Is this question based on inspection bias? This table describes the positive corona tests in all three open ports of country X, in two different days, plus for the total positive for all people entering the country via those three ports:
Day 2     Day 1      Port
1.7%      1.4%        A
0.7%      0.4%        B
0.9%      0.2%        C
-------------------------
1.2%      1.3%   Total entering the country

How is it possible that although the proportion of the corona positive tests, for each port, were higher in Day 2, but Day 1 has a higher proportion in the total!
What paradox is this?
 A: Consider the following scenario:
On Day 1, the ports of entry A, B, and C see 40000, 2000, and 2000 entries respectively. With a 1.4%, 0.4%, and 0.2% positive test rate in each of the ports, you will see 560, 8, and 4 positive tests. This yields a total of 572 positive tests, or 1.3% of 44000 entries in total.
On Day 2, the ports of entry A, B, and C now see 16000, 10000, 10000 entries respectively. With a 1.7%, 0.7%, 0.9% positive test rate in each of the ports, you will see 272, 70, and 90 positive tests. This yields a total of 432 positive tests, or 1.2% of 36000 entries in total.
This phenomenon is an example of Simpson's paradox, where (quoting Wikipedia) "a trend appears in several groups of data but disappears or reverses when the groups are combined."
Usually, the paradox arises due to unaddressed confounding variables (e.g. it is a public holiday in Country Y on Day 2 and people from that country prefer using Ports B and C) and assumptions made from early patterns that turn out to be not true (e.g. entries via a given port remain largely constant over days).
A: It is because the number of tested people can be different.
Imagine that on Day 2, 99% of your tests were done in Port C, and on Day 1, 99% of your tests in Port A. Then you would have on Day 2 roughly 0.9% and on Day 1 roughly 1.4%, which would even bee more "strange" than your actual data.
