Anyone give a little statistics help? (to my girlfriend)
Any stats experts round here? My girlfriend is doing her final project for her degree and stats isn't her strongpoint - and it certainly isn't mine I'll pass her over now to ask what she wants to know:
Hello there! sorry, this stuff totally eludes me...i know my coefficient is 0.420 but how do i know if it's statistically significant? i thought i had to compare the test score (t=1.308) with the crit value (.05, 1-tailed = 1.860) and if this crit value was equal or less than t it was significant. but i've been reading somewhere else and it says i have to compare my coefficient to a crit value table of coefficients to find out if it is stat. sig...also, what do i do with the p(1-tailed) number of 0.114.and what does it mean?? i'm so confused with all this, my head's all mashed up please help!
TEST SCORES
Subject Number Phoneme Segmentation Reading
1 7 18
2 4 9
3 4 6
4 3 24
5 4 14
6 7 39
7 12 20
8 5 18
9 4 13
10 4 10
Results of Spearman's rank order correlation
rs (corrected for ties) 0.420
Number of data pairs 10
Number of tied X ranks 2
Number of tied Y ranks 1
Degrees of freedom 8
t 1.308
p(1-tailed) 0.114
p(2-tailed) 0.227
tcrit (.05, 1-tailed) 1.860
tcrit (.01, 1-tailed) 2.896
tcrit (.05, 2-tailed) 2.306
tcrit (.01, 2-tailed) 3.355
[edited by - Starboy on June 6, 2004 6:02:43 AM]
quote:Original post by Starboy
i know my coefficient is 0.420 but how do i know if it''s statistically significant? i thought i had to compare the test score (t=1.308)
coefficient of I don''t know what, but assuming your t value is correct there, you need two things to tell if its statistically significant.
1. alpha value (looks like you''re using .05)
2. degrees of freedom for the t test
df is usually n (number of samples) -1 except is cases where you''re doing stranger things (linear regression tests, for example, which are n-2)
Now, I don''t know this critical value method, or how it could work with different degrees of freedom, but a simple calculation on a TI83 will give you a P value, which if it is less that alpha (.05) means its significant.
tcdf(1.308,999999,df) Gives you a one tailed P value.
CLAIRE: Hmmm, I''m still a little confused.
So, if alpha is .05, and the number of samples is 10, what''s the result? Is this statistically significant?
So, if alpha is .05, and the number of samples is 10, what''s the result? Is this statistically significant?
quote:Original post by Eelco
cant help you, but try the math & physics forum instead of the lounge.
Yeah, agreed. Moving it for you now.
From what I remember of A-level statistics (which isn''t much), the test score being below the critical value should be enough to demonstrate a significant result.
You have two hypothesis - the one you''re actually looking to prove ("Hone: Earphone sales are proportional to number of cars along the main street per hour"), and the ''null hypothesis,'' which is basically the opposite ("Hnull: Earphone sales are not proportional to number of cars along the main street per hour.").
You start by assuming that Hnull is true, and then find out what the chances of obtaining your result are under that circumstance. If the chance is below 5%, it''s a significant result.
As I say, though, my memory of this is very hazy. So please don''t trust it 100%.
Closing the thread. Homework/schoolwork posts are off topic. There are websites dedicated to homework help, and you should ask the question there. Review the Forum FAQ and the following thread:
Thread on Homework Policy
Graham Rhodes
Principal Scientist
Applied Research Associates, Inc.
Thread on Homework Policy
Graham Rhodes
Principal Scientist
Applied Research Associates, Inc.
This topic is closed to new replies.
Advertisement
Popular Topics
Advertisement