You're making the same mistake in reverse. The p-value is not a marker of correctness, but of confidence. If something doesn't reach the p-value cutoff, that do

You're making the same mistake in reverse. The p-value is not a marker of correctness, but of confidence. If something doesn't reach the p-value cutoff, that doesn't mean that the effect doesn't exist, just that it didn't meet the confidence level. A paper could be entirely correct in its hypothesis, yet not meet the confidence level in its experimentation.