T is a statistical value, because the characteristic of t distribution is that the farther away from the far point, the less likely it is to get this value. In regression analysis, our test hypothesis is "the coefficient of X =0 (at this time, X has nothing to do with Y)", so the larger the value of T, the better, because it is less likely to happen, so the relationship between X and Y.

Generally, the p value corresponding to the t value is a bilateral test in the report of univariate regression: that is to say, in the test of your regression, it is more likely that the value of t distribution is greater than the statistical value (plus absolute value) of t calculated by you. If the value of p is large, it means that the value of t is close to the origin, while the value of p is small, which means that the value of t is far from the origin (the greater the absolute value of t, the smaller the value of p). According to the above analysis, the smaller the p value, the better.