### Table 5: Canonical Correlations

"... In PAGE 27: ... The higher the correlation between the models, the less likely it is that either the Cox test or the Vuong can discriminate between them (see Anonymous 1999 for monte carlo simulation results). [ Table5 about here.] The Bayesian analysis echoes the results of the Cox and Vuong tests.... ..."

Cited by 4

### Table 6. Canonical redundancy analysis Canonical

2003

"... In PAGE 7: ... Although the first and second canonical functions are ignificant according to the above analysis, it is mmended that redundancy analysis be utilized to etermine which functions should be used in the terpretation [37]. Redundancy is defined as the ability of f independent variables, taken as a set, to explain the ariation in the dependent variables taken one at a time Table6 summarizes the redundancy analysis for the dent and independent variables for the two canonical unctions that were found to be significant by using the easure of model fit. The results indicate that the first onical function accounts for the highest proportion of otal redundancy (93.... ..."

Cited by 2

### Table 6. Canonical redundancy analysis Canonical

2003

"... In PAGE 7: ... Although the first and second canonical functions are ignificant according to the above analysis, it is mmended that redundancy analysis be utilized to etermine which functions should be used in the terpretation [37]. Redundancy is defined as the ability of f independent variables, taken as a set, to explain the ariation in the dependent variables taken one at a time Table6 summarizes the redundancy analysis for the dent and independent variables for the two canonical unctions that were found to be significant by using the easure of model fit. The results indicate that the first onical function accounts for the highest proportion of otal redundancy (93.... ..."

Cited by 2

### Table 4: Canonical Correlation Analysis for the Simulated Data: Method 3 and 4

"... In PAGE 9: ... Table 3 shows the results of the two simulations when an ordinary canonical correlation analysis was performed and when the first two of our four diagnostic methods were applied. Table4 shows the results when the last two of our four diagnostic methods were applied using both Huber and Campbell weighting systems. The distributions for the canonical correlations were reasonably normal allowing t-tests to be performed in order to compare these four methods.... In PAGE 9: ....0041 0.0041 0.0038 0.0047 0.0048 0.0045 Table4 shows the mean weights for the observations when the Robust Correlation and Robust Canonical Correlation methods are used as well as the average number of points deleted for the two simulations. The canonical correlation coefficients are all much higher than those found in Table 3 suggesting that methods 3 and 4 are more effective than methods 1 and 2.... ..."

### Table 7: Canonical Correlations

"... In PAGE 28: ... The nonnested tests should do a better job of discriminating between these models. [ Table7 about here.] The results of the Cox test are in Table 8, and they are quite clear.... ..."

Cited by 4

### Table 1 Canonical structure loadings

"... In PAGE 3: ... These structure loadings are commonly used when interpreting the meaning of the canonical variable because the structure loadings appear to be more stable, and they allow for the interpretation of canonical variable in the manner that is analogous to factor analysis. The structure loadings for the first two canonical variables are presented in Table1 . The first and the second canonical functions accounts for 37% and 17% of the variation in the discriminating variables.... ..."

### Table 3: Canonical Correlation Analysis for the Simulated Data: Method 1 and 2 Covariance Matrix for

"... In PAGE 9: ... We therefore expected that contaminated points would be more difficult to detect in the second simulation. Table3 confirms that the effect of removing contaminated points was to increase the canonical correlation. Table 3 shows the results of the two simulations when an ordinary canonical correlation analysis was performed and when the first two of our four diagnostic methods were applied.... In PAGE 9: ... Table 3 confirms that the effect of removing contaminated points was to increase the canonical correlation. Table3 shows the results of the two simulations when an ordinary canonical correlation analysis was performed and when the first two of our four diagnostic methods were applied. Table 4 shows the results when the last two of our four diagnostic methods were applied using both Huber and Campbell weighting systems.... In PAGE 9: ... The distributions for the canonical correlations were reasonably normal allowing t-tests to be performed in order to compare these four methods. As shown in Table3 deleting one of the contaminated points (method 1) caused a significant increase in the canonical correlation for both simulations. Using score regression diagnostics produced a significant improvement in the canonical correlation only in the case of the first simulation where, on average, 27.... In PAGE 9: ....0041 0.0041 0.0038 0.0047 0.0048 0.0045 Table 4 shows the mean weights for the observations when the Robust Correlation and Robust Canonical Correlation methods are used as well as the average number of points deleted for the two simulations. The canonical correlation coefficients are all much higher than those found in Table3 suggesting that methods 3 and 4 are more effective than methods 1 and 2. The best result (highest canonical correlation) is obtained with method 4 (robust canonical correlation) and the Campbell weights, however this method down-weights too many observations too heavily for comfort (60.... ..."

### Table 3: Canonical Sequence Analysis

1999

"... In PAGE 6: ... Add any additional property / value pairs you feel are needed for clarification. This activity is performed in Table3 for the enumeration of Table 2. The entries in the table associate a precise condition with each canonical sequence.... In PAGE 6: ... Finally, add the recursive rule to every function : . Given the enumeration in Table 2 and the analysis of Table3 , definitions for specification functions door and combo are derived in Table 4 and Table 5, respectively. A comparison reveals that these specification functions are almost identical to the specification functions derived earlier, except that in the earlier version of combo, the combination is not reset until the door is closed.... ..."

Cited by 1

### Table 1 - Extended usage-centered design notation for activity modeling

"... In PAGE 7: ... The objective is a simple notation that expresses clear distinctions where needed with minimal additions. The notation for activity modeling summarized in Table1 adds to the established notation already used in usage-centered design four new symbols for activities, actions, artifacts, and non-actor participants. It is important to keep in mind that these models are being introduced to maximize utility and efficiency in representing activity context for interaction design purposes rather than for software engineering.... In PAGE 9: ... competing activities involving shared resources in common with proximate activities 4. adjacent activities in the same setting but otherwise unrelated to proximate activities Activities are represented in an Activity Map by the block shape shown in Table1 . A line or arrow connecting one activity to another represents a relationship.... In PAGE 13: ... In this context, actions refer to goal-directed interactions among actors or players and between them and artifacts other than the system of reference. Actions are represented by a distinct symbol (the barred ellipse seen in Table1 ), a variation of the symbol already generally used to represent task cases. The Task Map, a model used to represent the interrelationships among task cases in usage- centered design, can be extended to incorporate activities and actions.... ..."

### Table 6: Logit analysis of the active enrollment decision

"... In PAGE 17: ... Then, the decision to enroll early should depend primarily on expected drug costs in 2006, which are highly correlated with drug costs in 2005. The second column of Table6 shows logit results for early enrollment (defined as being enrolled by March 2006). The results are as expected for rational individuals: Drug costs in 2005 are a very strong predictor of early enrollment while the socio- demographic variables have no significant impact.... In PAGE 17: ...osts. Those are also correlated with 2005 drug costs but weaker than 2006 costs. In addition to that, individual expectations, tastes, and the understanding of the penalty and its expected present value drive the decision whether to enroll late or not at all, given that early enrollment is not beneficial. The final column of Table6 shows logit results for whether individuals enroll late (April or May... In PAGE 18: ... They may reflect health and other expectations, information, and/or tastes. Taken together, the models in Table6 show that the strong predictive power of drug costs for total enrollment (column 1) is mainly driven by early enrollers (column 2), while the enrollment differentials by socio-economic variables are mainly driven by late enrollers (column 3). This is consistent with a view that most individuals understood at least the gross attributes of the initial enrollment alternatives and the incentives they faced.... In PAGE 25: ... The market shares we obtained from RPS-2006 data are well in line with those computed using official CMS data by Cubanski and Neuman (2006, Table 1). Table 10 shows results from OLS and quantile regressions for chosen plan premiums, using the same covariates as in Table6 . Higher pharmacy bills substantially increase chosen plan premiums, especially in the lower part of the APB distribution.... ..."