Type 1(i) versus Type 2(ii) Error Memory Aid

It may sound a bit silly but I remember is this way…I think of a court of law and say the worst error they could make is finding an innocent man guilty. In court it is “innocent until proven guilty”, so the null hypothesis is innocence.Finding an innocent man guilty is the same as “rejecting the null hypothesis. I then relate the number 1 to “worst” therefore resulting in a type 1 error of rejecting the null when true…..I.e. Finding an innocent man guilty.

One person told me they put the definition on their smartphone background/screensaver so they’d see it over and over.  I posted a note in front of me at my desk and repeated it daily.  Going to see it at L2 in conjunction with regression testing: ANOVA, t- stats and SE.

This is kind of a stupid mnemonic, but I remembered the difference by misspelling “turn”–“TIRN”–> Type I Rejects Null

We only use 1 significance level, so there’s only 1 alpha, which is the probability of a type 1 error. The bigger the alpha is, the bigger is our rejection region, which leads to a greater probability of a type 1 error, which is rejecting the null hypothesis, when it’s actually true. Mind you, we can test at difference significance levels, i.e 5% or 1%, so what falls between the region of 5% and 1% would mean a rejection of the null at the 5% level, and a failure to reject at the 1% level. But we’re often supplied with one significance level, so that’s a way to have the concept stick.

Type 1 error, corresponds with 1 significance level. Alpha has a positive relationship with type 1, and an inverse relationship with type II.

Sorry that should have read “finding an innocent man guilty is the same as rejecting the null hypothesis when it is true”

@corytuck I had issues with this as well. I just filed this as one of the last-minute things to rote-momorize right before I go into the exams. I did try and come up with memory aids (as you’re thinking of) but ended up confusing myself more.

i don’t know how helpful this will be, but i remember the difference by thinking of type II errors as the more complicated phrase. failing to reject when false is more complicated (in my mind) then reject when actually true