How to choose null hypothesis vs alternative hypothesis

How we can can chose bw null hypothesis and alternative hypothesis? what I understand from the text is that alternative hypothesis is what we want to check and make true like for example if we are testing risk should be less than some value then alternative hypothesis is less than condition and null would be greater than equal to. Am I correct or not? Please explain

yep

elcfa sorry to ask one stupid question have you done CFA or you are also appearing for exam [:P]

Done I and II, just having some spare time a few days so helping you guys out while I still can, with whatever I still remember.

thanks a lot because I personally felt that whatever explanation you was giving for any of query here was quite simple yet wonderful. Thats why I asked your exam status , well one more thing are you in any job now or preparing for level 3 June.

btw job and will sit on level 3 in June

Just want to make sure you don’t miss out on an easy question. Your goal in hypothesis testing is to REJECT THE NULL hypothesis. If they ask that, this is your answer, not accepting the alternative hypothesis. So if you want to check if something is statistically significant, your null hypothesis is that it is NOT. So the NULL hypothesis is that it’s NOT significant and your goal is to reject that. If you can reject it based on your analysis, you reject the null and have proven that this ‘something’ is statistically significant. Please note that this is the only way to approach it. rejecting the null is what you want. Hopefully this is clear. You mentioned that the alternative hypothesis is what you want to check which may be correct on itself but it’s the wrong way to approach it. If they ask what the goal is of hypothesis testing, saying that you want to check the alternative is the WRONG answer. Not because it’s not true, but because the goal is to reject the null.

ok thanks a lot now I am clear with this thing. One more related thing which I want to clarify from you guys is level of significance, suppose in the question they have asked to check the hypothesis for some level of significance. now if hypothesis is equal to vs not equal to then they chose the critical point dividing leve of significance by 2 and both on +ve and -ve side. But if condition is greater than or less than then sometime they chose critical point based on level of significance and sometime based on 1-level of significance. Can you tell me how to decide it???

please some1 reply to my query!!! I am doing practice for these question rite now

If it is one-tailed hypothesis test (e.g. greater and equal,) you only look up the value, not divide by two. Best with some examples. 1.65 is critical value for one tailed test at 5% level, while 1.96 critical value for two tailed test at 5% level (and critical value for one tailed test at 2.5% level) similarly for one tailed Z0.01=2.33, NOT 2.58 (it is for two level 1%)

The whole reasoning behind this stuff is that you CAN’T really use statistics to prove anything. You can only use it to SUPPORT one viewpoint or another. So what you are trying to support is usually the alternative hypothesis and you are expecting to reject the null. Make sense?

elcfa I am clear with this diving rule what doubt I have in my mind is that for z-test they check for 1-level of significance value in the Z table and for other test t,F they simple just take the value of level of significance. M i correct???

Not sure what you mean. At level I, relevant F tests can be two tailed e.g., equality of variance test where you have to divide significance level by 2 before looking up if you want to test two tailed. (At level II, F test is one - tailed ONLY for the test of at least ONE of the slope coefficients are <>0, but we are getting ahead of ourself). NB: you need to be careful about which table you are looking at. There are FOUR F-test tables at the back first one for two tailed 10% (one tailed 5%), second one for two tailed 5% (one tailed 2.5%) ,…