Below are recent Summer Undergraduate Research Program (SURP) projects completed by students studying Public Policy Analysis at Pomona College.
A Semi-simulation using Generalized Estimating Equations to Determine the Number of Visits Needed to Retain Statistical Significance
Jessie Levin ’18; Mentors: Eleanor Pullenayegum (McMaster University) and Brian Feldman (Hospital for Sick Children/University of Toronto)
Background:Paediatric clinical trials are arduous and costly due to the frequency of study visits, as patients may need to miss school. While it would be ideal to remove study visits, researchers risk sacrificing statistical power. Objective:To determine the minimum number of study visits needed to maintain reasonable precision. Methods: Generalized estimating equations were used to build a semi-simulation model where the predictor was the age of first joint bleed and the outcome was joint scores based on the Canadian Haemophilia Prophylaxis Study (CHPS) dataset where patients received quarterly joint health assessments for the first five years and semi-annually thereafter. To identify the minimum number of required visits, we removed half of the visits randomly or in an alternating fashion and compared the results with the intact dataset using robust standard errors (RSE) and estimates. Results: The relationship was preserved in the elbows, right knee and in the left ankle based on the RSE and estimates in the random dataset based on the estimates. Randomly removing half of the visits resulted in a larger increase in RSE for 5/6 of the analyses compared to the analyses where data was removed systematically. Conclusions: This data suggests that with the reduction of number of visits, the estimates decrease the precision and may no longer reflect the full model. Further investigations will see if reasonable precision is preserved when fewer visits are dropped.
Funding Provided By: STEMPREP Project (Southern Methodist University)
Land Ownership and Economic Advancement in Southwestern Provinces of the Dominican Republic
Tara Miller (2014); Mentor(s): Pierre Englebert
Abstract: Because land titling is believed to increase access to loans, which in turn increases investment and thus stimulates growth, many governments and organizations have chosen land titling as a main development priority in a number of Latin American countries, such as Bolivia and Peru. In the Dominican Republic, however, there has been little political and organizational will to start a formal land titling process. It is for this reason that I chose to assess whether formal land titling might support micro-level and macro economic development, focusing on rural communities in the Dominican provinces of Elias Piña, San Juan, Azua, and Barahona. In order to investigate this relationship I distributed a qualitative survey to 44 participants in communities in the regions of Elias Piña, San Juan, Azua, and Barahona. The communities were selected by partner agency contacts at Plan International and varied in size and terrain. The people selected to take the survey were involved in community projects with the organization and participated voluntarily. Participants who were able to read and write filled out their own survey and those that could not read and/or write dictated their answers. I am still in the process of analyzing the data to assess the different factors affecting both formalization of land ownership and receiving loans. Important results regarding methodology include elaborating on certain questions, making sure to only use participants who are out of school, and asking for explanations of specific reasons that participants did not apply for loans.
Funding Provided by: Schulz Fund for Environmental Studies
Skewed Lotteries: Examining Inequalities of Access in the New Boston Public Schools Assignment System
Thomas Conkling (2014); Mentor(s): Amanda Hollis-Brusky
Abstract: In an attempt to increase community involvement in schools and to reduce massive busing costs, Boston Public Schools will begin implementing a new school assignment system in August 2014. This new, MIT-designed assignment system, unlike any other in the country, divides schools into four tiers based on their quality, using their students’ achievement and improvement on state-administered standardized tests as proxies for merit. Students entering kindergarten will be given customized school choice menus based on their home address, with at least two schools in the top 25% of public elementary schools (Tier 1), four schools in the top 50% (Tiers 1 & 2), six schools in the top 75% (Tiers 1-3), and any schools within a one mile radius (All Tiers). Students will then rank their choices and enter a lottery to access these schools. Using 2011 and 2012 Massachusetts Comprehensive Assignment System data, this study creates the first citywide map detailing how this new system could allocate students to each school. Then, using 2007-2011 American Community Survey data, this study begins to examine inequalities of access to quality schools, based on zip code. Although the results are not definitive, children from one area of the city could be ten times as likely to attend a Tier 1 school as children from another part, suggesting that there could be great disparities in access to quality public schools. A greatly expanded version of this study will become my senior thesis.
Funding Provided by: Aubrey H. and Eileen J. Seed Student Research Fund
Medicaid Reform in Oregon -- Local Coordinated Care Success
Emily Hayes (2014); Mentor(s): David Menefee-Libey
Abstract: I examined the development of the Coordinated Care Model (CCM) within Deschutes County, Oregon, as part of the state’s reform and expansion of Medicaid under a $2 Billion grant from the federal Centers for Medicare and Medicaid Services. I found that the model has been effectively implemented in two ways that are crucial to policy success: timeliness and accuracy. In Central Oregon, PacificSource (the Medicaid payer) began instituting the policy shortly after passage of state legislation mandating the creation of Coordinated Care Organizations (CCOs) as replacements of standard Medicaid delivery systems. The dimension of accuracy was also positive-the way that the policy is being instituted is in line with the intended implementation. A major factor affecting accuracy is the alignment of the goals of primary actors local to the policy with the policy itself, and in this case that goal alignment has strongly supported the success of the program. Another factor in accuracy is the availability of necessary resources, namely human capital and money. The success of this program in implementation is impressive, but the localized reasons for success provide doubt that it would be able to be mirrored successfully all over the country. Were a CCM to be implemented in an area that was not receptive to reforms or underfunded it is unlikely that implementation would be similarly successful.
Funding Provided by: Faucett Catalyst Fund
The Effects of HIV Exposure and Ready-to-Use Therapeutic Food (RUTF) Supplementation on Infant Growth
Sophia Geffen (2013); Additional Collaborator(s): Sallie Permar*; Mentor(s): Pardis Mahdavi; Heather Williams
*Duke Human Vaccine Institute
Abstract: As of 2010, Malawi possessed an under-5 mortality rate of 92 deaths per 1,000 live births, one of the highest rates in the world (The World Bank, 2012) A significant number of these deaths are due to perinatal HIV exposure and infection. Therefore, it is of great interest to study the factors that enable children to achieve HIV-free survival. The CHAVI009 study investigated the effects of HIV exposure and ready-to-use therapeutic food (RUTF) supplementation on infant growth by feeding HIV positive mothers and their babies with RUTF and recording subsequent growth determinants. For my thesis, I will be using the CHAVI009 data to conduct statistical analysis and later place the outcomes of this clinical trail in the context of prevention of mother-to-child transmission (PMTCT) policies in Malawi and other developing nations. If the results of the study prove that RUTF is an effective means to counteract the growth effects of HIV-exposure, then this protocol could be used for PMTCT policies and programs worldwide.
Funding Provided by: Aubrey H. and Eileen J. Seed Student Research Fund