Whats the statistical aproach to knowing this?. I have heard its very comon for scientist the need of this , and It would be perfect in my own experiment.
2007-01-05 11:26:53 · 3 answers · asked by vatsq 2 in Science & Mathematics ➔ Other - Science
One of my favorite questions - there's not a simple answer.
We use t tests and F tests to tell the "probability" that the experimental results are due to the effects we hope they are (our hypothesis) or not!
But "truth" is a bit strong a word!
2007-01-05 11:37:32 · answer #1 · answered by bubsir 4 · 0⤊ 0⤋
It's called hypothesis testing. What you do is state a null hypothesis, an alternate hypothesis, a test procedure, a confidence level for testing the hypothesis, and then do the statistical test to test the validity of the null hypothesis. For example:
Null..H0: It's raining outside.
Alternate...H1: It's NOT raining outside.
Confidence Level: I want to be 95% confident (pretty certain) in my results.
Test: Stick my hand out the window, it comes back dry. I am fairly certain (95% confident) that "it's NOT raining outside."
The null hypothesis was not proved in this example; so, being the logical opposite, the alternate hypothesis "It's NOT raining outside" has reasonable assurance of not being false. There could be some error in my results; for example, suppose there was an awning above the window to stop the rain from falling on my outreached hand? But at 95% confidence, the chances for that error are minimized.
The test procedures in statistics are quite varied and sometimes quite complex. So we can't cover them here. You'd need to get an easy reading book like "Statistics for Dummies" to learn hypothesis testing.
FYI, your phrase "statistically random or true" is not a valid statement. Your data can be both statistically random and show your null hypothesis is true (NOT false). Random does not mean false in any logical sense. In fact, random variables are used to do hypothesis testing and the hypotheses might very well be true.
2007-01-05 19:54:01 · answer #2 · answered by oldprof 7 · 0⤊ 0⤋
Do you know much math? You really should study statistics instead of asking us. A serious experimenter wouldn't try to find that answer by asking here!! LOL
2007-01-05 19:37:46 · answer #3 · answered by wasabi_luvva 2 · 0⤊ 1⤋