### Table 1. Frequencies of unintentional injuries, Ontario, 1996.

2004

"... In PAGE 16: ... These categories include motor vehicle crashes, falls, poisoning, drowning and suffocation, fires, water transport, air and space, railway, pedal cycle, and a residual category other that includes motor vehicle non-traffic, natural and environmental, and recreational and other incidents categories. Table1 below summarizes the 1996 data on unintentional injuries for Ontario. The mortality and hospitalization data are based on nationally collected Canadian data.... In PAGE 16: ...Of the causes of injuries reported in Table1 , falls are the most frequent, followed by motor vehicle crashes. In the case of deaths, frequency is higher for motor vehicle crashes than for falls.... ..."

### Table 2. Direct and Indirect Costs ($ Millions) Resulting from Unintentional Injury, Ontario 1996.

2004

"... In PAGE 17: ... 28). Particularly startling in Table2 is the very high mortality costs of motor vehicle crashes in Ontario, largely due to the younger ages of these victims giving rise to large estimated future productivity losses. As context, SMARTRISK reports that Nationally, injury ranks third behind cardiovascular and musculoskeletal disease in terms of societal economic burden (p.... ..."

### Table 1: Sampling method versus full enumeration on a WWW network. Results of the sampling method of 3-node subgraphs compared to the full enumeration results, on a WWW network of the nd.edu domain (Barabasi and Albert 1999). The nodes represent web pages and the edges represent directed hyperlinks between pages. All 13 3-node connected subgraphs appear in the network. It can be seen that as few as 5000 samples (out of 287 million 3-node subgraphs) already give quite a good approximation of all the subgraph concentrations. Highlighted subgraphs were found to be network motifs.

2004

"... In PAGE 5: ...87x106. Running the algorithm with as few as 5000 samples gives a good approximation for all thirteen 3-node subgraph concentrations ( Table1 ). Even with 5000 samples, the 5 different network motifs (shaded subgraphs) are detected as significant vs.... ..."

Cited by 17

### Table 2. Top-5 most cited articles in each of the five Social Network Analysis clusters. TC=Times Cited; BC=Betweenness Centrality; H-L=Citation Half Life.

"... In PAGE 6: ... The span of five Social Network Analysis clusters over time. The most cited five articles from each cluster are shown in Table2 . The top-5 articles in C0 include Barabasi (1999) and Albert (2002); both are the seminal articles in the surge of the complex network analysis paradigm.... ..."

### Table 2. Top-5 most cited articles in each of the five Social Network Analysis clusters. TC=Times Cited; BC=Betweenness Centrality; H-L=Citation Half Life.

"... In PAGE 5: ... The span of five Social Network Analysis clusters over time. The most cited five articles from each cluster are shown in Table2 . The top-5 articles in C0 include Barabasi (1999) and Albert (2002); both are the seminal articles in the surge of the complex network analysis paradigm.... ..."

### Table 3: True values and posterior estimates from the simulated example. Mean values and standard deviations are based on HMC, mixture of 90% RWMH and 10% HMC (MIX) and data augmentation. Note that cutpoint values are given in the ordered parameterization.

"... In PAGE 16: ... The posterior was also sampled using the data augmentation procedure outlined in Albert amp; Chib (1993). This approach used the same priors as in the HMC Gibbs sampler, excepting the priors for cutpoints which employed a uniform prior over f0= 1 2 14 10g: Parameter estimates are presented in Table3 using the parameterization to facilitate comparisons with the Albert amp; Chib (1993) method. In comparing estimates, we nd that all 3 procedures have done fairly well in recovering the true parameter values, although in each case the 0 intercept term, corresponding to the rst cutpoint, is slightly biased with a smaller value.... ..."

### Table 3: True values and posterior estimates from the simulated example. Mean values and standard deviations are based on HMC, mixture of 90% RWMH and 10% HMC (MIX) and data augmentation. Note that cutpoint values are given in the ordered parameterization.

"... In PAGE 16: ... The posterior was also sampled using the data augmentation procedure outlined in Albert amp; Chib (1993). This approach used the same priors as in the HMC Gibbs sampler, excepting the priors for cutpoints which employed a uniform prior over f0= 1 2 14 10g: Parameter estimates are presented in Table3 using the parameterization to facilitate comparisons with the Albert amp; Chib (1993) method. In comparing estimates, we nd that all 3 procedures have done fairly well in recovering the true parameter values, although in each case the 0 intercept term, corresponding to the rst cutpoint, is slightly biased with a smaller value.... ..."

### Table 6. Blocking probability of regional emergency patients for each hospital, without cooperation

2005

"... In PAGE 14: ... Blocking probability regional emergency patients for each hospital with cooperation Now consider the hospitals without regional cooperation. Table6 presents the fraction of rejected regionals for each hospital. To achieve at most 1% of rejected regionals without cooperation, the Erasmus MC needs 10 emergency beds, the Albert Schweizer hospital 3 beds, the Sint Franciscus Gasthuis 4 beds, and the Dirksland Hospital one bed, resulting in 18 beds in total.... ..."

### Table 1: The payoff matrix of PD

"... In PAGE 2: ... Albert Tucker coined the name and developed the typical payoff matrix of this game in the 1950s. Table1 expresses the payoff matrix of PD in a symmet- ric two-player game, where R, T, S, and P represent reward, temptation, sucker, and punishment, respectively. Payoff re- lations ( T gt; R gt; P gt; S, 2R gt; T + S ) exist among the players, which leads to the dilemma.... ..."