Wed, Sep 05, 2018 - Page 8

Not so suspicious?

I was stopped by a sentence in a recent article attributed to an unnamed source of the Central Election Commission (CEC): “While checking the data for the first-threshold signatures, household registration offices also found that about 1 percent of the signatures in each of the three proposals belonged to people who are dead ... due to the large number of cases it is unlikely that those people passed away after signing the proposal” (“CEC finds irregularities in KMT proposals’ signatures,” Sept. 1, page 1).

What struck me was that it must take time to collect signatures from 1.45 million people (about 7.5 percent of the population over 17 years of age), and mortality in Taiwan reaches the 1 percent per year level at about age 65.

Using Ministry of the Interior data available on the Internet, the 2016 population of Taiwan (based on household registration data) and the abbreviated life table for the year 2014, and assuming that Chinese Nationalist Party (KMT) members have the same age distribution as the general population, the average number who would have died in one year amounts to 0.6 percent of the population of over 18.

If we make allowance for aging of the population between 2016 and 2018 and for KMT members being on average three years older than the general population, that would rise to more than 1 percent. For every additional year of difference between the KMT and general populations, the number increases by about 13 percent.

Allowing a difference of five years would bring the expected annual death rate to about 1.5 percent.

If the canvassing paid more attention to old people than young people at the beginning of the campaign, that could easily have raised the expected annual deaths to 2 percent of the relevant population.

The reported number of deaths might be surprising if we knew that the campaign took substantially less than one year and that the age distribution of KMT members differs from that of population over 18 by less than five years.

If the span of dates attached to the signatures is less than three months, the observation is probably sound; otherwise maybe the CEC should rethink that part of the argument.

Emilio Venezian