Linear mixed models and generalized linear mixed models are random-effects models widely applied to analyze clustered or hierarchical data. Generally, random effects are often assumed to be normally distributed in the context of mixed models. However, in the mixed-effects logistic model, the violation of the assumption of normally distributed random effects may result in inconsistency for estimates of some fixed effects and the variance component of random effects when the variance of the random-effects distribution is large. On the other hand, summary statistics used for assessing goodness of fit in the ordinary logistic regression models may not be directly applicable to the mixed-effects logistic models. In this dissertation, we present our investigations of two independent studies related to goodness-of-fit tests in generalized linear mixed models. First, we consider a semi-nonparametric density representation for the random effects distribution and provide a formal statistical test for testing normality of the random-effects distribution in the mixed-effects logistic models. We obtain estimates of parameters by using a non-likelihood-based estimation procedure. Additionally, we not only evaluate the type I error rate of the proposed test statistic through asymptotic results, but also carry out a bootstrap hypothesis testing procedure to control the inflation of the type I error rate and to study the power performance of the proposed test statistic. Further, the methodology is illustrated by revisiting a case study in mental health. Second, to improve assessment of the model fit in the mixed-effects logistic models, we apply the nonparametric local polynomial smoothed residuals over within-cluster continuous covariates to the unweighted sum of squares statistic for assessing the goodness-of-fit of the logistic multilevel models. We perform a simulation study to evaluate the type I error rate and the power performance for detecting a missing quadratic or interaction term of fixed effects using the kernel smoothed unweighted sum of squares statistic based on the local polynomial smoothed residuals over x-space. We also use a real data set in clinical trials to illustrate this application.

Ménière’s disease (MD) is an idiopathic inner ear disorder, characterised by episodes of vertigo, tinnitus, sensorineural hearing loss, and aural fullness in the affected ear. The relatively high variability of symptomological changes renders it difficult to confirm the MD diagnosis. The purpose of this study is to compare the diagnostic power of an instrumental method, electrocochleography (ECochG), and two subjective methods, including the criteria based on the clinical guidelines provided by the American Academy of Otolaryngology-Head and Neck Surgery Committee on Hearing Equilibrium (AAO-HNS CHE) and Gibson’s Score. A quota sampling method was used to include subjects. A total of 250 potential MD patients who were referred to the Department of Otolaryngology at the Christchurch Hospital between year 1994 and 2009 have had their signs and symptoms documented and ECochG testing completed. A selection of details obtained from both AAO-HNS CHE and ECochG assessment results were examined as a chart review in regard to its function as a diagnostic tool for MD. The between-method reliability was found to be high, with a few disagreements on individual diagnosis. Based on a receiver operating characteristic (ROC) curve analysis, the ECochG measures were shown to be pertinent to the diagnosis of MD. It was also found that patients tested “positive”, as compared with those tested “negative”, tended to show higher correlations among the four key symptoms of MD and among the ECochG measures derived from the auditory evoked responses to tone bursts at frequencies in close proximity to each other.

This thesis contributes to influence analysis in nonlinear regression and in particular the detection of influential observations. The focus is on a regression model with a known mean function, which is nonlinear in its parameters and where the function is chosen according to the knowledge about the process generating the data. The error term in the regression model is assumed to be additive. The main goal of this thesis is to work out diagnostic measures for assessing the influence of observations on various results from a nonlinear regression analysis. The obtained results comprise diagnostic tools for detecting observations that, individually or jointly with some other observations, are influential on the parameter estimates. Moreover, assessing conditional influence, i.e. the influence of an observation conditional on the deletion of another observation, is of interest. This can help to identify influential observations which could be missed due to complex relationships among the observations. Novelties of the proposed diagnostic tools include the possibility to assess influence of observations on a specific parameter estimate and to assess influence of multiple observations. A further emphasis of this thesis is on the observations' influence on the outcome of a hypothesis testing procedure based on Rao's score test. An innovative solution to the problem of visual identification of influential observations regarding the score test statistic obtained in this thesis is the so called added parameter plot. As a complement to the added parameter plot, new diagnostic measures are derived for assessing the influence of single and multiple observations on the score test statistic.

Current methods of determining spinal cord recovery in mice, post-directed injury, are qualitative measures. This is due to the small size and quickness of mice. This thesis presents a design for a gait acquisition and analysis system able to capture the footfalls of a mouse, extract position and timing data, and report quantitative gait metrics to the operator. These metrics can then be used to evaluate the recovery of the mouse. This work presents the design evolution of the system, from initial sensor design concepts through prototyping and testing to the final implementation. The system utilizes a machine vision camera, a well-designed walkway enclosure, and image processing techniques to capture and analyze paw strikes. Quantitative results gained from live animal experiments are presented, and it is shown how the measurements can be used to determine healthy, injured, and recovered gait.

An original composition in five movements for voice and a chamber orchestra of eleven instruments. The first movement is an overture; the second and fifth movements have text by Sara Teasdale; the third and fourth movements have text by Edna St. Vincent Millay.

Conventional charcoal and firewood are the main source of energy in Haiti. They provide up to 90% of the country’s energy for domestic and industrial use, resulting in severe environmental and health issues. The present study is initiated to better understand the reasons why two promising alternative technologies (improved cookstoves and alternative charcoal briquettes) have experienced low adoption in Haiti. The research was carried out in two districts in southern Haiti where the improved stoves and briquettes production units exist and where households benefited from a program distributing the improved stoves. This project contributes to the literature by gauging interest in the improved stove and briquettes, as well as their specific characteristics. It helps understand factors that affect the adoption and dis-adoption of the technologies. Additionally, the research measures tangible benefits for households that adopted the improved stoves. The study reveals that the use of the improved stoves lowers fuel expenditures by 14.6 cents/day to 23.6 cents/day. Haitian consumers are interested in both the stove and briquettes, but their willingness-to-pay depends on their personal characteristics such as location and income. The study has revealed two surprising results as well: Unnecessary dis-adoption of the stoves occurs because the two technologies were needlessly marketed together. Despite the target audience, which is poor and rural consumers, the improved stove is perceived as a rich, urban user’s technology.

There are two main purposes of this thesis. The first one is to get a better understanding of the aspects affecting customers’ loyalty in the telecommunication industry, in the context of when customers are using the services. The second purpose is to look into what the case company gains by having customers that are more loyal, where the degree of loyalty is measured by the Net Promoter Score metric. The methodology used to carry out the research is a case study with an approach that is both qualitative and quantitative. Where the quantitative approach has the largest share. Two datasets have been used in this thesis, one collected by the authors by sending out surveys and one collected beforehand at the case company. The survey created by the authors aim to let customers assess the satisfaction level with technical and non-technical aspects that can affect loyalty. The dataset that is already collected by the case company document the initial degree of loyalty of customers along with the revenue per customers over a period of years. The two datasets are used for the two different research purposes respectively. The statistical analysis for the data is conducted using the statistical tool Minitab. The findings for the first purpose are that our survey questions can be split into three categories using factor analysis. The categories are Perceived mobile multimedia quality, Perceived broadband multimedia quality and General perceptions. The first two categories are driving customer loyalty and the third category are indicators of customer loyalty. For the second purpose the findings are that the case company has different gains of more loyal customers depending on if the customers are either mobile or broadband customers. More loyal mobile customers stay longer as customers and also buy more. More loyal broadband customers only stay longer as customers. The practical implications of the findings are that the case company has to think of customer loyalty in new ways. There are more indicators of if a customer is loyal than the Net Promoter Score, these are for example customer satisfaction, perceived brand value, perceived overall quality, perceived customization etc. Therefore it would be better to measure customer loyalty not only with the Net Promoter Score Metric but to pick out 2-3 indicators to ask the customer and create an average index for all the questions that can represent the customers’ loyalty. Furthermore there are not a specific variable that drives customer loyalty more or less, several aspects are acting together in two high level groups. Another practical implication is that the gains of more loyal customers are higher for mobile customers since they buy more from the case company and stay longer as customers. However, for broadband customers, they only stay longer. Therefore the Net Promoter Score is not as useful to track for broadband customers. Either the broadband customers should have more opportunities to buy more or another metric should be used for broadband customers.

Syfte Syftet med föreliggande studie var att undersöka genomförbarheten av en hög-intensiv träningsintervention efter knäplastikoperation samt hur knäplastikopererade personers självvalda gånghastighet, maximal vadmuskelstyrka, vadmuskeluthållighet och självupplevda symptom, smärta, ADL funktion och hälsorelaterad livskvalité påverkas av tidigt insatt hög-intensiv styrketräning av vadmuskulaturen. Metod Femton forskningspersoner randomiserades till en träningsgrupp (n=7), som utförde hemträningsprogram i form av hög-intensiv vadmuskelträning samt till en kontrollgrupp (n=8) som utförde samma rehabiliteringsprogram som träningsgruppen utan hög-intensiv vadmuskelträning. En testprocedur genomfördes 3 respektive 12 veckor efter knäplastikoperationen. Självvald gånghastighet mättes genom ett 30 meters gångtest, vadmuskeluthållighet mättes med ett standardiserat tåhävningstest på ett ben, maximal isometrisk vadmuskelstyrka mättes med hjälp av en isometrisk dynamometer och självskattad symptom, smärta, ADL funktion och livskvalité utvärderades genom Knee injury Osteoarthritis Outcome Score (KOOS). Resultat Denna studie påvisar viss problematik med följsamheten i träningsregim för båda grupperna. Båda gruppernas sammanslagna resultat uppvisade signikant förbättrad självvald gånghastighet, maximal vadmuskelstyrka och uthållighet över tid. Trots detta ökade träningsgruppen den självvalda gånghastigheten signifikant mer än kontrollgruppen över tid. Träningsgruppen hade även mindre besvär på grund av smärta 12 veckor efter operationen. Båda gruppernas sammanslagna resultat förbättrades över tid gällande deras egen skattning av symptom, smärta, ADL funktion och livskvalité. Däremot förbättrade träningsgruppen alla variabler signifikant medan det för kontrollgruppen endast var livskvalitén som förbättrades signifikant över tid. Konklusion Tidigt insatt hög-intensiv vadmuskelträning efter knäplastikoperation är genomförbart och kan ha en viss positiv effekt på personens egen upplevelse av symtom, smärta, funktion och hälsorelaterad livskvalité. Självvald gånghastighet förefaller öka mer vid tidigt insatt hög-intensiv vadmuskelträning efter nio veckors träning jämfört med samma rehabiliteringsprogram utan hög-intensiv vadmuskelträning. / Aim The aim of this study was to determine the feasibility of a high-intensive exercise intervention following a knee arthroplasty, and also to evaluate possible effects on self-selected walking speed, calf muscle strength and self-rated symptoms, pain, ADL function and quality of life after high-intensive calf muscle training in an early post-operative phase. Method Fifteen participants were randomized into an exercise group (n=7), who performed a home-exercise program, consisting of high-intensive calf muscle exercise, and into a control group (n=8) who performed the same rehabilitation program but without the high-intensive calf muscle training. A test procedure was conducted after 3 and 12 weeks following the knee arthroplasty. Self-selected walking speed was measured by a 30 meter walking test, calf muscle endurance training was measured with a standardized one-leg heel-rise test, maximal isometric calf muscle strength was measured with an isometric dynamometer, and self-rated symptom, pain, ADL function and quality of life were evaluated with Knee injury Osteoarthritis Outcome Score (KOOS). Results There were problems with adherence to exercise regimes in both groups. The results of both groups demonstrated improved self-selected walking speed, calf muscle endurance and maximal isometric calf muscle strength over time. However, the exercise group increased self-selected walking speed more than the control group. The exercise group reported fewer disorders due to pain 12 weeks after surgery. The results of both groups showed improved self-rated symptoms, pain, ADL function and quality of life over time, however, the exercise group improved all these variables significantly, while the control group significantly improved the quality of life aspect. Conclusions Early initiated high-intensive calf-muscle training following a knee artroplasthy is feasible and may have positive effects on self-rated symptoms, pain, ADL function and quality of life. Also, self-selected walking speed increased more with early initiated high-intensive calf muscle training after nine weeks when compared to the same rehabilitation program without high-intensive calf muscle training.

The great challenge that exists in cross-cultural composition is finding commonalities of intonation, style, formality, and instrumentation. In the case of creating a composition for Persian tar and symphony orchestra, a number of challenges emerge. The goal of this composition is to find compatible musical elements between Persian and western music. The most difficult challenge of composing in this genre is finding compatible musical modes to generate optimum intonation. For this piece, the western pitch of Bb is used to help the intonation and tonal production of the tar. My approach to style involves a sensitive application using elements of classical Persian style and formality known as dãstgãh composition. Generally, Persian music places strong emphasis on improvisation, non-repetitive scalar melodies, variation of melodic and rhythmic material, tasteful ornamentation, and creating an inspiring musical atmosphere. Often music is composed with a story, poem or song in the mind of the performer. Typically, the instrumentation is very small and intimate, consisting of a tar performer, accompanied by a musician playing a daff (a Persian frame drum much like an Irish bodhran). I have composed a work that features many of the above characteristics. The Concerto for Tar and Orchestra encompasses many salient features found in Persian classical music, while also providing symphonic orchestration that adds complimentary western musical elements. Most of the instrumentation of the work adheres to standard orchestral instrumentation. I have included some Persian hand-drums to the orchestration (a tombak and a dohol) to play alongside standard symphonic percussion to add a more cross-cultural quality to the orchestration. The concerto also owes its inspiration to the story “Umar and the Harpist”, which is found in Jellaludin Rumi’s revered, masterful literary work entitled the Masnavi. I used the imagery found in this story to generate much of the dramatic contrast contained in the concerto. In the end, a composition has emerged which brings two distinct classical music traditions together in an effective cross-cultural form of expression.

預測模型已經被廣泛運用在日常生活中，例如銀行信用評比、消費者行為或是疾病的預測等等。然而不論在建構或使用預測模型的時候，我們都會在訓練資料或是測試資料中遇到遺失值的問題，因而降低預測的表現。面對遺失值有很多種處理方式，刪除、填補、模型建構以及機器學習都是可以使用的方法；除此之外，直接用某個成本去取得遺失值也是一個選擇。 本研究著重的議題是用某成本去取得遺失值，並且利用決策樹(因為其在建構時可以容納遺失值)來當作預測模型，希望可以找到用較低的成本的填值方法達到較高的準確率。我們延續過去Error Sampling中Uncertainty Score的概念與邏輯。提出U-Sampling來判斷不同特徵值的「重要性排序」。相較於過去Error Sampling用「受試者」(row-based)的重要性來排序。U-Sampling是根據「特徵值」(column-based)的重要性來排序。 我們用8組UCI machine Learning Repository的資料進行兩組實驗，分別讓訓練資料以及測試資料含有一定比例的遺失值。再利用U-Sampling、Random Sampling以及過去文獻所提及的Error Sampling作準確率和錯誤減少率的比較。實驗結果顯示在訓練資料有遺失值的情況，U-Sampling在70%以上的檔案表現較佳；而在測試資料有遺失值的情況，U-Sampling則是在87.5%的檔案表現較佳。 另外，我們也研究了對於不同的遺失比例對於上述方法的效果是否有影響，可以用來判斷哪種情況比較適用哪一種選值方法。希望透過U-Sampling，可以先挑選重要的特徵值來填補，用較少的遺失值取得就得到較高的準確率，也因此可以節省處理遺失值的成本。