Could you tell me how independent samples t-test could show random assignment worked well?

^^

I see. Thank you.

Personally, I agree that randomization helps to produce comparable treatment and control groups. However, the random assignment process doesn’t guarantee similarity. Because the process is random, we have little information about the similarity until the process finishes, especially in simple randomization. The treatment and the control groups may present differences in certain variables even after the random assignment. But as long as those variables are not correlated with the outcome, the differences won’t induce selection bias in theory. Therefore, we can conclude that treatment and control groups are comparable when results from the t-test show no statistically significant differences in all variables. But we cannot claim the failure of the random assignment only because we find significant differences through the t-test.