Doubt about Rigged election in South Korea South Korea’s 21st legislative elections were held on 15 April 2020.
There are two types of early voting systems in Korea. One is to vote to candidate in your own local area. The other is to vote to candidate in other local area where your address is registered but you cannot go. I will call each of this A and B. 
This is the result of early voting in Yeonsu District. 
I brought this data from National Election Commission. 
As you see, if we divide B by A, we have almost same result, 0.39. Not only in this area, but also in other areas, certain numbers appear like 0.29, 0.26 and so on. 
My question is that is it possible from a stochastic point of view? 

 A: If the voters are split randomly between A  and B, it would not be surprising that their voting behaviour be the same and (given the large numbers involved) that the proportions of votes to each party under each regime A  and B were also nearly the same.
However, I would expect that elder people with less mobility from their residence area, for instance, would fall mainly under regime A and youngsters (studying away from their homes, for instance) would fall under regime B. If elder people and youngsters vote differently (as would appear likely), I aggree with you that the result is surprising.
A: I am not very familiar with how election results should look like, e.g how big is the variation and how much it would differ from the null hypothesis contingency table. 
Looking the result, we can do a chi-sq test similar to what you have:
M = matrix(c(15797,6185,11335,4460,5296,2073),ncol=3)
chisq.test(M)

    Pearson's Chi-squared test

data:  M
X-squared = 0.052314, df = 2, p-value = 0.9742

If we ask the probability of get a result as close to the expected, i.e, a X-square lesser than 0.052314, it's 1 - 0.9742 = 0.0258. Normally we would do:
pchisq(0.052314,2)
[1] 0.02581787

However this is only 1 observation / experiment. Ideally you collect such statistics over many local areas and perform the same analysis, and ask if this result is a blip or there are indeed trends.
I can give a well known example, R.A Fisher noticed in Gregor Mendel's experimental data, for many experiments the number of seeds with a certain phenotype matches closely the expected. An exceptionally good fit of data to theory. He tested the probability of getting a chi square lesser than the observed for each experiment Mendel had, and postulated that if they are independent and followed the null hypothesis, the probability of getting an overall better result if all experiments are repeated would be 7/100000. More details about the analysis in this paper
Fisher even proposed:

"Although no explanation can be expected to be satisfactory, it
  remains a possibility among others that Mendel was deceived by some
  assistant who knew too well what was expected. This possibility is
  supported by independent evidence that the data of most, if not all,
  of the experiments have been falsified so as to agree closely with
  Mendel's expectations."

Reason for pointing out the above example is, even Fisher's analysis, it's still widely debated whether Mendel manipulated his data, because there are biological reasons that we still know little of. It goes beyond the statistics.
One cannot easily conclude from analysis of 1 election result that it is rigged. Even if you collect data over multiple areas, there are still many factors one need to consider, and take into account.
