The Internal Security Act and Hypothesis Testing

According to our Minister for Home Affairs, Teo Chee Hean, "more than 800 people were arrested under the Internal Security Act (ISA) in the 1970s of whom 235 were issued with Orders of Detention." 

While there have been many people speaking up against the ISA, I am more neutral on this. I understand why the ISA is necessary - to enable the Government to act quickly to preserve public order without the hassle of going through a trial. Lee Kuan Yew pointed out before that the absence of the ISA is akin to "go(ing) through the motions of a trial and let a guilty man off to continue his damage to society". Of course, in the hands of the wrong person (or Government), this could lead to a potential abuse of power where innocents can be seized just for opposing the Government.

It is interesting to note that such a large number of people were detained without trial. It is safe to assume that a percentage of these people are actually innocent, which, in statistics, is indicative of a Type I error. In this case, we can actually set up a hypothesis with H₀ being the null (or default) hypothesis and H₁ being the alternative hypothesis.

H₀: The person is innocent.

H₁: The person is guilty.

This is where statistical error comes in - a Type I or Type II error. A Type I error is when H₀ is rejected when it is actually true - i.e, the person is innocent but still convicted, whereas a Type II error occurs when H₀ is accepted when it is false - i.e, the person is declared innocent but is actually guilty. In this case, a Type I error is the most applicable, because it is quite possible that a significant number of the people seized are, in reality, innocent. An example would be the people closely related to the suspects that were seized simply because, well, they are closed related.

The level of significance, usually denoted by the symbol alpha (α), refers to the probability of a Type I error. In statistical problems involving hypothesis testing, the student is usually given the level of significance (which is usually 1% or 5%) and asked to test if H₀ should be rejected or not.

Of course, if we were to use the above situation in hypothesis testing, I think the level of significance would be significantly higher than 5%.