Skipping class in college and exam performance: Evidence from a regression discontinuity classroom experiment

In this paper we estimate the effect of class attendance on exam performance by implementing a policy in three large economics classes that required students scoring below the median on the midterm exam to attend class. This policy generated a large discontinuity in the rate of post-midterm attendance at the median of the midterm score. We estimate that near the policy threshold, the post-midterm attendance rate was 36 percentage points higher for those students facing compulsory attendance. The discontinuous attendance policy is also associated with a significant difference in performance on the final exam. We estimate that a 10 percentage point increase in a student's overall attendance rate results in a 0.17 standard deviation increase in the final exam score without adversely affecting performance on other classes taken concurrently.     

A pipelined architecture for distributed text query evaluation

Two principal query-evaluation methodologies have been described for cluster-based implementation of distributed information retrieval systems: document partitioning and term partitioning. In a document-partitioned system, each of the processors hosts a subset of the documents in the collection, and executes every query against its local sub-collection. In a term-partitioned system, each of the processors hosts a subset of the inverted lists that make up the index of the collection, and serves them to a central machine as they are required for query evaluation. In this paper we introduce a pipelined query-evaluation methodology, based on a term-partitioned index, in which partially evaluated queries are passed amongst the set of processors that host the query terms. This arrangement retains the disk read benefits of term partitioning, but more effectively shares the computational load. We compare the three methodologies experimentally, and show that term distribution is inefficient and scales poorly. The new pipelined approach offers efficient memory utilization and efficient use of disk accesses, but suffers from problems with load balancing between nodes. Until these problems are resolved, document partitioning remains the preferred method. Alistair Moffat was supported by the Australian Research Council, the ARC Special Research Center for Perceptive and Intelligent Machines in Complex Environments, and the NICTA Victoria Laboratory. William Webber and Justin Zobel were supported by the Australian Research Council. Ricardo Baeza-Yates was supported by Grant P01-029-F from Millennium Initiative of Mideplan, Chile; and by the University of Melbourne as a visiting scholar at the time this project was undertaken.     

Resilience of anthropometric data assembly strategies to imposed error

The aim of this study was to examine strategies for selecting a criterion value during anthropometric data assembly and their resilience to imposed error. Sixty-seven women aged 16-60 years were subjected to three separate series of measurements, which included six skinfolds and three girths. A random error term was added to the first of the three series of measurements to produce a pseudo-series containing error, termed the 'erroneous replicated series'. Five strategies were then used to determine the criterion value of each of the replication series: the first measurement, the mean of the first two measurements, the mean of all three measurements, the mean of the closest two measurements and the median of the three measurements. The technical error between the criterion values of the true and the flawed replication series for each of the selection strategies was calculated. We found that, for five of the skinfolds and all of the girths, the median value provided the smallest technical error of measurement between the criterion values for the true and erroneous replication series. We conclude that the strategy of selecting the median of three measurements is the most resilient to imposed error.