### Table 2. Multivariate survival analyses

"... In PAGE 5: ... Thus, the observed difference in log rank was found to occur by chance 179 times out of 1000. Multivariate results Results from fitting Cox proportional hazards models to pairs of factors univariately significant are summar- ized in Table2 . The table shows prognostic factors grouped by univariately significant factors.... ..."

### Table 4. Multivariate analysis of overall survival.

"... In PAGE 4: ... 63%) although the numbers were quite small. Multivariate analyses ( Table4 ) showed signifi- cantly better survival in patients with CML, low risk diseases, no acute GVHD and those transplanted with bone marrow as the source of stem cells. Few patients, only 7, were transplanted with PBPC.... In PAGE 4: ... A multivariate analysis per- formed adding 41 patients aged 41-49 years (inter- mediate age group), who fulfilled the same inclu- sion criteria used for the young-group, showed similar results (data no shown) when compared with the analysis of the two original groups (elder- ly and young). Discussion In this study 32 patients ( Table4 ) aged 50 years and older (elderly-group) were compared with a group of 92 young adults (aged 20-40), all condi- tioned with a myeloablative regimen and trans- planted from HLA identical siblings at a single institution. As far as we know, this is the largest single-center series of allogeneic SCT in patients 50 years and over published by a European team.... ..."

### Table 16: Best multivariate model

"... In PAGE 23: ...25 in univariate analysis were al- lowed to enter the model. Table16 shows the coefficients and statistical significance of the measures in the final model. The variables in the model have two multivariate outliers, which are not influential and were therefore retained for the model fitting.... In PAGE 25: ... This indicates that, as expected, larger classes tend to be more fault-prone. Results from a 10-Cross validation study We performed a 10-cross validation of the prediction model in Table16 . The 113 data points were randomly split into ten partitions of roughly equal size (seven partitions of 11 data points each, three partitions of 12 data points).... ..."

### Table 5. Median Survival Times (Weeks) Under Each Model Exponential Mixture Changepoint Kaplan-Meier

"... In PAGE 16: ... The surviving fraction of the mixture and changepoint models create horizontal asymptotes that clearly deviate from the monotonically-increasing-to-one tails of the exponential model. Table5 presents the median for each of these predictive distributions. Table 6 presents the value of the predictive CDFs at 104 weeks.... In PAGE 16: ... While the Kaplan-Meier Curves are included in these tables to serve as a point of reference and aid us in understanding the di erences in the models, comparisons between the Kaplan-Meier curves and the three parametric models should be done remembering that there is no sharing of information across centers in the Kaplan-Meier Curves. Figure 5 about here Table5 and Table 6 about here Table 5 also shows that within models the prior makes the largest di erence in estimating median survival for Center D. Moving from Prior I to Prior II pulls down the median by 13, 7, and 10 weeks for the exponential, mixture, and changepoint models respectively.... In PAGE 16: ... While the Kaplan-Meier Curves are included in these tables to serve as a point of reference and aid us in understanding the di erences in the models, comparisons between the Kaplan-Meier curves and the three parametric models should be done remembering that there is no sharing of information across centers in the Kaplan-Meier Curves. Figure 5 about here Table 5 and Table 6 about here Table5 also shows that within models the prior makes the largest di erence in estimating median survival for Center D. Moving from Prior I to Prior II pulls down the median by 13, 7, and 10 weeks for the exponential, mixture, and changepoint models respectively.... In PAGE 17: ...by 4, 2, and 4 weeks for the exponential, mixture, and changepoint models respectively. Exploring the heterogeneity across centers, Table5 shows that the median survival time is clearly the smallest for Center E and largest for Center D regardless of prior and model. Proceeding from column 2 to column 8 of Table 5, the range across the centers is 15-73, 19-60, 11-62, 13-55, 14-59, 18-49, and 5-72.... In PAGE 17: ... Exploring the heterogeneity across centers, Table 5 shows that the median survival time is clearly the smallest for Center E and largest for Center D regardless of prior and model. Proceeding from column 2 to column 8 of Table5 , the range across the centers is 15-73, 19-60, 11-62, 13-55, 14-59, 18-49, and 5-72. The heterogeneity is greatest for the Kaplan-Meier curves because no information is shared across the centers.... In PAGE 18: ... For example, look at Di erence B for Center A under Prior I. From Table5 , the Kaplan-Meier estimate of median survival for Center A was 49 weeks. Excluding Center A from the analysis, gave a predictive survival median for an unobserved center of 29 weeks.... ..."

### Table 3: Growing Mixture of Experts on the Multivariate Data-set. Number of Experts Ave. MSE

1999

Cited by 17

### Table 3: Growing Mixture of Experts on the Multivariate Data-set. Number of Experts Ave. MSE

### Table 7 Median Survival Times (Weeks) For an Unobserved Center When Observed Centers are Excluded Exponential Mixture Changepoint

"... In PAGE 17: ... The six models were rerun 5 times, each time excluding one treatment center. The impact of the exclusion is summarized in Table7 . This table presents the median survival time for an unobserved center, again denoted Center F.... In PAGE 18: ...ncluded is 34. Hence the di erence is 29-34=-5. Di erence B presents the error in our prediction if we used the median for an unobserved center to predict the Kaplan-Meier median for the excluded center. Table7 about here The table shows that median survival times for an unobserved center were most a ected by the exclusion of Centers D and E. The largest increase in median survival under each of the 6 models occurred when Center E was excluded, the largest decreases when Center D was excluded.... In PAGE 18: ... If we used this value to predict Center A apos;s median survival we would miss by 20 weeks. This is the Di erence B reported in Table7 . We see di erences for other centers ranging from -40 to +49.... ..."

### Table 8 Multivariate Logistic Regression Results for Troubled Company Models

"... In PAGE 24: ... Sound Firms Power comparisons for the troubled and sound firm samples using data for 1990 and 1991 are shown in Tables 6 and 7. Coefficient estimates and t-values for the corresponding multiple logistic regression models are shown in Table8 . The implications of the results for the troubled versus sound insurer samples generally are similar to those for the failed versus solvent insurer samples.... In PAGE 24: ... The most notable difference is that including RBC/S or its separate components with the FAST score increases power more often than was the case for the failed versus solvent insurers. Consistent with this result, the coefficient on RBC/S in the multiple logistic model that includes RBC/S and the FAST score (see Table8 ) is positive and highly significant for 1990 and 1991, and a likelihood ratio test rejects the null hypothesis that the coefficients on the separate RBC components are jointly equal to zero for both years. (SCOTT CHECK) These results are consistent with the hypothesis that the development of RBC... ..."

### Table 1: Some Members of the Scale Mixtures of Multivariate Normal Distributions and the Cor- responding Discrete Choice Models (See Appendix for the Notation)

in Parametric Discrete Choice Models Based on the Scale Mixtures of Multivariate Normal Distributions

"... In PAGE 4: ... More detailed descriptions can be found in Choy (1995) and Chen and Dey (1998). We refer readers to the Appendix for the notation used in Table1 and thereafter. (insert Table 1 around here) As in the case of MNP, an inherent unidenti cation problem exists for the new class of models based on SMMVN.... In PAGE 4: ... We refer readers to the Appendix for the notation used in Table 1 and thereafter. (insert Table1 around here) As in the case of MNP, an inherent unidenti cation problem exists for the new class of models based on SMMVN. Scaling the wi in (2) by a positive constant c doesn apos;t change the value of yi, but does result in a di erent parameterization (c ; c ( i) ).... In PAGE 9: ...in Table1 . It is easy to see that under MNT, ijyi; ; ; wi G J + v 2 ; 1 2 (wi?xi )0 ?1(wi?xi ) + v : For other models, however, no standard distributions exist for ijyi; ; ; wi.... ..."

### Table 2. Cox multivariate analysis of distant

"... In PAGE 4: ... 1B). A multivariate Coxregression analysis of distant disease-free survival, including lymph node status, menopausal status, tumor size, ERBB2 amplification, ERa status, two dummy variables for ERhC0 (versus ++ and + versus ++), and two interaction terms for ERh with ERa status ( Table2 ) showed a significantly worse distant disease-free survival for the ERhC0 group compared with ERh++ in the ERa-negative subgroup, with a hazard ratio (HR) of 14 [95% confidence interval (95% CI), 1.... In PAGE 5: ... 1D and F-G, respectively). From the Coxregression model with interaction between ERa and ERh ( Table2 ), the predictive value of ERa for distant disease-free survival in the different ERh subgroups could be interpreted. In agreement with the results from the Kaplan-Meier analysis, we found that ERa positivity was a significant independent predictive marker for improved distant disease-free survival only within the ERhC0 group (HR, 0.... In PAGE 5: ...egative tumors were similar (P = 0.68 for ERa and P =0.41for PgR, Kruskal-Wallis test), we concluded that residual low levels of ERa or PgR protein in the three ERh subgroups did not contribute to the prognostic effect of ERh within the ERa- negative tumors. In addition to ERh++ status compared with ERhC0 in the ERa- negative subgroup, four or more lymph nodes (compared with 0) and menopausal status had independent prognostic value after tamoxifen treatment in the multivariate analysis ( Table2 ). No other effects were significant; however, ERBB2 amplification and ERh+ status (compared with ERh++) in the ERa-negative subgroup showed a tendency (P lt; 0.... In PAGE 5: ... No other effects were significant; however, ERBB2 amplification and ERh+ status (compared with ERh++) in the ERa-negative subgroup showed a tendency (P lt; 0.10) to carry prognostic information ( Table2 ). S-phase fraction, age as a continuous variable, PgR, and DNA ploidy status were not included in the multivariate analysis as they showed no significant association with distant disease-free survival in univariate analysis.... In PAGE 6: ....21-0.89; P = 0.02, extrapolated from Table2 ). Moreover, it appeared that ERa negativity may confer a better outcome within the ERh strongly positive group (P = 0.... In PAGE 6: ....2; 95% CI, 0.67-40; P = 0.12, Table2 ). These findings raise an intriguing hypothesis that the best tamoxifen response is achieved when a tumor is expressing either ERa or ERh, but not both.... ..."