### Table 2 Empirical rejection rates under the null hypothesis Number of Regressors

### Table 1 Empirical rejection rates under the null hypothesis Number of Regressors

"... In PAGE 38: ... Though it appears that the White form of the statistic is uniquely beset with rejection rates that rise with n, both forms of the statistic actually reveal very similar distributional behavior. Taking a slice from Table1 , Figures 5 and 6 plot histograms of the full IM statistic in models with seven regressors. The vertical lines in the plots indicate where the asymptotic critical value is located, and for reference, the asymptotic chi-square distribution has also been superimposed.... ..."

### Table 6 The empirical size of 5% of tests with the null hypothesis of no cointegration

1997

"... In PAGE 17: ...alue 1.645. show little variation in size across the various values of h and p. Note that some of the results in Table6 have extremely severe size distortions. Pedroni (1995) also observed this; it is probably due to more challenging experimental designs employed in this section.... ..."

### Table 4: GARCH results for interest rate transmission under alternative exchange rate regimes: Developed Countries

"... In PAGE 17: ... Table 3 lists some of the key dates of the included countries with respect to their exchange rate arrangements. Table4 presents the results for the analysis using interest rate levels whereas Table 5 shows the empirical findings for the ECM framework.7 The ADF test reveals that the residuals of all of the models with interest rate levels are stationary, indicating that the results can indeed be interpreted as long run, cointegrating relationships between interest rate levels.... ..."

### Table 3 Empirical rejection rates under the null hypothesis Non-Parametric Bootstrap

"... In PAGE 47: ... This is clearly seen in the upper middle panel of Figure 14. This is responsible for the zero rate of rejection that is observed in Table3 . Raising k to 6 simply exacerbates this problem, and rejection of the null hypothesis is essentially precluded at this point.... ..."

### Table 4 Empirical rejection rates under the null hypothesis Non-Parametric Bootstrap

### Table 1: Empirical distribution of the test statistic C(d) under the null hypothesis of linear vector models with zt?d = y1;t?d and the recursive estimation starts at mo. The results are based on 10,000 replications.

1998

"... In PAGE 7: ... The sample sizes used are n = 150 and 300, and the number of replications is 10,000. Table1 shows the empirical percentiles of the test statistic C(d) and those of the corre- sponding chi-square distributions. In the test, we assume y1;t?d is the threshold variable, d 2 f1; 2; 3; 4g.... In PAGE 27: ...Table1 0: Results of threshold nonlinearity test for Iceland daily river ow data, using three possible threshold variables and three bivariate AR models. The sample size is 1095, the exogenous variables used are daily precipitation lagged 1, 2 and 3 days and daily temperature lagged 0 and 1 day, and the starting point of recursive least squares is 150, where p is the AR order, d is the delay and d:f: stands for degrees of freedom of the asymptotic 2 distribution.... In PAGE 28: ...Table1 1: Conditional least squares estimates and their t-ratios for a selected bivariate two- regime TAR(15) model for the Iceland river ow data. The threshold value is ?0:42394oC and the numbers of observations in each regime are 479 and 601, respectively.... ..."

Cited by 9

### Table 6: Results of hypothesis tests

2002

"... In PAGE 9: ....3. Hypothesis tests Log-linear analysis permits one to analyze categorical data in much the same manner as in analysis of variance. The sampling distribution underlying Table6 is a product of independent multinomials. According to Bishop, Fienberg and Holland, the kernel of the appropriate likelihood function is the same as that for a simple multinomial or a simple Poisson [7].... In PAGE 10: ... Thus, we do not reject the null hypothesis (H0), and hence conclude that there is no interaction between these two factors. For the remaining hypothesis tests, we selected out only the data for a particular pair of criteria (indicated by the column labeled Hypothesis in Table6 ) and then tested for an interaction between these two by fitting the model described with and without the corresponding fault-type/criterion term. Table 6 summarizes the results of these tests.... In PAGE 10: ... For the remaining hypothesis tests, we selected out only the data for a particular pair of criteria (indicated by the column labeled Hypothesis in Table 6) and then tested for an interaction between these two by fitting the model described with and without the corresponding fault-type/criterion term. Table6 summarizes the results of these tests. The column labeled Hypothesis states the null (H0) and alternative hypothesis (H1) for each test.... In PAGE 10: ... 6. Discussion The three hypotheses in Table6 that tested the effectiveness of each coupling-based criteria against Branch Coverage indicate that the coupling criteria are better at detecting the object-oriented faults used in the experiment. A remaining question is which of the three coupling criteria is the most effective.... ..."

Cited by 3

### Table 6: Results of hypothesis tests

2002

"... In PAGE 9: ....3. Hypothesis tests Log-linear analysis permits one to analyze categorical data in much the same manner as in analysis of variance. The sampling distribution underlying Table6 is a product of independent multinomials. According to Bishop, Fienberg and Holland, the kernel of the appropriate likelihood function is the same as that for a simple multinomial or a simple Poisson [7].... In PAGE 10: ... Thus, we do not reject the null hypothesis (H0), and hence conclude that there is no interaction between these two factors. For the remaining hypothesis tests, we selected out only the data for a particular pair of criteria (indicated by the column labeled Hypothesis in Table6 ) and then tested for an interaction between these two by fitting the model described with and without the corresponding fault-type/criterion term. Table 6 summarizes the results of these tests.... In PAGE 10: ... For the remaining hypothesis tests, we selected out only the data for a particular pair of criteria (indicated by the column labeled Hypothesis in Table 6) and then tested for an interaction between these two by fitting the model described with and without the corresponding fault-type/criterion term. Table6 summarizes the results of these tests. The column labeled Hypothesis states the null (H0) and alternative hypothesis (H1) for each test.... In PAGE 10: ... 6. Discussion The three hypotheses in Table6 that tested the effectiveness of each coupling-based criteria against Branch Coverage indicate that the coupling criteria are better at detecting the object-oriented faults used in the experiment. A remaining question is which of the three coupling criteria is the most effective.... ..."

Cited by 3

### Table 3: An Empirical Test for Zoning Fox Chapel

2007

"... In PAGE 30: ...Table3 . We test the null hypothesis that the coefficients on these new terms were jointly zero.... ..."