Search

In the past two decades, there have been calls for statistical reform in psychology. Three key concepts within reform are effect sizes, confidence intervals and statistical power. The aim of this thesis was to examine the use and knowledge of these particular concepts, to examine whether researchers are suitably equipped to incorporate them into their research. This thesis consists of five studies. Study 1 reviewed author guidelines across 100 psychology journals, to look for any statistical recommendations. Study 2 (n = 247) and Study 3 (n = 56) examined the use and knowledge of effect sizes using a questionnaire and online experiment. Study 4 surveyed psychology researchers on their use and knowledge of confidence intervals (n = 206). Similarly, Study 5 surveyed psychology researchers on their use and knowledge of power analyses and statistical power (n = 214). Typically, psychology journals expect authors to report effect sizes in their work, although there are fewer expectations related to confidence intervals. Power analyses are also frequently encouraged for sample size justification. Self-reported use of effect sizes, confidence intervals and power analyses was high, while common barriers to use included a lack of knowledge, a lack of motivation, and the influence of academic peers. While knowledge of effect sizes was quite high, they appear to only be understood in relatively limited contexts. In contrast, both confidence intervals and statistical power appear to be frequently misunderstood, and many researchers find power analysis calculations difficult. Researchers would benefit from increased education and support to encourage them to confidently adopt an assortment of statistics in their work, and more effort must be made to prevent statistical changes from becoming a new series of tick-box exercises that do not improve the integrity of psychological research.

Adopting ambitious renewable energy targets has profound social, economic, and environmental implications, at local and global scales. Indeed, these targets have raised deep questions about social justice in capitalist societies attempting to pursue clean energy transitions. To understand how these transitions occur, we must understand dynamics of community acceptance that are linked to the politics of local approval of siting Renewable Energy Technologies (RETs). As a result, it is vital to identify links to justice as an important explanatory factor affecting community acceptance of RETs. Wind energy offers an emblematic case of these dynamics, highlighting the tensions between the technical and the social. At a technical level, the efficiency of wind energy technologies has rapidly improved, becoming a relatively cheap renewable resource central to many energy transition and climate change mitigation strategies. Nonetheless, despite widespread public support for wind energy, low success rates in planning applications are threatening the expansion of wind energy production. This is because wind farm developments often face strong social opposition. While this puzzle has been studied in high-income settings, there is little work that adequately explains the sources of this resistance in developing countries. This research thus explores the factors affecting whether and how communities accept wind energy developments in Southern Mexico. It does so by drawing together three lenses: the energy justice framework, the capability approach, and power analysis. These lenses help examine how shifting actions and power relations maintain or transform conflicts around RETs in developing contexts. This study looks at the case of wind energy siting in three Indigenous communities neighbouring wind installations in the Isthmus of Tehuantepec, Oaxaca state, Mexico. The three localities offer a valuable comparison, as they have low, medium, and high levels of wind farm community acceptance. The study draws on fieldwork conducted between September 2017 and June 2018, primarily consisting of 103 semi-structured interviews and a medium-size questionnaire-based survey (N= 382). Operationalising the capabilities approach, this data helps understand local perceptions and concerns regarding capabilities and wellbeing, and links this to three elements of energy justice: distribution, recognition and procedures. Two intertwined findings emerge from this research in the case of wind energy in Southern Mexico, with broader significance. Firstly, just energy transitions require recognition of locally-valued forms of justice. Energy infrastructure siting processes and outcomes must incorporate these understandings of well-being. Secondly, it is essential to understand the power relations in renewable energy processes. Community acceptance entails bundles of changing actions and positions, shaped by internal power relations and those between communities, the state, and wind energy developers. Importantly, these power dynamics can create barriers to expanding valuable capabilities for some stakeholders, thereby reducing the social acceptance of wind energy, and diminishing the possibility for a just energy transition. Together, these findings contribute to Latin American case studies on social dynamics around wind energy, which has significant implications for theorising about energy justice and sustainability pathways from the Global South. This study offers a bottom-up perspective to just transitions by emphasising the capabilities that local people have reason to value, in the context of power dynamics between developers, governments and communities.

This study provides an account of the agendas, participation and influence of management, teachers, students and parents at a primary school in Tigray, Ethiopia. A literature review revealed gaps in the knowledge of these stakeholders’ involvement in school leadership structures in the current national policy context. A broader review of the major traditions of school research informed the design of this ethnographic case study. Fieldwork at ‘Ketema School’ took place over an eight-month period in 2014, and involved participant observation, informant-led interviews, and the collection of institutional documents. Data collection focused on the meetings of various bodies, and the activities of a single class in Grade 6 and 7. Inductive analysis of the case data was supported by Atlas.ti. The study reveals a convergence of understandings about the purposes and processes of the school consistent with the state-authorised model of schooling for national development. Structures and processes of surveillance and control incentivise and normalise compliance with government directives. These include positions of distributed leadership and mechanisms of mutual surveillance and internal accountability through which teachers and students share responsibility for supervising peers and colleagues. For example, the student leaders of the ‘one-to-five’ networks perform an academic support and behavioural control function in relation to their peers; and gim gima is a practice of public critique used for exposing misconduct. Meetings and other participative spaces enable members of the school community to share their views on conditions in school according to their interests and priorities; however, these forums are dominated by management agendas, and school-level decisions are restricted by a strong external policy context. This study extends knowledge of school leadership practices in Ethiopia and informs wider debates around community participation, accountability and school autonomy in developing countries. Recommendations are made for sharing and strengthening democratic practices and for future research.

Bluetooth Low Energy (BLE) has emerged as one of the main wireless technologies used in low-power electronics, such as wearables, beacons, and devices for the Internet of Things (IoT). BLE’s energy efficiency characteristics and ease of use interface are essential features for the design of ultralow-power devices. The integration of BLE with various sensors is certainly the main aspect for the development of efficient solutions, and monitoring the power consumption of the device in all the cycles of operation is important for decisions such as advertising interval, sensor routine and data transmission.Recent work of BLE power analysis focuses on the theoretical aspects of the advertising and scanning cycles, with most results being presented in the forms of mathematical models and computer software simulations. Such models and simulations are particularly important for the understanding of the technology. However, many times they leave real applications out of scope.This thesis covers the implementation of a multi-sensor Bluetooth Low Energy system, and the study of the communication protocol regarding its power consumption and RF performance. The implementation consists of a battery powered custom Printed Circuit Board (PCB) featuring the Texas Instruments CC1352P7 as the main SoC and three different sensors capable of measuring 6 different parameters. The sensors are the BME688 from Bosch, and the ADXL343 and AD5941 from Analog Devices. The characterization process was performed using Keysight EXR Oscilloscope and 208A Spectrum Analyzer.The proposed design has dimensions of 28 mm × 35 mm. The current consumption of the implemented design with one sensor in operation is 4.1mA and 7.1mA, and 6.9 mA combined.

Background: The complexity and variety of inhaled tobacco products have increased significantly with the introduction of Electronic Nicotine Delivery Systems (ENDS). Product characteristics and human behavior are the main factors affecting products' emissions, consumption, and health effects. Combining the understanding of these two factors informs the next generation of regulations of these products. Quantifying use patterns helps health care professionals provide informed treatment of users addicted to nicotine and tobacco products. The literature lacks comprehensive studies to characterize ENDS hardware and operation, such as heating element resistance, battery performance, power, and thermal management. Human behavior is measured as use topography, a temporal signal of the interaction between the user's mouth and the product. Traditional analysis of use topography focused on short-term puff dynamics, and overlooks extended session dynamics. The literature lacks the appropriate parameters and tools to quantify session dynamics over a long duration (i.e. days, weeks, months). Method and validation: This dissertation focuses on two aims: The first aim is to design methods to characterize the hardware and operation of modern ENDS devices and test their effects on ENDS performance. Parts of ENDS devices are repurposed to produce product-specific testing apparatus of effective coil resistance for several ENDS products. The effects of manufacturing variation in coil resistance on coil lifetime and Total Particulate Matter (TPM) are measured for one of the most popular pod-style ENDS. A method for dynamic measurement of electrical signals in modern ENDS is presented and validated for ENDS power management characterization. The second aim is to design and validate algorithms for quantifying chronic use topographies associated with inhaled tobacco products. The quantifying tool is designed based on autocorrelation to quantify chronic topography parameters including session period, and session duration as random distributions. These distributions provide insights into session topography dynamics over a day, a week, a month, and longer. Results: Significant variations in coil resistance were observed within and between brands such as the difference of 0.593 [Ω] (p < 0.001) between JUUL and Vuse ALTO. The mean resistance and standard deviation of the coil assemblies was shown to be μ= 1.031 (σ= 0.067) [Ω] for Vuse ALTO and μ= 1.624 (σ= 0.033)[Ω] for JUUL. The variation in coil resistance between products and within products can have significant impacts on aerosol emissions. Dynamic measurement of power in Vuse ALTO showed that voltage is delivered to the coil as pulses of ~119.5 Hz with duty cycles of 0.3 - 0.6 which found to be correlated with the energy and change in temperature in coil. The mass of the generated aerosol per puff was correlated with the energy per puff. Quantifying session topography method successfully worked with puff period, session period and session duration with mean absolute percentage error of 1.18 [%], 2.92 [%], 12.59 [%] respectively. The method showed resiliency to session dynamic variation with accepted percentage of 99.98 [%], 92.39 [%], and 76.21 [%] for puff period, session period and session duration respectively. The method appears sufficiently valid and robust for analysis of natural environment human subject behavioral studies of tobacco product use. Conclusions: The absence of e-liquid in the pod is an important factor in causing coil failure. Small bits of the degraded coil could be potentially introduced to the aerosol. Energy is an important, if not the most important, contributor to the yield generated form an ENDS. A method was demonstrated to dynamically measure coil temperature based on temperature coefficient of resistance (TCR). It was also shown that coil temperature can be controlled by the ENDS by changing the energy delivered to the coil per pulse which is intern controlled by duty cycle of the pulses using the PWM algorithm. The quantifying session topography method provides high value in investigating the effects of user environment such as a day of week on session dynamics. It also demonstrated an example of using the method for quantifying how product characteristics such as e-liquid flavors and nicotine concentration may moderate use behavior.

Well-conducted field experiments, broadly construed to contain both randomized controlled trials and quasi-experiments, involve extensive planning with substantive deliberation. Such deliberation has the potential to fuel and strengthen the analysis stage of the study. Each field experiment is unique in its own manner, from the subgroups on which effects are expected to concentrate to the design of the study itself. Reliance on off-the-shelf methods to analyze field experiments may exclude this potentially valuable information that, if handled properly, would provide a greater opportunity to detect an effect. In this dissertation, we propose two novel methods that look to extract information unique to a specific study and translate it into additional power. We demonstrate these methods on a large-scale education intervention aimed at correcting the stalled reading trajectories of early elementary students. The first method, Power-maximizing Weighting for Repeated-measurements and Delayed-effects (i.e. PWRD) aggregation, converts the theory of change behind a class of education interventions into a test statistic that maximizes the asymptotic relative efficiency over standard methods, thus providing greater power. The scheme emphasizes cohorts and years-of-follow-up on which effects are expected to accrue with appropriate attention paid to the relative precision of estimates within cohorts. We find through a simulation study mirroring the design of the reading intervention that this method provides stark gains in power over methods common to education research. While PWRD aggregation increases power, confidence interval estimation is more difficult. To alleviate this problem, we partition our parameter space into three regions: equivalence, superiority, and inferiority. In the first, we employ PWRD aggregation to provide the greatest opportunity to detect an effect. In the latter two regions, we employ a method common to the domain in which the research occurs such that when we are able to detect an effect, interpretation of the point estimate and confidence interval proceeds in a standard fashion. The second method we propose is a dry run simulation scheme that creates a pseudo-experiment replicating the initial randomized trial in a manner that preserves blinding to impact estimates. This procedure, which uses real rather than synthetic data, provides a sandbox in which various models may be tested to discover the model specification that most precisely estimates an artificially imposed treatment effect. The dry run method, similar in motivation to cross-validation and uniformity trials, allows the statistician advising field experiments to estimate expected losses for each of a variety of methods, enabling them to elect a novel or unfamiliar method if it demonstrably outperforms methods more familiar to the broader team. When applied to the reading intervention that motivated dry runs, results from this method challenged preconceived notions about covariate choice, suggesting we control for covariates beyond pre-test scores.

Trees outside forests (TOF) are increasingly recognized as an important resource for carbon storage. TOF include trees used in agricultural windbreaks, urban ornamentals, and trees in non-forest ecosystems like prairie. This study used tree data from the US Forest Service’s Forest Inventory and Analysis (FIA) database to detect variation in forest carbon stocks and tree canopy cover between trees in forest ecosystems and trees in prairie ecosystems. A longitudinal transect was established, extending from Bismarck, ND to Duluth, MN to capture a gradient of mixed temperate forest in the east, prairie in the west, and a transitional zone in the middle. A number of potential transect sizes were evaluated using a series of power analyses (alpha = 0.05, power = 0.88, 0.83) and these determined that a transect radius of 50 km (total transect height 100 km) and transect length of 660 km was sufficient to capture the gradient and provide statistically significant results if differences existed. All FIA plots within the transect which had tree measurements taken from 2012 – 2018 (n = 4,155) were then used in a series of random forest analyses. The response variables of interest were carbon content (in megagrams per hectare) and percent live canopy cover, both at the plot level. Twenty-seven predictor variables were assessed: a few plot condition indicators from the FIA data, but mostly climate variables (30-year climate normals). Six random forest analyses were run: three examining canopy cover as the response variable using all plots, forest plots, and non-forest plots, and three examining carbon content within the same groupings. The power analysis lent confidence to the establishment of an effective study area and transect size; however, the random forest analyses were ultimately unable to consistently predict tree canopy cover or carbon content at the plot level. Although the random forest analyses did not provide statistically significant evidence for variation at the plot level, the patterns they revealed between which climate predictors performed best under forested and non-forested conditions are intriguing and may invite further investigation.

From intelligence in romantic partners to empathy in political leaders, people can readily think about what they like in other people. These abstract ideas about liking can be useful in many situations, but do they always align with people’s concrete experiences of liking? In this dissertation, I distinguish between ideas and experiences in interpersonal evaluation, and I argue that this distinction is useful because they can predict different kinds of decisions and help explain social psychological phenomena that appear paradoxical. In Chapter 1, I situate ideas versus experiences in a broader theoretical framework on attributes, and I conducted a series of studies showing that (1) ideas about liking for personal attributes can be affected by incidental features of the context, and that (2) ideas about liking versus experiences of liking predict different outcomes in the context of romantic attraction. In Chapter 2, I zero in on empathy and consider why people like the idea of empathy but not always those who show it. Across seven experiments, I found evidence that empathy has evaluative consequences for the empathizer beyond the empathic dyad. Findings from these experiments suggest that although people are often encouraged to empathize with disliked others, they are not always favored for doing so. In Chapter 3, I present a discussion and tutorial on a statistical technique central to my empirical work on ideas versus experiences, structural equation modeling (SEM). Specifically, I conducted a simulation study showing key factors that influence statistical power to detect true effects in SEM, and I introduce a free online app that helps researchers conduct power analysis for SEM. Taken together, these three chapters offer theoretical, empirical, and methodological advances for the study of ideas versus experiences in interpersonal evaluations. Unpacking the distinct roles that ideas versus experiences play can help us understand the apparent disconnect between what people think they like and what drives their liking across many domains.

With increased growth of the Internet of Things (IoT) and physical exposure of devices to adversaries, a class of physical attacks called side-channel analysis (SCA) has emerged which compromises the security of systems. While security claims of cryptographic algorithms are based on the complexity of classical cryptanalysis attacks, they exclude information leakage by implementations on hardware platforms. Recent standardization processes require assessment of hardware security against SCA. In this dissertation, we study SCA based on deep learning techniques (DL-SCA) as a universal analysis toolbox for assessing the leakage of secret information by hardware implementations. We demonstrate that DL-SCA techniques provide a trade-off between the amount of prior knowledge of a hardware implementation and the amount of measurements required to identify the secret key. A DL-SCA based on supervised learning requires a training set, including information about the details of the hardware implementation, for a successful attack. Supervised learning has been widely used in power analysis (PA) to recover the secret key with a limited size of measurements. We demonstrate a similar trend in fault injection analysis (FIA) by introducing fault intensity map analysis with a neural network key distinguisher (FIMA-NN). We use dynamic timing simulations on an ASIC implementation of AES to develop a statistical model for biased fault injection. We employ the model to train a convolutional neural network (CNN) key distinguisher that achieves a superior efficiency, nearly $10times$, compared to classical FIA techniques. When a priori knowledge of the details of hardware implementations is limited, we propose DL-SCA techniques based on unsupervised learning, called SCAUL, to extract the secret information from measurements without requiring a training set. We further demonstrate the application of reinforcement learning by introducing the SCARL attack, to estimate a proper model for the leakage of secret data in a self-supervised approach. We demonstrate the success of SCAUL and SCARL attacks using power measurements from FPGA implementations of the AES and Ascon authenticated ciphers, respectively, to recover entire 128-bit secret keys without using any prior knowledge or training data.

High-dimensional models that incorporate sparsity assumptions arise naturally in numerous modern applications ranging from cybersecurity to statistical genetics. In such data models, statisticians are typically tasked with (a) detecting the presence of sparse signals, and (b) estimating the support -- i.e., locations of the non-zero components -- of the signals. In the past two decades, researchers have studied the theoretical limits for the two statistical problems, and discovered interesting phase transition phenomena where the problems are found to be solvable if and only if the signals have certain configurations in terms of their sparsity and signal size. This dissertation contributes a spectrum of new phase-transition results under various statistical risks, providing a largely complete picture of the fundamental limits in the sparse signal detection and support recovery problems. A major contribution of this thesis is a complete solution to the exact signal support recovery problem for the general class of thresholding estimators. Specifically, it is shown that if the signal magnitude is larger than a certain boundary as a function of the signal sparsity, then thresholding estimators can recover the signal support exactly as the signal dimension grows to infinity. Conversely, for signal magnitudes below that same boundary, no thresholding estimators can recover the support asymptotically. This phase-transition result is surprisingly universal and it holds under very general error-dependence models. Threshold-based support estimators were also shown to be finite-sample Bayes-optimal for errors with log-concave densities and sub-optimal for heavier tailed errors. As a result, complete minimax characterizations of the statistical difficulty in support recovery problems are obtained. The study of exact support recovery under broad error-dependence conditions is a first of its kind. It is made possible by developing new probabilistic results on a concentration of maxima phenomenon known as relative stability. A complete characterization of the relative stability phenomenon for Gaussian error arrays in terms of their correlation structure is established. Finally, the manifestation of the phase transition phenomena in the field of statistical genetics is studied. The established results provide a first theoretical explanation of the long empirically observed trade-off between allele frequency and penetrance, and quantifies the fundamental statistical limits of discoverability in genome-wide association studies. Moreover, one surprising result of practical significance is that balanced designs in association studies rarely deliver optimal power. Developed are explicit formulas for power calculations, as well as an interactive web-based software application for performing power analysis and finding optimal study designs.

The Advanced Encryption Standard (AES) is the current standard for symmetric key ciphers and is algorithmically secure. Side channel attacks that target power consumption can reveal the secret key in AES implementations. Masking data with random variables is one of the main methods used to thwart power analysis attacks. Data can be masked with multiple random variables to prevent higher-order attacks at the cost of a large increase in area. This thesis tests the plausibility of using varied finite field construction to prevent power analysis attacks as an alternative to masking. Initially, a design using finite field architecture as the sole countermeasure was investigated. This was followed by varied field construction in conjunction with a low entropy masking scheme. Neither approach provided an acceptable trade off between security and area. Analysis then turned to a combined Boolean and multiplicative masking scheme. Varied construction provided little gain for multiplicative masking. However, varied constructions were found to greatly increase security when used in conjunction with a Boolean random mask. A novel masking scheme for AES resistant to second-order attacks is proposed. Instead of an additional mask, variation in finite field construction is exploited to increase resistance to second-order attacks in Boolean masked shares. As a result, the area requirement is substantially reduced. For an example AES encryption, the proposed design is 12% smaller compared to the previous best design, with a small drop in achievable security level.

From a digital world where we make phones and twitter talk to each other, we are moving towards building a network where everyday objects communicate and share data among themselves. This platform known as Internet of Things (IoT) is a fast growing technology today where devices are connected via internet. In the next few years, the studies anticipate that billions of devices will be connected through internet and we will be heavily dependent on IoTs which will serve as the backbone of almost every sector[31]. Producing secure connections and secure devices is the need of the hour, and most of the devices as part of this connection are small-sized devices with low computational powers, while carrying private and sensitive information. Implementing low-level techniques to provide security such as WDDL[58], SDMLp[44] can be very expensive and affect its range of usage. The key objective of the thesis is to propose a Secure design methodology for such small devices which are used in today's high connectivity networks.This dissertation focuses on producing secure designs for small-to-medium sized Finite State Machines and the scope of the work is within the analysis of Power based side-channel attacks. The proposed algorithm for FSM state assignment, removes the reliance of its power consumption from the data carried within the FSM. The designer will be able to design the FSM with required security level at the cost of area. The thesis further discusses generation of state encodings for Low-power design without compromising on security. All the experiments have been performed on BenGen [27] and MCNC benchmarks [63] and the merit of the proposed algorithm has been proved.

Power dissipation has now become the most critical design constraint. Up till now, in the design flow of any SoC, power estimation and analysis came into the picture only after the completion of RTL synthesis. However, design optimization for low power is most suitable before synthesis. Each decrease in process geometry makes dynamic power targets harder to achieve. Also, changes made later in the design for power optimization lead to costly re-spin. It is better to pin-point power related problems in the design as early as possible when they can still be fixed. It also reduces risk by ensuring that the design meets power goals before embarking on its implementation. A novel approach is presented in this thesis which introduces power analysis at the RTL stage itself using PSL assertions. This will enable the SoC designer to optimize the design from a low power perspective at a very early stage (RTL) in the design flow where the scope of modification is maximized and the cost minimized.

In experimental research, planning studies that have sufficient probability of detecting important effects is critical. Carrying out an experiment with an inadequate sample size may result in the inability to observe the effect of interest, wasting the resources spent on an experiment. Collecting more participants than is necessary may unnecessarily put too many participants at risk and potentially detect an effect size that is not clinically meaningful. Therefore, obtaining the most accurate estimates of necessary sample size prior to applying for research funding and carrying out experiments is of utmost importance. Educational studies often select whole classrooms for participation. When clusters of individuals are assigned to experimental conditions, the design is referred to as a cluster randomized clinical trial. An over-estimation or under-estimation of the number of clusters needed can have large budgetary or logistical consequences. Accurate sample size estimates, especially in large-scale studies, can help researchers carry out high-quality research and make valid conclusions. Statistical power is the probability of rejecting a false null hypothesis so that a researcher can correctly conclude that an effect has been observed in a study. Three different methods of estimating power are examined in this study, including (a) formulaic power functions, (b) empirical simulation studies, and (c) constructing exemplary datasets. Formula-based methods make assumptions that may not be met in practice. This study assessed (a) the extent to which failure to account for practical data conditions such as attrition, unequal treatment group sizes, and unequal cluster sizes bias estimates of anticipated power and sample size; and (b) if these effects were moderated by the amount of variability that is attributable to between-cluster differences. The empirical simulation study and exemplary dataset methods showed that attrition and unequal treatment group sizes have substantial effects on estimates of power and sample size. Unequal cluster sizes did not appreciably affect power. Higher levels of bias were found when less variability was attributable to between-cluster differences. Power analyses based on a formulaic approach that fail to account for attrition or treatment group imbalance may severely overestimate power and underestimate the sample size necessary to observe important effects. Adviser: James A. Bovaird

Annual commercial fur harvest data suggests a decline in South Carolina's mink (Mustela vison) populations over the past century (Butfiloski and Baker 2005). In 1990, DNR began conducting assessments of South Carolina's coastal marshes, revealing mink populations to be scarce to absent in the northern coastal regions (Baker 1999). Since 1999, restoration efforts have re-established mink populations in the Cape Romain National Wildlife Refuge and additional areas in the northern coastal marshes of South Carolina. Because mink are a species of high conservation priority throughout the coastal zone of South Carolina, it is important to develop and implement a reliable population monitoring program that will direct future management and aid in the recovery of coastal mink populations (Kohlsaat et al. 2005). Spotlight survey data were used to evaluate the relationship of various environmental variables and survey counts during each year of the study (2008 and 2009). Results indicated that actual tide height (P< 0.001) and creek size (P< 0.001) were correlated to mink observations during spotlight counts during 2008 (r2= 0.291), while actual tide height (P= 0.026), creek size (P= 0.039), and Julian date (P< 0.001) were significant variables during 2009 (r2= 0.225). Actual tide height, predicted tide height, and creek size were also used with 2008 spotlight counts to predict survey counts during 2009. Results showed that predicted mink counts, based on actual tide heights (r2= 0.133, P< .001) and predicted tide heights (r2= 0.11, P< 0.001), were correlated to observed counts for 2009 spotlight surveys. A method was also developed to determine optimal survey tide height. Results from the analysis indicated that at a tide height of at least 1.85 meters (6.05 feet) above mean lowest low water level (MLLW), the probability of observing mink was significantly greater (P= 0.035) than at a lower tide height. Survey data were also used in power analyses to determine the probability of spotlight surveys to detect changes in mink counts over multiple survey seasons. Results indicated that an increase in annual spotlight survey effort, length of monitoring effort (5 or 10 years), a less restrictive probability of making a type I error (i.e. alpha level), and a decrease in the standard deviation of survey counts improves the probability of detecting annual changes (increasing and decreasing) in mink survey counts.

This study examines the social actors and issues involved in constructing and contesting the National School Lunch Program (NSLP), in order to identify whose interests are involved in shaping an institution which transmits dietary habits and food knowledge to the nation’s children through the mid day meal. For the historical analysis, I collected data from historical accounts of the NSLP, congressional hearings, laws, and newspaper articles. For the contemporary analysis, I interviewed 15 actors representing organizations key to federal NLSP policy making. To frame my analysis, I utilize a model of power, based on the work of Arts and Van Tatenhove (2004), and the work of Burstein (1991), who describes issue creation and movement in policy domains. The key findings of this study are that actors with the most financial resources (e.g. the food industry) do not automatically achieve their interests in the policy making process. In fact, at key times of contestation, economically powerful actors form alliances and adjust their agenda in reaction to the use of other forms of power by economically weaker actors. This information can help economically weaker actors (e.g. the farm to school movement) understand how to increase their influence in the policy domain.

The rapid increase in the use of embedded systems for performing secure transactions, has proportionally increased the security threats which are faced by such devices. Side channel attack, a sophisticated security threat to embedded devices like smartcards, mobile phones and PDAs, exploits the external manifestations like processing time, power consumption and electromagnetic emission to identify the internal computations. Power analysis attack, introduced by Kocher in 1998, is used by adversaries to eavesdrop on confidential data while the device is executing a secure transaction. The adversary observes the power trace dissipated/consumed by the chip during the encryption/decryption of the AES cryptographic program and predicts the secret key used for encryption by extracting necessary information from the power trace. Countermeasures proposed to overcome power analysis are data masking, table masking, current flattening, circuitry level solutions, dummy instruction insertions, balancing bit-flips, etc. All these techniques are either susceptible to multi-order side channel attacks, not sufficiently generic to cover all encryption algorithms, or burden the system with high area cost, run-time or energy consumption. The initial solution presented in this thesis is a HW/SW based randomised instruction injection technique, which infuses random instructions at random places during the execution of an application. Such randomisation obfuscates the secure information from the power profile, not allowing the adversary to extract the critical power segments for analysis. Further, the author devised a systematic method to measure the security level of a power sequence and used it to measure the number of random instructions needed, to suitably confuse the adversary. The proposed processor model costs 1.9% in additional area for a simplescalar processor, and costs on average 29.8% in runtime and 27.1% in additional energy consumption for six industry standard cryptographic algorithms. This design is extended to a processor architecture which automatically detects the execution of the most common encryption algorithms, starts to scramble the power waveform by adding randomly placed instructions with random register accesses, and stops injecting instructions when it is safe to do so. This approach has less overheads compared to previous solutions and avoids software instrumentation, allowing programmers with no special knowledge to use the system. The extended processor model costs an additional area of 1.2%, and an average of 25% in runtime and 28.5% in energy overheads for industry standard cryptographic algorithms. Due to the possibility of removing random injections using large number of samples (due to the random nature, a large number of samples will eliminate noise), the author proposes a multiprocessor 'algorithmic' balancing technique. This technique uses a dual processor architecture where two processors execute the same program in parallel, but with complementary intermediate data, thus balancing the bitflips. The second processor works in conjunction with the first processor for balancing only when encryption is performed, and both processors carry out independent tasks when no encryption is being performed. Both DES and AES cryptographic programs are investigated for balancing and the author shows that this technique is economical, while completely preventing power analysis attacks. The signature detection unit to capture encryption is also utilised, which is used in the instruction injection approach. This multiprocessor balancing approach reduces performance by 0.42% and 0.94% for AES and DES respectively. The hardware increase is 2X only when balancing is performed. Further, several future extensions for the balancing approach are proposed, by introducing random swapping of encryption iterations between cores. FPGA implementations of these processor designs are briefly described at the end of this thesis.

Floating Point Arithmetic is extensively used in the field of medical imaging, biometrics, motion capture and audio applications, including broadcast, conferencing, musical instruments and professional audio. Many of these applications need to solve sparse linear systems that use fair amounts of matrix multiplication. The objective of this thesis is to implement double precision floating point cores for addition and multiplication .These cores are targeted for Field Programmable Gate Arrays because FPGAs give the designer good control over the number of I/O pins and utilization of on chip memory. FPGAs are also comparable to floating point processors in their power consumption. The multiplier and adder cores conform to the IEEE 754 standard for double precision. The design is implemented on Xilinx ISE 8.2i and has been simulated on ModelSim 6.1i.The thesis pays significant attention to the analysis of the adder and multiplier cores in terms of pipelining and area so as to maximize throughput in any manner possible. It further throws light on variations of power with pipelining. Power measurements are done using XPower provided by ISE.

In longitudinal studies, both repeated measures (RM) and hierarchical linear model (HLM) can be applied. Yet, it is not clearly determined which of RM and HLM should be applied in the balanced design of a longitudinal study. The purpose of this study was to explore the interrelation of HLM and RM in their theoretical development and to compare the two approaches to the evaluation of fixed effects in the balanced design of longitudinal data analysis. This was done through a Monte Carlo (MC) study of the empirical power of HLM and RM in three fixed effect tests: Two-group treatment effect, time effect and treatment-by-time interaction. In this research, RM included traditional RM (TRM) and updated multivariate RM (UMRM). Two covariance structures examined under UMRM were UN and AR(1). Specifically, this paper compared the power of HLM, AR(1), UN and TRM in the three fixed effect tests within three factors: Effect sizes, sample sizes and G matrices. The results indicated that HLM, AR(1) and UN had similar power patterns but different from TRM. TRM was significantly more powerful than the other three in the treatment and time tests, but significantly less powerful than the three in the interaction test. Among HLM, AR(1) and UN, UN power seemed to rank the highest, AR(1) the second, and HLM the lowest, in all three tests under defined conditions. Nevertheless, the pairwise power differences of HLM, AR(1) and UN are not all significant. Additionally, findings from Type I error rates, bootstrap bias, standard errors, confidence intervals and model fit statistics (AIC and BIC) were examined for the four models. Limitations, conclusions, recommendations for future study were also provided.

Hardware devices implementing cryptographic algorithms are finding their way into many applications. As this happens, the ability to keep the data being processed or stored on the device secure grows more important. Power analysis attacks involve cryptographic hardware leaking information during encryption because power consumption is correlated to the key used for encryption. Power analysis attacks have proven successful against public and private key cryptosystems in a variety of form factors. The majority of the countermeasures that have been proposed for this attack are intended for software implementations on a microcontroller. This project focuses on the development of a VHDL tool for investigating power analysis attacks on FPGAs and exploring countermeasures that might be used. The tool developed here counted the transitions of CLB output signals to estimate power and was used to explore the impact of possible gate-level countermeasures to differential power analysis. Using this tool, it was found that only a few nodes in the circuit have a high correlation to bits of the key. This means that modifying only a small portion of the circuit could dramatically increase the difficulty of mounting a differential power analysis attack on the hardware. Further investigation of the correlation between CLB outputs and the key showed that a tradeoff exists between the amount of space required for decorrelation versus the amount of decorrelation that is desired, allowing a designer to determine the amount of correlation that can be removed for available space. Filtering of glitches on CLB output signals slightly reduced the amount of correlation each CLB had. Finally, a decorrelation circuit was proposed and shown capable of decorrelating flip-flop outputs of a CLB, which account for less than 10% of the CLB outputs signals.