# p value of 0.000 on SPSS

#### viciousorphan

##### New Member
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.

#### svetlanazvyagintsev

##### New Member
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

##### New Member
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 really wanted to know if the pvalue of 0.000 is significant and is not the result of some error on my part while using SPSS. If some body could clarify that it will help.

#### trinker

##### ggplot2orBust
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).

#### Dason

##### Ambassador to the humans
In real life p is never = 0 (Dason will probably come up with some absurd counter case where this does not hold true).
It's true for the most part but... yes you can come up with cases where p is legitimately 0.

#### hlsmith

##### Omega Contributor
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.

#### trinker

##### ggplot2orBust
@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.