### Table 3. Tests of Predicted Distributions

"... In PAGE 16: ... Thus the behavior of six out of 30 or 20% of the subject pairs is consistent with the subgame perfect equilibrium of the self-regarding preferences model. The first row of Table3 reports the means and standard deviations of the amounts sent by first movers and returned by second movers in treatment A. The fourth row of the table reports results from a Kolmogorov test of the hypothesis that amounts sent are equal to minus five; the test implies rejection of the hypothesis.... In PAGE 17: ... In treatment C, 21 out of 30 subjects chose zero euros, six subjects returned positive amounts, and three subjects returned negative amounts. The second and third rows of Table3 report the means and standard deviations of the amounts sent and returned in treatments B and C. The Kolmogorov test reported in row six does not imply rejection of the hypothesis that amounts sent in treatment B are equal to minus five.... In PAGE 17: ... The Kolmogorov test reported in row six does not imply rejection of the hypothesis that amounts sent in treatment B are equal to minus five. The Kolmogorov test reported in the seventh row of Table3 does not imply rejection of the hypothesis that amounts returned are equal to zero. Thus the tests do not reject the predictions of the self-regarding preferences model with data from treatments B and C.... In PAGE 18: ... Only two of the subjects exhibited altruistic motivation by sending positive amounts in treatment B. The last row of Table3 reports tests of the hypothesis that amounts sent in treatment B are greater than or equal to zero. Not surprisingly, the tests imply rejection of the hypothesis that amounts sent are non-negative.... ..."

### Table 2: Predictions for an Exponential Distribution

1979

"... In PAGE 25: ... Though we do not normally expect exponentially-distributed task sizes in PPF, the distri- bution is easily analysed. Table2 suggests that the metrics are slightly less accurate for an exponential distribution than a normal distribution.... ..."

Cited by 1

### Table 5: Extreme tail probability using the empirical data distribution, (fully) Bayesian (FB) predictive distribution and approximate Bayesian (AB) predictive distribution

2004

Cited by 1

### Table 3.3: Kullback{Leibler Divergence for the predictive distribution.

1996

### Table 4 Predictive distribution of vector of interest t-GARCH model

"... In PAGE 24: ...0051 0.0485 Table4 - True: NLR // Estimated: NLR T=50 T=100 Misspecification Test Mean %reject(.05) Mean % reject (.... ..."

### Table 8 Post-predictive distribution of vector of interest GARCH model

"... In PAGE 26: ...xamining the resulting data plots gives results nearly identical to those in Figs. 1-3. Thus, the PR modeler is likely to conclude that a NLR with a trend might be the appropriate model. The results in Table 7 and Table8 indicate clearly that this model is indeed statistically adequate and any inferences based on it will be reliable. The one exception is the estimate of R2 which is still misleading as computer programs always take deviations from a constant mean for yt when estimating Var(yt) (see McGuirk, et al (1993)).... In PAGE 27: ...0056 0.0489 Table8 - True: NLR with trend // Estimated: NLR with trend T=50 T=100 Misspecification Test Statistic %reject(.05) Statistic %reject(.... ..."

### Table 6: Kuiper test of the predicted return distribution

2001

"... In PAGE 19: ... We use the Kuiper test because it is more sensitive to the tails that are particularly important for risk management. The test results on a 5% level can be found in Table6 . As for the FOEL test, constant volatility and the normal distribution are mostly rejected, whereas any of the real stochastic volatility models seems to perform well in con- junction with the hyperbolic distribution.... ..."

Cited by 3

### Table 4: Prediction of models: distributions and accuracies

"... In PAGE 7: ...onditions on orderings. Nevertheless, ordering overcame much of the don apos;t-care value problem. Another modeling activity involved a detailed examination of the ability of rules induced to classify the training examples after removing the category attribute (also called a resubstitution test). When the DP as in owchart (a) (see Figure 1) was used in generating the examples and enforcing some ordering constraints on attributes, CN2 created an ordered set of rules that when used to classify the examples gave the results in Table4 . (CN2 can create an ordered set of rules that must be executed sequentially or an unordered set which can be run in parallel.... ..."

### Table 2. Distribution of prediction types for the two models

"... In PAGE 8: ... We have then 4 cases to analyse: PFTK and SQRT in both the underestimation and then overestimation case. Table2 shows the Table 2. Distribution of prediction types for the two models... ..."