p reported in the output as .000
Locked
p reported in the output as .000
 
|
|
Re: p reported in the output as .000
|
|
Tony, I’ve not heard that before. I’d be curious to know where you heard this. I would argue that we know for sure that it’s less than .001, but beyond that, we don’t know what it’s exact value is. I don’t
have an APA version 6 manual near me at the moment, but I know that numerous websites discussing how to handle this say that you give an exact value down to .000 where it should be written as <.001. My understanding was that originally it was not clear how
to handle this, and this was the later suggestion by the APA. Many journals still prefer the 5th edition recommendation over the 6th edition change for the reasons I mentioned in my earlier email.
It should be noted that if the reported statistics are part of a large results table, 6th edition still suggests going with the *’s (†,
*, **, ***) to represent the p values, as it makes for a less messy table. My interpretation of this would be that, in the results table, if having a separate probability column made things too busy, then revert to the 5th edition form (in my own
work, this is almost always the case). However, within your results writeup, you would likely give exact p values in the text, but only down to .001, where it then reverts to P<.001. Matthew J Poes Research Data Specialist Center for Prevention Research and Development University of Illinois From: SPSSX(r) Discussion [mailto:[hidden email]]
On Behalf Of Anthony Babinec My understanding is that this should be interpreted as < .0005. Tony Babinec |
Re: p reported in the output as .000
|
|
As a check, run the following small job: data list / y(f6.5). begin data. .00010 .00020 .00030 .00040 .00049 .00050 .00051 end data. formats y( f6.3). list var y. .0005 or .00051 rounds up to .001. Tony Babinec From: SPSSX(r) Discussion [mailto:[hidden email]] On Behalf Of Poes, Matthew Joseph Tony, I’ve not heard that before. I’d be curious to know where you heard this. I would argue that we know for sure that it’s less than .001, but beyond that, we don’t know what it’s exact value is. I don’t have an APA version 6 manual near me at the moment, but I know that numerous websites discussing how to handle this say that you give an exact value down to .000 where it should be written as <.001. My understanding was that originally it was not clear how to handle this, and this was the later suggestion by the APA. Many journals still prefer the 5th edition recommendation over the 6th edition change for the reasons I mentioned in my earlier email. It should be noted that if the reported statistics are part of a large results table, 6th edition still suggests going with the *’s (†, *, **, ***) to represent the p values, as it makes for a less messy table. My interpretation of this would be that, in the results table, if having a separate probability column made things too busy, then revert to the 5th edition form (in my own work, this is almost always the case). However, within your results writeup, you would likely give exact p values in the text, but only down to .001, where it then reverts to P<.001. Matthew J Poes Research Data Specialist Center for Prevention Research and Development University of Illinois |
Re: p reported in the output as .000
|
|
Sorry if I wasn’t clear, That’s not what I was referring to. Of course you are right that the value is less than .005. I was asking where you heard that you report it that way. My understanding was that the
reporting recommendations for peer reviewed journals indicated not reporting to 4 decimal places. APA requests that it be reported as P<.001, Federal Education Department developed their recommendations for the Education literature through What Works Clearing
house using a revised version of the APA 5th edition suggestions, etc. In other words, you might interpret it as such (though I’d stand by that this is completely unimportant), but you would never report it as such in any social science peer reviewed
outlet that I know of (I’ve also been involved in public health and political science, and I never saw guidelines suggesting anything different from what I said earlier).
Matthew J Poes Research Data Specialist Center for Prevention Research and Development University of Illinois From: SPSSX(r) Discussion [mailto:[hidden email]]
On Behalf Of Anthony Babinec As a check, run the following small job: data list / y(f6.5). begin data. .00010 .00020 .00030 .00040 .00049 .00050 .00051 end data. formats y( f6.3). list var y. .0005 or .00051 rounds up to .001. Tony Babinec From: SPSSX(r) Discussion
[hidden email]
On Behalf Of Poes, Matthew Joseph Tony, I’ve not heard that before. I’d be curious to know where you heard this. I would argue that we know for sure that it’s less than .001, but beyond that, we don’t know what it’s exact value is. I don’t
have an APA version 6 manual near me at the moment, but I know that numerous websites discussing how to handle this say that you give an exact value down to .000 where it should be written as <.001. My understanding was that originally it was not clear how
to handle this, and this was the later suggestion by the APA. Many journals still prefer the 5th edition recommendation over the 6th edition change for the reasons I mentioned in my earlier email.
It should be noted that if the reported statistics are part of a large results table, 6th edition still suggests going with the *’s (†, *, **, ***) to represent the p values, as it makes for
a less messy table. My interpretation of this would be that, in the results table, if having a separate probability column made things too busy, then revert to the 5th edition form (in my own work, this is almost always the case). However, within
your results writeup, you would likely give exact p values in the text, but only down to .001, where it then reverts to P<.001. Matthew J Poes Research Data Specialist Center for Prevention Research and Development University of Illinois |
Re: p reported in the output as .000
|
In reply to this post by Anthony Babinec
When I see p's so extreme, I take that as a flag that there may be an
error in that analysis. I'd go back and check things. What do boxplots look like? Is there that much visual difference? Is the comparison so obvious that there was to reason to do the study? E.g., the weight of the average 5 year old differs from the weight of the average 10 year old? Although mathematically p's can be very small, I do not believe that most readers would be able to differentiate beyond .0005 or even .001. YMMV. Art Kendall Social Research Consultants On 1/11/2012 12:49 PM, Anthony Babinec wrote: > > As a check, run the following small job: > > data list / y(f6.5). > > begin data. > > .00010 > > .00020 > > .00030 > > .00040 > > .00049 > > .00050 > > .00051 > > end data. > > formats y( f6.3). > > list var y. > > .0005 or .00051 rounds up to .001. > > Tony Babinec > > [hidden email] <mailto:[hidden email]> > > *From:*SPSSX(r) Discussion [mailto:[hidden email]] *On > Behalf Of *Poes, Matthew Joseph > *Sent:* Wednesday, January 11, 2012 11:20 AM > *To:* [hidden email] > *Subject:* Re: p reported in the output as .000 > > Tony, I’ve not heard that before. I’d be curious to know where you > heard this. I would argue that we know for sure that it’s less than > .001, but beyond that, we don’t know what it’s exact value is. I > don’t have an APA version 6 manual near me at the moment, but I know > that numerous websites discussing how to handle this say that you give > an exact value down to .000 where it should be written as <.001. My > understanding was that originally it was not clear how to handle this, > and this was the later suggestion by the APA. Many journals still > prefer the 5^th edition recommendation over the 6^th edition change > for the reasons I mentioned in my earlier email. > > It should be noted that if the reported statistics are part of a large > results table, 6^th edition still suggests going with the **’s (*†, *, > **, ***) to represent the p values, as it makes for a less messy > table. My interpretation of this would be that, in the results table, > if having a separate probability column made things too busy, then > revert to the 5^th edition form (in my own work, this is almost always > the case). However, within your results writeup, you would likely > give exact p values in the text, but only down to .001, where it then > reverts to P<.001. > > *Matthew J Poes* > > Research Data Specialist > > Center for Prevention Research and Development > > University of Illinois > ===================== To manage your subscription to SPSSX-L, send a message to [hidden email] (not to SPSSX-L), with no body text except the command. To leave the list, send the command SIGNOFF SPSSX-L For a list of commands to manage subscriptions, send the command INFO REFCARD
Art Kendall
Social Research Consultants |
| Free forum by Nabble | Edit this page |