Home > Statistics Every Writer Should Know > The Stats Board > Discusssion
level of significance
Message posted by wasco on February 7, 2001 at 12:00 AM (ET)
If you define alpha as the level of significance and you are looking for the z statistic. the problem is if alpha is .05, then is z = 1.645 or 1.96. does it depend on a one sided or two sided test?
READERS RESPOND:
(In chronological order. Most recent at the bottom.)
Re: level of significance
Message posted by nancy diehl on February 8, 2001 at 12:00 AM (ET)
Yes, if you have a one-sided test, then all the alpha value (e.g. .05) goes in
one tail of the distribution, hence the Z value relating to 5% of the area in
the tail of a normal distribution is 1.645. If you have a two tail test, then
you need to split the .05 value in half; .025 in each tail. The Z-value
associated with .025 area under the normal curve in the tail of the distribution
is 1.96.
Re: level of significance
Message posted by dkayombo (via 64.86.50.240) on June 13, 2001 at 7:55 AM (ET)
what is the level of significance
Re: level of significance
Message posted by JG (via 128.8.22.17) on June 16, 2001 at 5:11 AM (ET)
Level of significance is the smallest alpha value that will allow you to reject the hypothesis or the largest alpha value that will allow you to accept the hypothesis. That is, if you wish to accept the hypothesis give your data, then level of significance is the smallest alpha value that you can have. The level of significance can also be thought of as how significant your data is as far as allowing you to reject the hypothesis on the basis of the available data. A large level of significance means that if you reject the hypothesis on the basis of the present data you will have a large probability - alpha value - of being wrong. This is equivalent to telling a judge - if you convict this person on the basis of the available evidence then there is the following probability - level of significance - that you will convict an innocent person. Another way of saying this is - a large level of significance means a high probability that on the basis of available evidence the accused is actually innocent.
Re: level of significance
Message posted by Ruby (via 131.247.66.57) on July 16, 2001 at 4:57 PM (ET)
I did a sample test which provided a printout for testing whether the mean low stock price of NYSE stocks differs from $30 and whether the mean high stock price of NASDAQ stocks is greater than $40. For both of these--what will the conclusion be if I used a level of significance(alpha) of 0.05 for this test?
Re: level of significance
Message posted by JG (via 128.8.22.156) on July 16, 2001 at 8:56 PM (ET)
Any test either accepts or rejects a hypothesis. Alpha=.o5 means that 5% of the time you will reject a hypothesis that should have been rejected.
Your $5 contribution helps cover part the $500 annual cost of keeping this site online.
