Steven R. Lerman

The Estimation of Choice Probabilities from Choice Based Samples

Ti-H CONCERN of this paper is the estimation of the parameters of a probabilistic choice model when choices rather than decision makers are sampled. Existing estimation methods presuppose an exogeneous sampling process, that is one in which a sequence of decision makers are drawn and their choice behaviors observed. In contrast, in choice based sampling processes, a sequence of chosen alternatives are drawn and the characteristics of the decision makers selecting those alternatives are observed. The problem of estimating a choice model from a choice based sample has suibstantive interest because data collection costs for such processes are often considerably smaller than for exogeneous sampling. Particular instances of this differential occur in the analysis of transportation behavior. For example, in studying choice of mode for work trips, it is often less expensive to survey transit users at the station and auto users at the parking lot than to interview commuters at their homes. Similarly, in examining choice of destination for shopping trips, surveys conducted at various shopping centers offer significant cost savings relative to home interviews.2 While interest in transportation applications provided the original motivation for our work, it has become apparent that choice based sampling processes can be cost effective in the analysis of numerous decision problems. In particular, wherever decision makers are physically clustered according to the alternatives they select, choice based sampling processes can achieve economies of scale not available with exogeneous sampling. Some non-transportation decision problems in which decision makers do cluster as described include the schooling decisions of students, the job decisions of workers, the medical care decisions of patients and the residential location decisions of households. Realization of the sampling cost benefits of choice based samples presupposes of course that the parameters of the underlying choice model can logically be inferred from such samples and that a tractable estimator with desirable statistical properties can be found. We shall, in this paper, confirm the logical supposition, develop a suitable estimator, and characterize the behavior of existing, exogeneous sampling, estimators in the context of choice based samples. An outline of the presentation and summary of major results follows.

The iLab Shared Architecture: A Web Services Infrastructure to Build Communities of Internet Accessible Laboratories

The Massachusetts Institute of Technologys iLab project has developed a distributed software toolkit and middleware service infrastructure to support Internet-accessible laboratories and promote their sharing among schools and universities on a worldwide scale. The project starts with the assumption that the faculty teaching with online labs and the faculty or academic departments that provide those labs are acting in two roles with different goals and concerns. The iLab architecture focuses on fast platform-independent lab development, scalable access for students, and efficient management for lab providers while preserving the autonomy of the faculty actually teaching the students. Over the past two years, the iLab architecture has been adopted by an increasing number of partner universities in Europe, Australia, Africa, Asia, and the United States. The iLab project has demonstrated that online laboratory use can scale to thousands of students dispersed on several continents.

Wireless Laptops as Means For Promoting Active Learning In Large Lecture Halls

Abstract This paper reports on a study that examined the use of wireless laptops for promoting active learning in lecture halls. The study examined students’ behavior in class and their perceptions of the new learning environment throughout three consecutive semesters. An online survey revealed that students have highly positive perceptions about the use of wireless laptops, but less positive perceptions about being active in class. Class observations showed that the use of wireless laptops enhances student-centered, hands-on, and exploratory learning as well as meaningful student-to-student and student-to-instructor interactions. However, findings also show that wireless laptops can become a source of distraction, if used for non-learning purposes.

Computing in Higher Education: The Athena Experience

Project Athena at MIT is an experiment to explore the potential uses of advanced computer technology in the university curriculum. About 60 different educational development projects, spanning virtually all of MITs academic departments, are already in progress.

Hedonic theory, bid rents, and willingness-to-pay: Some extensions of Ellickson's results

Abstract An alternative to Ellicksons multinomial logit model of household bids for dwelling units is derived by making use of observable information on the price paid by the winning bidder. The alternative specification makes it possible to estimate willingness-to-pay for housing attributes, which cannot be obtained from Ellicksons model. Some methodological issues arising from Ellicksons method of grouping households are also examined.

PCoord: network position estimation using peer-to-peer measurements

Several recently emerged Internet services make use of application-level or overlay networks. Examples of such services include overlay multicast, structured peer-to-peer lookup services, and peer-to-peer file sharing. Many of these services could benefit from enabling participating end hosts to estimate their relative network locations within the overlay. We present PCoord, a peer-to-peer network coordinate system for overlay topology discovery and distance prediction. The goal of PCoord is to allow participating peer nodes in an overlay network to collaboratively construct an accurate geometric model of the overlay network topology in a completely decentralized peer-to-peer fashion. We evaluate the PCoord approach through extensive simulations using both real network measurements and simulated topologies. Our results indicate that the constructed geometric model can give accurate pair-wise distance prediction and nearest neighbor discovery. In particular, using a simulated overlay network consisting of over 3,400 peer nodes, our results indicate that over 90% of the peers can predict their closest peers by probing only a small fraction of the global peer population.

A decomposition algorithm for the all-pairs shortest path problem on massively parallel computer architectures

The shortest path problem is a classical and important combinatorial problem with many applications. We examine the solution of the all-pairs shortest path problem on a SIMD fine-grained, massively parallel computer, the Connection Machine. We develop a network decomposition algorithm for the problem and implement it on the Connection Machine. The algorithm consists of the following phases: (1) Decompose the network into subnetworks and identify the set of cutset nodes associated with each subnetwork. (2) For every cutset node determine the shortest path to all other nodes in the network. (3) For each subnetwork solve the all-pairs shortest path problem. (4) Combine results from the two preceeding phases to obtain the final shortest path distances. We discuss mapping strategies of the network to the processors and examine the sensitivity of the execution time of the algorithm to different decomposition strategies. We compare running times on different architectures and draw conclusions on appropriate network decomposition strategies. (Copies available exclusively from MIT Libraries, Rm. 14-0551, Cambridge, MA 02139-4307. Ph. 617-253-5668; Fax 617-253-1690.)

Random Utility Models of Spatial Choice

Random utility models are now in widespread use for analyzing decisions such as mode to work. However, their application to problems of spatial choice have been far fewer and have faced some methodological obstacles specific to choices among numerous alternatives. This paper reviews both the major problems and the alternative solutions which have been developed. Particular attention is given to uses of multinomial logit analysis and its variants primarily because of their potential for dealing with large choice sets without imposing unrealistic computational burdens on the model estimation process.

A model system for forecasting patronage on demand responsive transportation systems

Abstract Demand-responsive transportation (DRT) systems represent a broad class of public transportation options characterized by spatial and/or temporal flexibility in serving demand. These systems offer the potential to provide service in lower density areas. This paper presents a model system which provides DRT system designers with a behaviorally consistent and validated patronage forecasting procedure. The model system consists of three components: a work trip demand model; a non-work trip demand model; and a DRT service model. These components are applied in an equilibrium framework. The demand components of the system are disaggregate demand models. The work trip model forecasts modal split, while the non-work model forecasts trip frequency, time of day, mode and destination choice. The service model forecasts both wait time and ride time. The model was estimated on data from Rochester, New York, and validated against data from Davenport, Iowa and La Habra, California. Total daily trips by DRT appear to be forecast ±20–30%. A sketch planning model was also developed and validated against six other DRT sites with errors of a similar magnitude. The model system can analyze options such as varying the vehicle fleet or service area over the day, using differing vehicle sizes, alternate pricing policies, and improving vehicle speeds.

Project ATHENA as a Next Generation Educational Computing System

Academic computing can be divided into three main segments: administration, research, and education. Historically, computing for these three segments has been provided by large timesharing systems. (Many universities have uncoupled administrative computing from research or educational computing for security reasons, but they still use the common technology of timesharing.) More recently, the price and performance of personal computers have made them attractive for educational computing, usually on a stand-alone basis.