I ve come across some papers listing p value of 0.000 as a result. Well, you can always list it as 0.0001
I am using SPSS to define relationship between two variables. Previous research has indicated a significant relationship between them. Spearman correlation gives a p value of 0.000 and linear regression also gives the same value. What does this mean. For my paper i have set p<0.05 to be significant. Although the obvious conclusion is that it is significant but i am a bit thrown off by the value of 0.000.
Any help will be appreciated.
I ve come across some papers listing p value of 0.000 as a result. Well, you can always list it as 0.0001
viciousorphan (03-10-2013)
If you have lots of power (big effect and/or large n) it's common to get a p = .0000 in SPSS. Caution that this does not mean p = 0. It emans that due to rounding p ~ 0. I think the way to approach this is to say p < (insert your predetermined alpha level in here). In real life p is never = 0 (Dason will probably come up with some absurd counter case where this does not hold true).
"If you torture the data long enough it will eventually confess."
-Ronald Harry Coase -
viciousorphan (03-11-2013)
It is just a very very small p-value, definitely smaller than 0.05. Like they said above it gets rounded to this value. I would put in your paper <0.001, given your reported 0.000.
Another scenario that you can see some times from a program is a p-value: 1.000. I usually list this one as p-value: >0.999 or if the values truly are the exact same you can put 1.000 (e.g., exact same proportions in contingency tables.
Stop cowardice, ban guns!
viciousorphan (03-11-2013)
@hlsmith I agree with all points except the recommendation to use p < .001. The alpha level set by the poster is .05 thus because hypothesis testing answers a yes/no question of significance. It's common practice to have lots of < p values but I really think this violates the intended use. IMO it should be p < alpha or p = (exact value). But that's arguing straws i think.
"If you torture the data long enough it will eventually confess."
-Ronald Harry Coase -
Tweet |