholds are unique in providing this kind of evidence. The conclusion whether thresholds are to be considered truly ordered or disordered can be based on empirical point estimates of thresholds. Alternatively, statistical tests can be carried out taking standard errors of threshold estimates into account. Such tests might either stress the need for evidence of ordered thresholds or the need for a lack of evidence of disordered thresholds. Both approaches are associated with unacceptably high error rates, though. A hybrid approach that accounts for both evidence of ordered and disordered thresholds is suggested as a compromise. While the usefulness of statistical tests for a given data set is still limited, they provide some guidance in terms of a modified response scale in future applications.

Dissertação (mestrado)—Universidade de Brasília, Departamento de Estatística, 2013. / Submitted by Albânia Cézar de Melo (albania@bce.unb.br) on 2013-09-23T16:37:07Z No. of bitstreams: 1 2013_RomuloAndradeSilva.pdf: 4141117 bytes, checksum: 94af0b4f18ed2268977af4871560d723 (MD5) / Approved for entry into archive by Guimaraes Jacqueline(jacqueline.guimaraes@bce.unb.br) on 2013-09-25T12:14:02Z (GMT) No. of bitstreams: 1 2013_RomuloAndradeSilva.pdf: 4141117 bytes, checksum: 94af0b4f18ed2268977af4871560d723 (MD5) / Made available in DSpace on 2013-09-25T12:14:02Z (GMT). No. of bitstreams: 1 2013_RomuloAndradeSilva.pdf: 4141117 bytes, checksum: 94af0b4f18ed2268977af4871560d723 (MD5) / Uma das pressuposições, no processo de estimação dos parâmetros dos modelos tradicionais de resposta ao item, é a independência condicional entre as respostas de diferentes indivíduos. Porém, muitas vezes essa pressuposição é relaxada, por exemplo, quando aplicada em larga escala nas avaliações de sistemas educacionais, o que pode ocasionar variabilidade extra não considerada pelos modelos usuais. A proposta é usar potenciais fontes de heterogeneidade como variáveis explicativas de um efeito aleatório multiplicativo no modelo de Rasch. Esse efeito, consequentemente, acomodará a superdispersão presente nos dados e controlará a pressuposição de independência condicional entre clusters de indivíduos. O modelo foi ajustado aos dados da Prova Brasil 2007, trazendo novas interpretações de grupos. Contudo, a nova abordagem probabilística de considerar informações extras dos indivíduos no momento do ajuste se mostra útil na fase de calibração dos itens. ______________________________________________________________________________ ABSTRACT / One of the assumptions in the estimation process of the parameters of the traditional models of itemresponse is conditional independence between the responses of di erent individuals. Nonetheless, this assumption is often relaxed, for example,when applied to large-scale evaluations of educational systems, which can cause extra variability not considered by usual models. The proposal is to use potential sources of heterogeneity as explanatory variables in a random e ect multiplicative Rasch model. This e ect, therefore, will accommodate the overdispersion in the data and control the conditional independence assumption between clusters of individuals. The model was adjusted to data from ProvaBrasil 2007 (Brazil Test 2007), bringing new interpretations of groups. However, the new probabilistic approach on considering extra information of individuals at the time of adjustment proves to be useful in the calibration phase of the items.

Thesis advisor: Larry H. Ludlow / The focus on science, technology, engineering, and mathematics fields has noticeably increased in recent decades with the rapid growth in technology. Mathematical knowledge and competence is in many ways a gateway to scientific and technical development, and therefore careers (Prieto & Delgado, 2007). Unfortunately, national and international assessments of mathematics achievement (e.g., Kena et al., 2016; NCES, 2015; OECD, 2013) show that schools in the United States are not helping students achieve acceptable levels of mathematical and numerical proficiency. Therefore, we need to understand how various factors, including mathematics anxiety, affect student performance and persistence in STEM. To do this, a valid and reliable measure of mathematics anxiety is needed. Existing instruments to measure mathematics anxiety have been insufficient in several ways, including in their conceptualization of the construct and use of classical test theory over Rasch measurement theory methods.In this study, an instrument – the Comparative Mathematics Anxiety Scale (CMAS) – was developed to measure the three-dimensional conceptualization of mathematics anxiety that Dr. Caroline Vuilleumier and I originated. A unique, comparative item format adapted from Ludlow et al. (2014, 2019) and Rasch measurement theory (Rasch, 1960/1980) were utilized to mitigate some of the limitations of existing instruments. The overarching research question and three sub-research questions explored whether the CMAS could measure mathematics anxiety in a valid, reliable, and meaningful way. This study employed a seven-step iterative scale development process and was accomplished across three rounds. Ultimately, twenty-three third-person items were developed to capture the emotional-attitudinal, mental-cognitive, and physical-somatic dimensions of mathematics anxiety. Using the Rasch rating scale model, the outcome was the 23-item CMAS that reliably and validly measures increasing levels of three dimensions of mathematics anxiety. The distribution of the items mostly confirmed their hypothesized order and the Rasch measurement theory principles. The scale also provides meaningful interpretations of what a raw score means regarding a student’s experience of emotional-attitudinal, mental-cognitive, and physical-somatic mathematics anxiety. Overall, the findings suggest that the novel approach of combining Rasch measurement theory with third-person items and comparative response options can be successful in developing a scale that measures an important construct. Furthermore, the scale can provide the evidence needed in the provision of interventions and in research to reduce students’ overall experience of mathematics anxiety. / Thesis (PhD) — Boston College, 2021. / Submitted to: Boston College. Lynch School of Education. / Discipline: Educational Research, Measurement and Evaluation.

Thesis advisor: Larry H. Ludlow / Interaction with faculty is one of the most important aspects of completing an undergraduate degree (Chambliss & Takacs, 2014). At traditional colleges and universities, much of this interaction takes place within the classroom. However, out-of-class communication (OCC) is also an important part of the college-going experience. Participation in OCC has been associated with many positive undergraduate outcomes, such as motivation (Komarraju et al., 2010) and course grades (Micari & Pazos, 2012). Prior measurement instruments related to OCC suffered from limitations with respect to construct definition and methodological procedures that limit the interpretability and utility of the scores they yield. My dissertation ameliorates these issues in constructing a new instrument that measures students’ perceptions of faculty availability outside of class. This instrument is built using Rasch/Guttman Scenario (RGS) scale methodology, which brings together the frameworks of Rasch measurement (Rasch, 1960/80) and Guttman facet theory design (Guttman, 1954; Guttman 1959). Two scales, each containing seven short scenarios that function as items, were constructed: the Physical Accessibility Scale (PAS) and the Social Engagement Scale (SES). Together, these two scales comprise the Out-of-Class Availability Scales (OCAS). Three facets of physical accessibility and social engagement are identified and represented within the items: arranged meetings, chance encounters, and email. The OCAS development process and analysis results presented within my dissertation suggest that the RGS methodology is useful for capturing students’ perceptions of faculty availability outside of class. The OCAS can also be used by others to conduct future research on the topic of OCC. Because they measure students’ perceptions of availability and not frequency of OCC, the OCAS have value as a potential faculty evaluation tool. Even if students choose not to interact with a particular faculty member outside of class, they would still ideally find that faculty member available for such interaction should the need arise. Finally, the RGS scale development process ensures that OCAS scores are accompanied by qualitative descriptions, which enhances their utility and measurement value. / Thesis (PhD) — Boston College, 2020. / Submitted to: Boston College. Lynch School of Education. / Discipline: Educational Research, Measurement and Evaluation.

Mémoire numérisé par la Direction des bibliothèques de l'Université de Montréal.

The current study examined and compared the use of Rasch measurement, common factor analysis (EFA), and confirmatory factor analysis (CFA) in establishing construct validity evidence based on internal structure with multi-item scales measuring middle and secondary students' achievement goal orientation and academic self-efficacy. 1054 complete responses were received to 18 items measuring mastery, performance-approach, performance-avoidance, and self-efficacy. Items in each subscale were first analyzed by each method as a unidimensional unit. Items were then analyzed by each method as a multidimensional unit. Results showed that when scales were analyzed individually, all three methods corroborated unidimensionality; however, when all items were analyzed together, the multi-factor model identified through EFA was not supported by the Rasch or CFA analysis. While EFA provided the best information about individual item functioning, Rasch provided important, additional information about rating scale functioning and item fit that helps diagnose poorly performing items. Results also support the use of EFA in evaluating the suitability of a scale for meeting the unidimensionality requirement of item response models. While the mastery and efficacy scales performed in a manner consistent with the motivation literature, the approach and avoidance subscales did not. Conclusions from the study include the need for better explication of conjoint use of classical and modern test theory methods in instrument development, more exposure of current and future researchers to the foundations of measurement theory, and more research about the saliency of measuring performance-avoidance.

Satisfaction is a central concept in marketing. However in recent years, satisfaction has come under increasing criticism. Its ability to predict post-purchase behaviour has not been established and the importance of the word satisfaction to consumers has been questioned. Current satisfaction measures are inadequate, as they fail to discriminate between respondents, with the majority of respondents regularly endorsing the most positive answer category available. The limited discrimination of existing scales suggests that only a small part of the unfavourable/favourable evaluation, rather than the entire dimension is being measured. The overwhelming use of the most positive answer category, in traditional scales, illustrates that they fail to capture highly positive evaluations. Affective Response to Consumption (ARC) is conceptualised as an extension to satisfaction. The conceptualisation shifts the emphasis from a scale relying on one, rather weak, emotional feeling -- satisfaction -- to a multitude of emotional feelings, including highly positive terms. A scale measuring ARC is developed in an alternative measurement paradigm -- Rasch Modelling -- to the dominant paradigm for scale development in marketing -- Classical Test Theory. The characteristics of Rasch Modelling are particularly useful, when measuring a concept like ARC, that captures the entire dimension of unfavourable/favourable evaluations and includes terms of markedly different intensity.

The domestic dog, Canis lupus familiaris, is a unique model system for the dissection of hereditary diseases. Selective breeding practices have created more than 300 distinct breeds of dogs, born from a desire to create specific physical and behavioral characteristics. Breeds represent closed breeding populations and the extensive records maintained for members of each breed (e.g., multi-generational pedigrees, veterinary medical records) present an incredible tool for genetic research. Two closed populations were used in the work presented here: a colony of mixed-breed dogs segregating resistance and sensitivity to cholesterol feeding, and a purebred pet population of Havanese experiencing a high frequency of developmental abnormalities. Estimates of heritability were calculated for each disease to evaluate the degree of phenotypic variation attributable to genetics among dogs in the populations used. A heritability of 0.55 (± 0.16) was identified for cholesterol resistance and sensitivity in the mixed-breed colony. The small sample size prevented the use of complex segregation analyses to examine mode of transmission. A heritability of 0.36 (± 0.26) was calculated for the composite phenotype in the Havanese, encompassing the spectrum of abnormalities in the breed. Polygenic inheritance was identified for the composite phenotype, but the action of a major gene was identified by complex segregation analyses in the Havanese. Complex diseases preclude the use of a candidate gene approach, owing to the multitude of genes involved in the disease process. Whole genome screens provide a practical approach to the identification of chromosomal region(s) associated with a disease phenotype by narrowing the search for candidate gene(s). The Minimal Screening Set – 2 (MSS-2) was used in the present studies to evaluate the segregation of microsatellite markers in pedigrees for both the mixed-breed colony and the Havanese. No significant LOD scores were identified, though suggestive LOD scores were obtained in both analyses. A canine-specific oligonucleotide microarray was used to create gene expression profiles for developmental abnormalities in the Havanese and for cholesterol sensitivity in the mixed-breed colony dogs. Distinct expression profiles were generated for each group, and several genes of interest were identified as being both differentially expressed (>±2-fold change) and statistically significant (p-value<0.05).

This study is aim to develop and validate the Creative Life Style Check List by analysis of the literature scale and integration expert opinions. This scale for the self-report scale, a total of 28 questions, uses Likert five-point scale scoring. There were four subscale, ¡§wildly life experiences ¡§, ¡§imagination¡¨, ¡§new creation and expression¡¨, and ¡§openness mind¡¨. Purposive sampling method to extract the Sun Yat-sen University, Ministry of students as the study sample, the pre-pilot scale a total of 304 samples, a total of 983 formal scale samples. The MPCM analysis by the Rasch model is to examine the scale of reliability and validity. Content validity Infit MNSQ is between 0.75 ~ 1.3logits, each subscale internal consistency from .831 to .893. Scale internal consistency reliability was .93. Construct validity of the comparison of PCM and MPCM that estimated residuals of MPCM smaller than PCM. That means four-dimensions model more fit than one-dimension model. The correlation between each subscale is in the range of 0.684 ~ 0.861. The validity generalization, the scale of the boys and girls difficulty estimate differences in the range of 0.000 to 0.198, on behalf of this scale is no gender differential item functioning (DIF). Participants separated reliability coefficient of 0.844 to 0.900 that means this scale can stability measure the location of participants of the creativity construct. Difficulty estimated values of each subscale in the formal questionnaire and the pre-test scale ranged from .815 to .944 is high correlation. And p <0.05 show the scale has cross-sample stability.

The domestic dog, Canis lupus familiaris, is a unique model system for the dissection of hereditary diseases. Selective breeding practices have created more than 300 distinct breeds of dogs, born from a desire to create specific physical and behavioral characteristics. Breeds represent closed breeding populations and the extensive records maintained for members of each breed (e.g., multi-generational pedigrees, veterinary medical records) present an incredible tool for genetic research. Two closed populations were used in the work presented here: a colony of mixed-breed dogs segregating resistance and sensitivity to cholesterol feeding, and a purebred pet population of Havanese experiencing a high frequency of developmental abnormalities. Estimates of heritability were calculated for each disease to evaluate the degree of phenotypic variation attributable to genetics among dogs in the populations used. A heritability of 0.55 (± 0.16) was identified for cholesterol resistance and sensitivity in the mixed-breed colony. The small sample size prevented the use of complex segregation analyses to examine mode of transmission. A heritability of 0.36 (± 0.26) was calculated for the composite phenotype in the Havanese, encompassing the spectrum of abnormalities in the breed. Polygenic inheritance was identified for the composite phenotype, but the action of a major gene was identified by complex segregation analyses in the Havanese. Complex diseases preclude the use of a candidate gene approach, owing to the multitude of genes involved in the disease process. Whole genome screens provide a practical approach to the identification of chromosomal region(s) associated with a disease phenotype by narrowing the search for candidate gene(s). The Minimal Screening Set – 2 (MSS-2) was used in the present studies to evaluate the segregation of microsatellite markers in pedigrees for both the mixed-breed colony and the Havanese. No significant LOD scores were identified, though suggestive LOD scores were obtained in both analyses. A canine-specific oligonucleotide microarray was used to create gene expression profiles for developmental abnormalities in the Havanese and for cholesterol sensitivity in the mixed-breed colony dogs. Distinct expression profiles were generated for each group, and several genes of interest were identified as being both differentially expressed (>±2-fold change) and statistically significant (p-value<0.05).