Reed Digital Collections (Reed College)
Not a member yet
8045 research outputs found
Sort by
Thinking With Fire: Burning Landscapes of Knowledge and Power
No full texthttps://rdc.reed.edu/v1/resources/cdba3269-db8e-4e7d-a91f-724def655b26/thumb/128.jpgManaging, understanding, using, and living with the forest fire regimes of the U.S. West Coast has been an important concern of both scientific study and Indigenous ecological knowledge. Implicated in ways of studying and knowing fire are epistemological and ideological aspects of different knowledge systems. The scientific management of fire, entangled with settler-colonial ideological legacies, has created the profoundly harmful state of fire exclusion in the ecosystems of the West Coast. This thesis explores knowledge as it relates to fire as an agentive and transformative force. Thinking with fire allows for the imagining of transformed landscapes of knowledge and power
Lone Pine Talk: TRAP-backing and Connection to Place in Rural California
No full texthttps://rdc.reed.edu/v1/resources/6a2f78b7-f177-4bf4-a497-44b12082885e/thumb/128.jpgStudies of California English often focus on speakers in urban, coastal regions of the state, leaving out the rural speakers. This thesis explores the retraction of the TRAP vowel—a salient feature of California English—among locals in Lone Pine, California, a rural town in the Owens Valley at the base of the Eastern Sierras. I interview eight individuals on site in Lone Pine and analyze the patterning of their TRAP vowel position (backer or fronter) with their urban/rural orientation, level of rootedness, and age. My findings support my hypothesis that urban-orientation, lower rootedness, and younger age predict a backer TRAP; however, individual variation turns out to be primarily responsible for the variation in the production of TRAP. This can be explained by style, identity, and personae, viewing the speakers’ linguistic choices as reflective of their own, unique connection to place
Supervised Machine Learning to Predict Student Proficiency in Oregon Public Schools
No full texthttps://rdc.reed.edu/v1/resources/4ea38e30-3bb5-4ea8-aa59-c47305d41713/thumb/128.jpgThe Every Student Succeeds Act (ESSA) requires all states to annually measure the academic performance of public schools in the United States, including Oregon. One critical metric of school performance is the ”percentage proficient” score, which measures the percentage of students who achieve proficiency in English Language arts (ELA) and Math standardized tests for the 2018-19 year. This thesis compares the use of random forest models and decision trees to predict student proficiency for Oregon public schools using various student and school information. These variables include gender, ethnicity, faculty resources, and others. By comparing random forests and decision trees we aim to provide models that are both interpretable and accurate
How the “Trans Panic Defense” Shapes Attitudes toward Anti-Transgender Violence
No full texthttps://rdc.reed.edu/v1/resources/6e9880a0-36ec-4bc4-9244-bdcd027426a7/thumb/128.jpgThus far in 2023, almost 500 anti-LGBTQ+ bills have been introduced in the U.S., 150 specifically targeting the transgender community. The trans panic defense—a legal strategy in which the assailant of a violent crime against a transgender individual is framed as the victim of the assaultee’s “deception” regarding their gender identity—capitalizes on anti-LGBTQ+ bias. The Human Rights Campaign recommended legislative action against the use of this defense, but it remains legally viable in 35 states and five U.S. territories. Building on previous research on victim framing, I investigated the efficacy of the trans panic defense and its underlying psychological mechanisms. In a preregistered experiment, online U.S. adult participants (N = 605) read a fictionalized news report about a physical assault that framed the female assaultee as the victim (of violence), the male assailant as the victim (of deception), or neither individual as a victim (baseline). In the report, the assaultee was described as transgender or was presumptively cisgender. Participants expressed more support for the assailant and recommended a lighter sentence when he was framed as the victim of deception, relative to baseline, and when the assaultee was framed as the victim of violence. Participants also expressed more support for the assailant and recommended a lighter sentence when the assaultee was transgender than when she was cisgender. Importantly, these two effects combined to yield the most favorable impressions of the assailant when he was framed as the victim of deception by a transgender assaultee (i.e., the trans panic defense). These findings provide evidence for the generalizability of victim framing effects and have implications for mitigating increasing rates of violence fueled by transphobia
Towards Analysis of Parallel Caching with Variable Size Items
No full texthttps://rdc.reed.edu/v1/resources/9600c619-b742-481c-b40b-f7acfd560272/thumb/128.jpgIn this paper we present the problem of parallel caching and explore the challenges with generalizing existing results to include items with variable sizes. We do this by building on prior work in that area and the related area of green paging. Starting with the basic model of sequential caching, we build up to parallel caching, where p processors are simultaneously using the cache, and green paging, where we want to minimize the total cache usage while maintaining a minimum allocation. We review existing literature on these models and put together a description of each result, along with a discussion of whether it generalizes to variable sizes.For results that do not generalize to variable size items, we demonstrate why and propose ways of accommodating the now missing proofs in the new model. We also show which steps do generalize, making some small progress towards a full generalization. One of the features that we can generalize is the close correspondence between green caching and parallel caching. We can also generalize one step of the proof that a green paging algorithm BLIND has an optimal competitive ratio, and can list a smaller number of specific conditions that need to be proved in order to complete the generalization.By collecting these conditions, we leave a subset of the original results that need to be proved in the size model in future work. These goals can be the main focus of future work in this area, and we end by detailing our attempts at proving those remaining results
Shakers and Movers: A Case Study in Earthquakes, Communities, Politics, and Redevelopment
No full texthttps://rdc.reed.edu/v1/resources/4375d861-70d9-408d-824b-bffccd4a4eae/thumb/128.jpgIn this thesis, I explored the nature of recovery in the housing market of Christchurch, NZ following the Canterbury Earthquake Series. I sought to understand the complexities between the society, economy, and policy that resulted in the housing crisis in Christchurch today. Through this exploration, it is clear that the overwhelming need to focus on redevelopment and reconstruction of the city enter, as well as the need to move on from the disaster, resulted in a large portion of the city remaining, to this day, unestablished. The possibility that the city employs a ‘Building Back Better’ strategy, wherein a city makes a disproportionate jump forward, was limited due to the constraints on housing supply, preparedness, and market forces. I utilize the understandings gained in the situation in Christchurch to understand the preparedness of Portland, Oregon, in anticipation of ‘The Big One’—a subduction zone earthquake capable of reaching up to a 9.0 magnitude
Three Metaphysical Fables
No full texthttps://rdc.reed.edu/v1/resources/a2395bd4-42f8-47a3-b222-eb856a3f9925/thumb/128.jp
Multithreaded Prefix-Free Parsing for Multi-Genome Index Construction
No full texthttps://rdc.reed.edu/v1/resources/6eb5a2cf-5f8a-482c-8f3b-875f154ed237/thumb/128.jpgPrefix-free parsing (PFP) is a compression-aware tool developed by Boucher et al. as a method for computing the Burrows-Wheeler transform (BWT) on large input texts [1]. PFP is particularly designed for when the input text size takes up a significant portion of the computer’s memory or exceeds it. This is of use in bioinformatics, for example, when analyzing several genomes at once.This thesis describes the theory behind and importance of the BWT. We then describe PFP. Finally, we contribute two new methods of performing the critical parsing phase of PFP using multiple threads of execution. We give experiments that demonstrate the improvement our methods make on the performance of “bigbwt”, the code distributed with the original PFP paper [2]
Quantifying the Association between Open Spaces and Neighborhood Gentrification in Portland, Oregon
No full texthttps://rdc.reed.edu/v1/resources/f213e35e-63e3-4bb5-ab21-b623c581a4fe/thumb/128.jpgThis thesis studies the issue of 'green gentrification' by quantifying the association between open spaces such as parks, natural areas, and golf courses, and neighborhood gentrification in the city of Portland, OR. As the multifaceted process of gentrification is largely mediated through the channel of increasing property values, gentrification was defined, for the purposes of this analysis, as the price premium homeowners are willing to pay to live in a particular neighborhood. I adopted a twostage methodology. In the first stage, I estimated 30 hedonic models, one for each year in the study period (1990 – 2019), on individual home sale data in order to create a panel dataset of gentrification metrics. In the second stage, I estimated four models with the measure of gentrification calculated in the first stage as the dependent variable. Spatial data on open spaces, compiled by Oregon Metro, was used to construct open space variables used as independent variables in the second stage. The results from the second stage analysis suggest that, overall, a medium amount of open space in a neighborhood is associated with neighborhood gentrification, with parks contributing to gentrification and natural areas to reverse gentrification
Between Resurrections
No full texthttps://rdc.reed.edu/v1/resources/6f188504-aa2b-49c0-84fd-7f5aef9c4dbc/thumb/128.jpgIn this work, I catalog my art practice and corresponding research centered around three tenets: magic lanterns, interference, and grief.My research of magic lanterns is rooted in anticolonial analysis of their operations to transpose and manipulate information across space, specifically in its inception in the Baroque period and the way that it was received by European audiences in the Victorian period. I take a critical view of the Catholic institution that propagated them so extensively, and connect their mode of oppressive media to my own experience with grief.Mechanistically, I deal with grief by repeatedly recreating parts of myself that represent what I’ve lost. That repetition is obsessive, and in these self-similar reiterations, the larger-scale changes and patterns are what is isolated, cherished, and displayed. This process is best expressed with the operation of interference and moiré, which I discuss both scientifically and experientially. I use the technique of moiré extensively in my paintings, treating them as analog generative works. I discuss as a whole my art practice, obsessive nature, religious motivations, and most centrally, my personal process of coping and healing from personal loss through a repetitive and reproductive artistic process