If the absolute value of your calculated t-statistic exceeds the critical value from the standard normal distribution, you can reject the assumption that the error terms are homoskedastic. reject the null hypothesis. safely assume that your regression results are significant. conclude that most of the actual values are very close to the regression line.