whats Randy talking about?

[chris&barb]

i was reading through this old thread and dont understand what Randy is talking about or how he came up with this, http://www.reefcentral.com/forums/showthread.php?s=&threadid=190787&highlight=crystal+sea+marine+mix

the thread and the topic are not important. its this quote, what it means and how he found/figured it out that i want to know.

When I do a t-test on those two results in Ron's article, I get a p of 0.36. That means there is a 36% chance that the apparent difference between the two is a random event (and that there is no actual difference). Most folks like to see p less than 0.05 (less than a 5% chance of random differences explaning the results) before claiming any significance. I know that Ron agrees with this.

what is a "t-test" what is "p" how do you preform this, or where can you find this info on research papers?

im particularly interested in finding out this(t-test or p) from this paper http://links.jstor.org/sici?sici=0006-3185(198110)161:2<213:SDAAOZ>2.0.CO;2-N&size=LARGE

if i can i would like to find out about the mortality differences between the two groups and what that really means.

 

[mcvicker]

Sorry sent this too early. There are several ways to look at data statistically. One very good test is how similar to things are to each other such as a correlation. The t test shows the probability that there is a correlation.
-chris


http://www.statsoft.com/textbook/stbasic.html

 

[bc_slc]

the "t-test" is shorthand for a "Students t-test," named after a guy named Student (not after all those people studying). The principle of the t-test is to look at two groups and decide if the average between them is statistically different. The "p" value is a variable that gets calculated from the t-test which is used to tell you how "significant" the difference between the two groups are - or to phrase it another way - the p value tells you the likelihood that chance alone is to blame for the differences seen in the group.

For example if you looked at 100 people who broke their hip and 100 people who broke their arm, the t-test could tell you if the average age of the people in each group is statistically different between two groups. If you ran the t-test on the ages of the people in these two gorups and the p value was 0.86, then the difference is not statistically significant or another way - "chance alone" may explain the difference between the two groups.

However, if everyone that breaks their arm is a young kid on a trampoline and everyone that breaks their leg is an old lady and there is a big difference between their ages, then the p value would likely be <0.05....the cutoff set by most researches. This value means that there is only a 5% chance that chance alone can explain your findings. A p value of 0.001 would tell you that the findings being attributed only to chance, are very very low and your findings are highly significant (actually 1 in 1000 due to chance alone.

Mat

 

[chris&barb]

thanks for the replies. i will read through the link (for the next year ) and try to get a better understanding of statistics.

Mat, i see what you are saying. you gave a very nice explanation.

what i dont understand is how he can come up with a mathematical equation by looking at group A compared to group B, and determine that X is of no significance?

 

[Boomer]

Thanks Mat

Chris, it is a statics, something I'm not big on. Maybe Mat can help you some more