stat wars 2010

May aattend ba?! Kelan to? Go beat 'em StatSoc!

1-tailed or 2-tailed

Nakakatuwa talaga pag ang mga geeks eh nagdedebate. Labasan ng mga evidentiary (may term bang ganito?) supports! Todo research agad sa internet at labas ng mga nagkakapalang libro. At pag hindi pa rin nagkasundo eh tawag na sa lifeline. Thanks ulit Wendz! Anyways, the question is: ano ba mas strict? 1-tailed or 2-tailed tests?! Thanks Pepsi for giving us the chance to review our rusty statistical test theories. LOL

So ano nga bang sagot? 2-tailed tests are stricter. Why? According to Sir Wendz, 1-tailed tests are just directional. In order to reject the null hypothesis, we have to prove just one scenario: either mu1 is greater than or less than mu2. But in 2-tailed tests, you have to prove both. You have to prove that mu1 is less than AND at the same time greater than mu2. By then it's safe to say that both means are statistically different. Hmm. Tama ba Wendz? Patay baka ibalik ako sa Stat115 nito haha.

If you want to test the difference of two means or proportions at 95% level of significance using 1-tailed, then  just test it at 90% level of significance at 2-tailed. Same goes if you want to test parity of two means using 1-tailed test, then test it at 80% level of significance using 2-tailed.

There you go, my very first Stat-is-Geek entry. Will try to post more. Maybe about survey and sampling so that presidentiables like Gordon can shut the hell up! Hindi ako galit. Hindi po talaga!. LOL