Once again, have you taken a statistics class before?
Clearly you do not understand how a test is performed. You don't test populations, you test sample sizes (of size n) because measuring entire populations is virtually impossible or impractical. Allow me to explain the most rudimentary concept of statistics you can possibly learn in a statistics class:
1. You need to test something.
2. You determine what population it applies to.
3. You acquire a sample from the population. This sample must not be very small, or else you increase the chances of the data and results being inaccurate.
4. You have a control group and a "treatment" group (the "treatment" group can be named anything, it is just the group that you are testing. The control group is a group not affected by your test, it is just something to compare your "treatment" group with after the test is performed.)
5. You randomly (ideally, simple random sample technique) assign members of the sample size to the two groups.
6. Perform the test.
7. Get p-value.
8. If p-value is less than determined alpha (as I posted earlier, 10% is the highest alpha), you can then assume the event is not due to chance, therefore it is statistically significant.
The most rudimentary concept in statistics. If you still do not understand it after taking a statistics course, well, good luck to you.
It is. The part that he bolded is the part that I quoted and linked to the wiki article (post #740, this thread), before he posted. I posted it above him and he acts like I never seen it. Bizarre.
Anyway, I know 5% is just the "conventional" or common level of significance. But doesn't that make it more appropriate in context?
Last edited by ecofarm; 12-22-11 at 03:32 PM.
And despite your apparent inability to read your own link, it says nothign at all about 5% of a population being statistically significant, which is what your claim was when you said:
Statistical significance is not a measure of a total population. It means that a research finding is not likely to have occurred by chance. It does not mean "5%", as you claimed.
What the 5% (or whateverp-valueAlpha that is chosen) actually means is that there is a 95% certainty that the results were not due to random chance. 5% of the time, though, the results are actually just due to random chance.
When you called "color-blindness" statistically significant, you were using the term incorrectly.
Editted to correct terminology error.
Last edited by Tucker Case; 12-22-11 at 03:45 PM.
He never said "5% of a population." I'm not sure if he was referring to a study with the color-blindness, but whether he was or not I'm pretty sure it was assumed that any data or hypothetical reasoning was from a sample, not a population.
This is what I first saw, which (at least in statistics) doesn't make much sense. Especially because 10% on each side of a normal curve is almost always the maximum any statistician would use in determining if something occurred by chance (normally) or due to something else.
I should say that you were right about "5% of a population" being inaccurate and that I made the mistake of thinking you did not understand what you were saying. My bad!
To be fair, ecofarm never said 5% of the population. That was Tucker Case. I don't believe that ecofarm meant 5% of the population, instead it meant you could prove the hypothesis that "partial (especially minor) color blindness isn't abnormal" if you used 5% as your level of significance.
Either way, you are not arguing the same thing. When something falls within two standard deviations (or 95%) it is considered normal. That is different than saying 95% of the population. Tucker Case argued that using the two standard deviations as a model, then color blindness would be considered normal, which is inaccurate, unless the statistical analysis was done on the entire population. I would wager that if you did a statistical analysis on a group (you choose the size), that color blindness would not fall within the two standard deviations.
Hail to the King baby!
Ecocentrism . Democratic Peace . Ecofeminism . Deep Ecology
Dube . Tosh . Marley