Robustness issues in multilevel regression analysis

A multilevel problem concerns a population with a hierarchical structure. A sample from such a population can be described as a multistage sample. First, a sample of higher level units is drawn (e.g. schools or organizations), and next a sample of the sub‐units from the available units (e.g. pupils in schools or employees in organizations). In such samples, the individual observations are in general not completely independent. Multilevel analysis software accounts for this dependence and in recent years these programs have been widely accepted. Two problems that occur in the practice of multilevel modeling will be discussed. The first problem is the choice of the sample sizes at the different levels. What are sufficient sample sizes for accurate estimation? The second problem is the normality assumption of the level‐2 error distribution. When one wants to conduct tests of significance, the errors need to be normally distributed. What happens when this is not the case? In this paper, simulation studies are used to answer both questions. With respect to the first question, the results show that a small sample size at level two (meaning a sample of 50 or less) leads to biased estimates of the second‐level standard errors. The answer to the second question is that only the standard errors for the random effects at the second level are highly inaccurate if the distributional assumptions concerning the level‐2 errors are not fulfilled. Robust standard errors turn out to be more reliable than the asymptotic standard errors based on maximum likelihood.     

Outcomes from use of an evidence-based practice guideline

The aim of evidence-based guidelines is primarily to improve patient outcomes without adding to the existing cost of care because both payers and policymakers want to identify health care costs that do not result in benefit to the patient. The purpose of the reported project was to generate a practice guideline for the treatment of uncomplicated acute cystitis in a female population, to determine the extent to which the guideline would be used by providers and to measure the cost and quality of outcomes from its use. A retrospective chart review was used to gather pre-guideline practice and cost data. Measurements included the type, frequency, and duration of antibiotic therapy and the use of urine cultures and both complications and routine followup visits. The implementation of an outpatient practice guideline resulted in a significant change in antibiotic prescribing and a trend toward a change in ordering cultures and clinic followup. There was also a significant decrease in treatment costs.     

A New Approach to Gravity Anomaly Separation: Differential Markov Random Field (DMRF)

In this paper the authors introduce a novel approach to stochastic image processing, denoted as Differential Markov Random Field (DMRF), which has been applied to gravity anomaly separation problems. The advantages of the method are that it introduces only little distortion into the shape of the original image and that it is not affected significantly by factors such as the overlap power spectra of regional and residual fields. Testing of the proposed meth using synthetic examples gave excellent results.     

Review of Microstructural Evidence of Magmatic and Solid-State Flow

Evidence of magmatic flow includes: (a) parallel to sub-parallel alignment of elongate euhedral crystals (e.g., of feldspar or hornblende) that are not internally deformed, (b) imbrication (‘tiling’) of elongate euhedral crystals that are not internally deformed, (c) insufficient solid-state strain in regions between aligned or imbricated crystals to accommodate phenocryst rotation, (d) elongation of microgranitoid enclaves without plastic deformation of the minerals, (e) magmatic flow foliations and elongate microgranitoid enclaves deflected around xenoliths, and (f) schlieren layering (if due to flow sorting) in the absence of plastic deformation of the minerals involved. These features are consistent with rotation of crystals in a much weaker medium, namely a melt phase, at a stage when the magma has become viscous enough to preserve the alignment. Evidence of solid-state flow includes: (a) internal deformation and recrystallization of grains, (b) recrystallized “tails,” (c) elongation of recrystallized aggregates (e.g. of quartz and mica), (d) grainsize reduction, (e) fine-grained folia anastomosing around less deformed relics, (f) microcline twinning, (g) myrmekite, (h) flame perthite, (i) boudinage of strong minerals, typically with recrystallized aggregates of weaker minerals (e.g. quartz and mica) between the boudins, (j) foliation passing through, rather than around enclaves, and (k) heterogeneous strain with local mylonitic zones. Several criteria suggest “submagmatic flow,” including recrystallized feldspar, inferred transitions from magmatic imbrication to solid-state S/C arrangements, evidence of c-slip in quartz, and especially evidence of migration of residual melt into lower-pressure sites. Recent experimental studies indicate that a change from grain-supported flow to suspension flow typically occurs in deforming magmas at melt contents of between 20% to 40%, and that large amounts of strain may accumulate in magmas without being recorded by the final fabric. At lower melt percentages, perhaps as low as a few percent, depending on the minerals and their shapes, strain may be accommodated by: (a) melt-assisted grain-boundary sliding, (b) contact-melting assisted grain-boundary migration, (c) strain partitioning into melt-rich zones, (d) intracrystalline plastic deformation (c-slip in quartz indicating plastic deformation at temperatures near the granite solidus), and (f) transfer of melt to sites of low mean stress. The only indication of strain in the absence of crystal plasticity may be an alignment of crystals. Moreover, magmatic flow microstructures may be destroyed by fracturing, crystal plasticity and recrystallization before the magma reaches its solidus. Many rocks show evidence of solid-state flow superimposed on magmatic flow. Evidence of magmatic flow is commonly preserved in deformed felsic metamorphic rocks: for example the alignment of rectangular K-feldspar megacrysts and of microgranitoid enclaves. However, absence of alignment does not preclude a magmatic origin for K-feldspar megacrysts in felsic gneisses, as magmatic flow may cease before the magma becomes viscous enough to preserve an alignment.     

Aggregation bias, compositional change, and the border effect

Borders affect the composition, not only the level, of interregional trade. In disaggregated U.S. Commodity Flow data, border effects vary substantially across commodities. Substantial border–induced compositional change suggests the possibility that standard estimates suffer from aggregation bias arising from endogenous industry location patterns and the presence of zero observations in commodity–level trade. Adjusting for these effects reduces the estimate of the aggregate border effect from 20.9 to 5.7. JEL Classification: F14, F15
Biais d'agrégation, changement de composition, et effet de frontières. Les frontières affectent la composition et pas seulement le volume de commerce inter-régional. Une analyse des données désagrégées des flux de commerce de biens des Etats-Unis montre que l'effet de frontières varie substantiellement selon les biens. Le changement de composition du commerce engendré par les frontières suggère la possibilité que les estimations usuelles souffrent d'un biais d'agrégation résultant des patterns de localisation industrielle endogènes et de la présence d'observations nulles dans le commerce de certains biens. Un ajustement pour tenir compte de ces effets suggère que les effets de frontières passent de 20.9 à 5.7.     

The efficiency of top agents: An analysis through service strategy in tennis

We consider the question whether top tennis players in a top tournament (Wimbledon) employ an optimal (efficient) service strategy. While we show that top players do not, in general, follow an optimal strategy, our principal result is that the estimated inefficiencies are not large: the inefficiency regarding winning a point on service is on average 1.1% for men and 2.0% for women, implying that–by adopting an efficient service strategy–players can (on average) increase the probability of winning a match by 2.4%-points for men and 3.2%-points for women. While the inefficiencies may seem small, the financial consequences for the efficient player at Wimbledon can be substantial: the expected paycheck could rise by 18.7% for men and even by 32.8% for women. We use these findings to shed some light on the question of whether economic agents are successful optimizers.     

An investigation of the giving behavior of loyal,lapsed, and non‐givers to a religious organization

• This paper provides a comparative study of the giving behavior of current donors (CDs), lapsed donors (LDs), and non‐donors (NDs) in a Catholic diocese in New England. Focus groups were conducted and surveys were administered to Catholics to compare the attitudes and behaviors of lapsed and active donors as well as those who had not given to a recent diocesan annual campaign. The results show significant differences between the groups. Active donors indicate a much greater level of satisfaction with the church, are more aware of its programs, and feel they are more recognized for their gifts than do lapsed or NDs. These findings suggest that relationship fundraising, and creating trust, are critical to forging and maintaining valued donor relationships. The program implemented by the diocese as a result of the study is described.
• Marketing for religious organizations
• Lapsed donors
• Nonprofit marketing
• Marketing for churches

Copyright © 2009 John Wiley & Sons, Ltd.