Gary Simon

Hematocrit as a predictor of significant injury after penetrating trauma

A study was undertaken to determine if there are differences in hematocrit (HCT) during the initial evaluation of patients with and without significant visceral intrathoracic or intraabdominal injury after penetrating trauma and, if so, the predictive value of this parameter. Sixty consecutive adults with potentially significant penetrating trauma who presented to an urban municipal trauma center during a 10-week period were studied. Diagnostic variables were recorded for all patients at risk for significant injury, defined as intrathoracic or intraabdominal injury requiring surgical intervention. Patients were ultimately grouped on the basis of operative findings or observation to discharge. Acquired variables included vital signs, initial HCT (HCT-0), HCT at 15 minutes (HCT-15), HCT at 30 minutes (HCT-30), and fluid administered. Twenty-one patients had significant injuries (INJ), and 39 did not (NO-INJ). INJ patients had lower HCT values than NO-INJ on presentation (35% +/- 6% and 41% +/- 5%, respectively). At presentation, a low HCT was predictive of significant injury, but a normal HCT did not preclude injury. The changes in HCT from arrival to 15 minutes, between INJ and NO-INJ patients, were similar (-1.5% +/- 3% and -0.6% +/- 3% respectively). Only when the decrease in HCT was > or =6.5% from presentation measurements was it predictive of injury. During the first 15 minutes a decrease in HCT of > or =6.5% had a positive predictive value and specificity of 1.0. The change in HCT between 15 and 30 minutes was less useful. There was a large difference between the amounts of fluid given to injured and uninjured patients, which may have been responsible for some of the differences in HCT between the two groups. These results show that HCT may have some diagnostic utility during the early management of penetrating trauma. Presentation with an HCT below normal, or an early decrease in HCT, is an indicator of potential injury. Although many patients with serious internal injuries do not manifest large decreases early after presentation, those who do have a high probability of internal injury. The lower the HCT, or the greater the decrease, the greater the probability that a significant injury exists.

Diagnostic Plots for Missing Data in Least Squares Regression

Abstract The usual approach to handling missing data in a regression is to assume that the points are missing at random (MAR) and use either a fill-in method to replace the missing points or a method using maximally available pairs in the sample covariance matrix. We derive limits for the values of the least squares estimates of the coefficients (and their associated t statistics) when there are missing observations in one carrier. These limits are derived subject to a constraint on the relationship of the missing data to the present data. Calculating these limits while varying this constrained value results in a series of diagnostic plots that can be used to study the potential effect of the missing points on the regression (without assuming that the points are MAR). Simulations are performed to illustrate the use of the plots, and two real data sets are analyzed. The more general case of missing data in more than one carrier is also discussed.

A unified maximum likelihood approach to document retrieval

Empirical work shows significant benefits from using relevance feedback data to improve information retrieval (IR) performance. Still, one fundamental difficulty has limited the ability to fully exploit this valuable data. The problem is that it is not clear whether the relevance feedback data should be used to train the system about what the users really mean, or about what the documents really mean. In this paper, we resolve the question using a maximum likelihood framework. We show how all the available data can be used to simultaneously estimate both documents and queries in proportions that are optimal in a maximum likelihood sense. The resulting algorithm is directly applicable to many approaches to IR, and the unified framework can help explain previously reported results as well as guide the search for new methods that utilize feedback data in IR.

“Last licks”: Do they really help?

Much has been written about the home field advantage in sports. Baseball and softball are unusual games, in that the rules are explicitly different for home versus visiting teams, since by rule home teams bat second in each inning (they have “last licks”). This is generally considered to be an advantage, which seems to be contradicted by the apparent weakness of the home field advantage in baseball compared to that in other sports. In this article we examine the effect of “last licks” on baseball and softball team success using neutral site college baseball and softball playoff games. We find little evidence of an effect in baseball, but much greater evidence in softball, related to whether a game is close late in the game. In softball games that are tied at the end of an inning, batting last seems to be disadvantageous later in the game, apparently related to the chances of the team scoring first to break the tie. Since the database also includes games where one team is playing on its home field, we are also able to say something about benefits from playing at home that are not related to “last licks.”

Were the 1996–2000 Yankees the Best Baseball Team Ever?

Statistics Working Papers Series

Spatial Regression Models Using Inter-Region Distances in a Non-Random Context

This article considers spatial data z( s 1), z( s 2),…, z( s n ) collected at n locations, with the objective of predicting z( s 0) at another location. The usual method of analysis for this problem is kriging, but here we introduce a new signal-plus-noise model whose essential feature is the identification of hot spots. The signal decays in relation to distance from hot spots. We show that hot spots can be located with high accuracy and that the decay parameter can be estimated accurately. This new model compares well to kriging in simulations.

Map Neighbor Counts

This paper investigates a conjecture of Lowell regarding the number of neighbors on maps divided into regions. Lowell speculated that border regions have on average three neighbors and that nonborder regions have on average six neighbors. This paper investigates the conjecture and, for a large category of maps, an exact relationship is obtained. Exceptions to this large category are considered, and a further exact relationship is obtained. The results of this paper may find applications not only for geographic maps, but also for models involving planar networks.

Confidentiality guaranteed: a noninvasive procedure for collecting sensitive information

We present a scheme in which subjects select balls from each of two urns, reporting only the result from the urn appropriate to their response to a sensitive question. Subjects can thus be induced to respond to the sensitive question while preserving confidentiality. Design aspects are discussed.