SPSSX Discussion

Level of significance

Classic

List

Threaded

17 messages Options

Humphrey Paulie

Level of significance

Dear all,
When sig.=00.00 in an analysis, what level of significnace can or should we report?
p<.01, p<.001, p<.001? or something else?
Humphrey

Ryan

Re: Level of significance

Humphrey,

In general, if sig.=.00, then you can conclude that p<.005.

Ryan

Humphrey-6 wrote

Dear all,
When sig.=00.00 in an analysis, what level of significnace can or should we report?
p<.01, p<.001, p<.001? or something else?
Humphrey

Granaas, Michael

Re: Level of significance

In reply to this post by Humphrey Paulie

report what you know, if SPSS reports sig = 0.000 then you know that p < 0.001. It may be even smaller, but you cannot know how much smaller.

Alternatively you can reset the output to more decimal places, but I don't see the point.

Michael

****************************************************
Michael Granaas             [hidden email]
Assoc. Prof.                Phone: 605 677 5295
Dept. of Psychology         FAX: 605 677 3195
University of South Dakota
414 E. Clark St.
Vermillion, SD 57069
*****************************************************

From: SPSSX(r) Discussion [[hidden email]] On Behalf Of Humphrey Paulie [[hidden email]]
Sent: Monday, May 17, 2010 4:47 AM
To: [hidden email]
Subject: Level of significance

Dear all,
When sig.=00.00 in an analysis, what level of significnace can or should we report?
p<.01, p<.001, p<.001? or something else?
Humphrey

statisticsdoc

Re: Level of significance

In reply to this post by Humphrey Paulie

if SPSS reports a p value of .000 you can report significance at p < .001. Few publications require lower alpha levels to be reported.
Best
Steve Brand
www.StatisticsDoc.com

=====================
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

Re: Level of significance

In addition to the good advice in the previous post I would like to add:
In my experience few people can really make fine grained distinctions in reading probability statements or in comparing confidence intervals. Also in the social and behavioral sciences reporting too many significant digits in probabilities is a fallacy of precision.

One of my favorite quotes when discussing the fallacy of precision is from Aristotle:
In Ethics I.3, he says, "we must be satisfied to indicate the truth with a rough and general sketch: when the subject and the basis of a discussion consist of matters which hold good
only as a general rule, but not always, the conclusions reached must be of the same order. . . . For a well-schooled man is one who searches for that degree of precision in each kind of study which the nature of the subject at hand admits..."

Art Kendall
Social Research Consultants

On 5/17/2010 11:54 AM, Statisticsdoc Consulting wrote:

if SPSS reports a p value of .000 you can report significance at p < .001. Few publications require lower alpha levels to be reported.
Best
Steve Brand
www.StatisticsDoc.com

=====================
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

===================== 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

Martin Holt

Re: Level of significance

In reply to this post by Ryan

I don't agree with Ryan, below. If p were 0.009, and truncation were in operation, you would get 0.00 yet 0.009 > 0.005.

Martin

From: rblack <[hidden email]>
To: [hidden email]
Sent: Monday, 17 May, 2010 13:37:17
Subject: Re: Level of significance

Humphrey,

In general, if sig.=.00, then you can conclude that p<.005.

Ryan

Humphrey-6 wrote:

>
>
>
> Dear all,
> When sig.=00.00 in an analysis, what level of significnace can or should
> we report?
> p<.01, p<.001, p<.001? or something else?
> Humphrey
>
>
>
>
>
>
>
>
>
>

--
View this message in context: http://old.nabble.com/Level-of-significance-tp28581317p28582789.html
Sent from the SPSSX Discussion mailing list archive at Nabble.com.

=====================
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

Handel, Richard W.

Re: Level of significance

In reply to this post by Art Kendall

Art Kendall wrote:

“Also in the social and behavioral sciences reporting too many significant digits in probabilities is a fallacy of precision.”

This is slightly off topic, but I recently read an interesting article that touches on the practice of reporting too many digits for correlations.

http://orm.sagepub.com/cgi/content/abstract/12/4/687

Cheers,

Rick

Martin Holt

Fw: Level of significance

In reply to this post by Humphrey Paulie

Sorry,

I posted this, but inadvertently not to the group. Hence my later reply.

Not sure I come up to Art's method :)

Martin

----- Forwarded Message ----
From: M HOLT <[hidden email]>
To: Humphrey Paulie <[hidden email]>
Sent: Monday, 17 May, 2010 16:13:11
Subject: Re: Level of significance

Hi Humphrey,

Firstly, there is no need for the double zero at the start. One zero is all that is needed. I'd question where this number came from: can you trust whichever package is giving you this ?

Two things lurk in the background.

1. Is this truncation ? (simply cutting off the last numbers to give you "00.00" whatever their value). In which case it could be 00.00999999.....

2. Or is it rounding ? (following certain rules, saying that you will round to "00.00" level of accuracy.)

If it's rounding, the rule usually followed is that if the number is >=0.005 then it will be rounded up to 0.01. As yours wasn't rounded like this, then either it is <0.005, or a different rounding rule is in place.

http://en.wikipedia.org/wiki/Rounding gives a good overview of all of this.

Whichever of the two that you look at, a level of significance of p<0.01 covers both possibilities. So I would go for that.....but I still think 00.number is dubious.

Best Wishes,

Martin Holt

From: Humphrey Paulie <[hidden email]>
To: [hidden email]
Sent: Monday, 17 May, 2010 10:47:16
Subject: Level of significance

Dear all,
When sig.=00.00 in an analysis, what level of significnace can or should we report?
p<.01, p<.001, p<.001? or something else?
Humphrey

Bruce Weaver

Re: Level of significance

Administrator

In reply to this post by Martin Holt

Martin Holt wrote

I don't agree with Ryan, below. If p were 0.009, and truncation were in operation, you would get 0.00 yet 0.009 > 0.005.

Martin

Martin, the values in the "Sig." column of SPSS output are rounded, not truncated. E.g., I just generated some output, and then edited the values in the "Sig." column.

Actual
Value Displayed (to 3 decimals, the default)
.00050 .001
.00049 .000
.00090 .001

Cheers,
Bruce

--
Bruce Weaver
bweaver@lakeheadu.ca
http://sites.google.com/a/lakeheadu.ca/bweaver/

"When all else fails, RTFM."

PLEASE NOTE THE FOLLOWING:
1. My Hotmail account is not monitored regularly. To send me an e-mail, please use the address shown above.
2. The SPSSX Discussion forum on Nabble is no longer linked to the SPSSX-L listserv administered by UGA (https://listserv.uga.edu/).

Martin Holt

Re: Level of significance

Thanks, Bruce. I must admit I had thought that this couldn't be SPSS (starting with two zero's, "00.00"), and that Humphrie must be using some freeware....does SPSS really use "00. " ? (My copy of SPSS expired some time ago, but I like the list.)

Best Wishes
Martin

From: Bruce Weaver <[hidden email]>
To: [hidden email]
Sent: Monday, 17 May, 2010 19:13:11
Subject: Re: Level of significance

Martin Holt wrote:
>
> I don't agree with Ryan, below. If p were 0.009, and truncation were in
> operation, you would get 0.00 yet 0.009Â > 0.005.
>
> Martin
>

Martin, the values in the "Sig." column of SPSS output are rounded, not
truncated. E.g., I just generated some output, and then edited the values
in the "Sig." column.

Actual
Value Displayed (to 3 decimals, the default)
.00050 .001
.00049 .000
.00090 .001

Cheers,
Bruce

-----
--
Bruce Weaver
[hidden email]
http://sites.google.com/a/lakeheadu.ca/bweaver/
"When all else fails, RTFM."

NOTE: My Hotmail account is not monitored regularly.
To send me an e-mail, please use the address shown above.
--
View this message in context: http://old.nabble.com/Level-of-significance-tp28581317p28586558.html
Sent from the SPSSX Discussion mailing list archive at Nabble.com.

=====================
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

Kornbrot, Diana

Re: Level of significance

Re: Level of significance In most versions of SPSS I’ve seen the defualt for p values is 3 sign figures
So 0.000 occurs fairly frequently
I usually look up the actual p in EXCEL and I believe SPSS rounds up or down in conventional way
Thus if SPSS give .000 then p < .0005 as p = .00051 [0r even .0005)would be rounded to p = .001
Similarly p = .00 can be interpreted as p <.005
Current recommended practice is to give EXACT p –value and ALSO always to give effect size
Best
Diana

On 17/05/2010 19:32, "M HOLT" <m861holt@...> wrote:

Thanks, Bruce. I must admit I had thought that this couldn't be SPSS (starting with two zero's, "00.00"), and that Humphrie must be using some freeware....does SPSS really use "00. " ? (My copy of SPSS expired some time ago, but I like the list.)

Best Wishes
Martin

From: Bruce Weaver <bruce.weaver@...>
To: SPSSX-L@...
Sent: Monday, 17 May, 2010 19:13:11
Subject: Re: Level of significance

Martin Holt wrote:
>
> I don't agree with Ryan, below. If p were 0.009, and truncation were in
> operation, you would get 0.00 yet 0.009Â > 0.005.
>
> Martin
>

Martin, the values in the "Sig." column of SPSS output are rounded, not
truncated. E.g., I just generated some output, and then edited the values
in the "Sig." column.

Actual
Value    Displayed (to 3 decimals, the default)
.00050    .001
.00049    .000
.00090    .001

Cheers,
Bruce

-----
--
Bruce Weaver
bweaver@...
http://sites.google.com/a/lakeheadu.ca/bweaver/
"When all else fails, RTFM."

NOTE: My Hotmail account is not monitored regularly.
To send me an e-mail, please use the address shown above.
--
View this message in context: http://old.nabble.com/Level-of-significance-tp28581317p28586558.html
Sent from the SPSSX Discussion mailing list archive at Nabble.com.

=====================
To manage your subscription to SPSSX-L, send a message to
LISTSERV@... (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

Professor Diana Kornbrot
  email: d.e.kornbrot@...
   web:    http://web.mac.com/kornbrot/iweb/KornbrotHome.html
Work
School of Psychology
University of Hertfordshire
College Lane, Hatfield, Hertfordshire AL10 9AB, UK
    voice:     +44 (0) 170 728 4626
    mobile:   +44 (0) 796 890 2102
    fax          +44 (0) 170 728 5073
Home
19 Elmhurst Avenue
London N2 0LT, UK
   landline: +44 (0) 208 883 3657
   mobile:   +44 (0) 796 890 2102
   fax:         +44 (0) 870 706 4997

msherman

Re: Level of significance

When I open up the number of decimal places within SPSS 18.0 it maxes out = at 32 decimal places. Thus, if I have a large N I might find that the p

value (not the significance level) could go out quite far. For instance, I h= ad a correlation of .528 with an N of 142 and got the following output for = the p value.

.000000000007345309847637664000000000

Would one then go ahead and report it as p < . .000000000008

Martin F. Sherman, Ph.D.

Professor of Psychology

Director of Masters Education: Thesis Track Loyola College of Arts and Sciences

Loyola University Maryland

4501 North Charles Street

222 B Beatty Hall

Baltimore, MD 21210-2601

410-617-2417 office

410-617-5341 fax

[hidden email]>

www.loyola.edu<http://www.loyola.edu

From: SPSSX(r) Discussion [mailto:[hidden email]] On Behalf Of kornbrot
Sent: Monday, May 17, 2010 2:52 PM
To: [hidden email]
Subject: Re: Level of significance

In most versions of SPSS I’ve seen the defualt for p values is 3 sign figures
So 0.000 occurs fairly frequently
I usually look up the actual p in EXCEL and I believe SPSS rounds up or down in conventional way
Thus if SPSS give .000 then p < .0005 as p = .00051 [0r even .0005)would be rounded to p = .001
Similarly p = .00 can be interpreted as p <.005
Current recommended practice is to give EXACT p –value and ALSO always to give effect size
Best
Diana

On 17/05/2010 19:32, "M HOLT" <m861holt@...> wrote:

From: Bruce Weaver <bruce.weaver@...>
To: SPSSX-L@...
Sent: Monday, 17 May, 2010 19:13:11
Subject: Re: Level of significance

Martin Holt wrote:
>
> I don't agree with Ryan, below. If p were 0.009, and truncation were in
> operation, you would get 0.00 yet 0.009Â > 0.005.
>
> Martin
>

Martin, the values in the "Sig." column of SPSS output are rounded, not
truncated. E.g., I just generated some output, and then edited the values
in the "Sig." column.

Actual
Value    Displayed (to 3 decimals, the default)
.00050    .001
.00049    .000
.00090    .001

Cheers,
Bruce

-----
--
Bruce Weaver
bweaver@...
http://sites.google.com/a/lakeheadu.ca/bweaver/
"When all else fails, RTFM."

NOTE: My Hotmail account is not monitored regularly.
To send me an e-mail, please use the address shown above.
--
View this message in context: http://old.nabble.com/Level-of-significance-tp28581317p28586558.html
Sent from the SPSSX Discussion mailing list archive at Nabble.com.

=====================
To manage your subscription to SPSSX-L, send a message to
LISTSERV@... (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

Re: Level of significance

the reported p value is a number that the obtained p value is less than. p < .001 says that less than 1 tenth of a percent of the total area under the curve is beyond the cut-off(s). The computer cannot understand the substantive nature of your data so the programmers report more digits to cover all the bases.

In these circumstances I suggest reporting "p < .01". With that kind of N, I would orally report the correlation as .5ish. Offhand I cannot think of practical circumstances where reporting the third digit on a correlation does anything other than make it harder for readers to get a notion of what you found.

It is my belief that for most social and behavioral science applications and for the cognitive capacities of most mortals reporting any more than 3 digits for a p is deceitful.

Art Kendall
Social Research Consultants

On 5/17/2010 3:08 PM, Martin Sherman wrote:

Re: Level of significance

When I open up the number of decimal places within SPSS 18.0 it maxes out = at 32 decimal places. Thus, if I have a large N I might find that the p

value (not the significance level) could go out quite far. For instance, I h= ad a correlation of .528 with an N of 142 and got the following output for = the p value.

.000000000007345309847637664000000000

Would one then go ahead and report it as p < . .000000000008

Martin F. Sherman, Ph.D.

Professor of Psychology

Director of Masters Education: Thesis Track Loyola College of Arts and Sciences

Loyola University Maryland

4501 North Charles Street

222 B Beatty Hall

Baltimore, MD 21210-2601

410-617-2417 office

410-617-5341 fax

[hidden email]>

www.loyola.edu<http://www.loyola.edu

From: SPSSX(r) Discussion [[hidden email]] On Behalf Of kornbrot
Sent: Monday, May 17, 2010 2:52 PM
To: [hidden email]
Subject: Re: Level of significance

In most versions of SPSS I’ve seen the defualt for p values is 3 sign figures
So 0.000 occurs fairly frequently
I usually look up the actual p in EXCEL and I believe SPSS rounds up or down in conventional way
Thus if SPSS give .000 then p < .0005 as p = .00051 [0r even .0005)would be rounded to p = .001
Similarly p = .00 can be interpreted as p <.005
Current recommended practice is to give EXACT p –value and ALSO always to give effect size
Best
Diana

On 17/05/2010 19:32, "M HOLT" <m861holt@...> wrote:

Thanks, Bruce. I must admit I had thought that this couldn't be SPSS (starting with two zero's, "00.00"), and that Humphrie must be using some freeware....does SPSS really use "00. " ? (My copy of SPSS expired some time ago, but I like the list.)

Best Wishes
Martin

From: Bruce Weaver <bruce.weaver@...>
To: SPSSX-L@...
Sent: Monday, 17 May, 2010 19:13:11
Subject: Re: Level of significance

Martin Holt wrote:
>
> I don't agree with Ryan, below. If p were 0.009, and truncation were in
> operation, you would get 0.00 yet 0.009Â > 0.005.
>
> Martin
>

Martin, the values in the "Sig." column of SPSS output are rounded, not
truncated. E.g., I just generated some output, and then edited the values
in the "Sig." column.

Actual
Value    Displayed (to 3 decimals, the default)
.00050    .001
.00049    .000
.00090    .001

Cheers,
Bruce

-----
--
Bruce Weaver
bweaver@...
http://sites.google.com/a/lakeheadu.ca/bweaver/
"When all else fails, RTFM."

NOTE: My Hotmail account is not monitored regularly.
To send me an e-mail, please use the address shown above.
--
View this message in context: http://old.nabble.com/Level-of-significance-tp28581317p28586558.html
Sent from the SPSSX Discussion mailing list archive at Nabble.com.

=====================
To manage your subscription to SPSSX-L, send a message to
LISTSERV@... (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

Professor Diana Kornbrot
  email: d.e.kornbrot@...
   web:    http://web.mac.com/kornbrot/iweb/KornbrotHome.html
Work
School of Psychology
University of Hertfordshire
College Lane, Hatfield, Hertfordshire AL10 9AB, UK
    voice:     +44 (0) 170 728 4626
    mobile:   +44 (0) 796 890 2102
    fax          +44 (0) 170 728 5073
Home
19 Elmhurst Avenue
London N2 0LT, UK
   landline: +44 (0) 208 883 3657
   mobile:   +44 (0) 796 890 2102
   fax:         +44 (0) 870 706 4997

Art Kendall
Social Research Consultants

Mike

Re: Level of significance

I believe the requirement for exact p-values is based on the American Psychological

Association (APA) style guide. The following is a quote from page 114 in the 6th ed

of the APA publication manual:

|When reporting p values, report exact p values (e.g., p=.031) to two or three

|decimal places. However, report p values less than .001 as p< .001. The tradition

|of reporting p values in the form p< .10, p< .05, p< .01, and so forth, was appropriate

|in a time when only limited tables of critical values were available. However, in

|table "p<" notation may be necessary for clarity (see section 5.16).

I believe that reporting "exact p values" may be due in part to the work by

Robert Rosenthal that popularize the notion that one could convert p-values

into effect size measures. NOTE: I quote from the first printing of the

6th edition which has a remarkable number of errors in it and one should

consult a later printing to see if this rationale has changes.

-Mike Palij

New York University

[hidden email]

----- Original Message -----

From: [hidden email]

To: [hidden email]

Sent: Monday, May 17, 2010 4:17 PM

Subject: Re: Level of significance

the reported p value is a number that the obtained p value is less than. p < .001 says that less than 1 tenth of a percent of the total area under the curve is beyond the cut-off(s). The computer cannot understand the substantive nature of your data so the programmers report more digits to cover all the bases.

In these circumstances I suggest reporting "p < .01".   With that kind of N, I would orally report the correlation as .5ish. Offhand I cannot think of practical circumstances where reporting the third digit on a correlation does anything other than make it harder for readers to get a notion of what you found.

It is my belief that for most social and behavioral science applications and for the cognitive capacities of most mortals reporting any more than 3 digits for a p is deceitful.

Art Kendall
Social Research Consultants

On 5/17/2010 3:08 PM, Martin Sherman wrote:

When I open up the number of decimal places within SPSS 18.0 it maxes out = at 32 decimal places. Thus, if I have a large N I might find that the p

value (not the significance level) could go out quite far. For instance, I h= ad a correlation of .528 with an N of 142 and got the following output for = the p value.

.000000000007345309847637664000000000

Would one then go ahead and report it as p < . .000000000008

Martin F. Sherman, Ph.D.

Professor of Psychology

Director of Masters Education: Thesis Track Loyola College of Arts and Sciences

Loyola University Maryland

4501 North Charles Street

222 B Beatty Hall

Baltimore, MD 21210-2601

410-617-2417 office

410-617-5341 fax

[hidden email]>

www.loyola.edu<http://www.loyola.edu

From: SPSSX(r) Discussion [[hidden email]] On Behalf Of kornbrot
Sent: Monday, May 17, 2010 2:52 PM
To: [hidden email]
Subject: Re: Level of significance

In most versions of SPSS Ive seen the defualt for p values is 3 sign figures
So 0.000 occurs fairly frequently
I usually look up the actual p in EXCEL and I believe SPSS rounds up or down in conventional way
Thus if SPSS give .000 then p < .0005 as p = .00051 [0r even .0005)would be rounded to p = .001
Similarly p = .00 can be interpreted as p <.005
Current recommended practice is to give EXACT p value and ALSO always to give effect size
Best
Diana

On 17/05/2010 19:32, "M HOLT" <m861holt@...> wrote:

Thanks, Bruce. I must admit I had thought that this couldn't be SPSS (starting with two zero's, "00.00"), and that Humphrie must be using some freeware....does SPSS really use "00. " ? (My copy of SPSS expired some time ago, but I like the list.)

Best Wishes
Martin

From: Bruce Weaver <bruce.weaver@...>
To: SPSSX-L@...
Sent: Monday, 17 May, 2010 19:13:11
Subject: Re: Level of significance

Martin Holt wrote:
>
> I don't agree with Ryan, below. If p were 0.009, and truncation were in
> operation, you would get 0.00 yet 0.009Â > 0.005.
>
> Martin
>

Martin, the values in the "Sig." column of SPSS output are rounded, not
truncated. E.g., I just generated some output, and then edited the values
in the "Sig." column.

Actual
Value    Displayed (to 3 decimals, the default)
.00050    .001
.00049    .000
.00090    .001

Cheers,
Bruce

-----
--
Bruce Weaver
bweaver@...
http://sites.google.com/a/lakeheadu.ca/bweaver/
"When all else fails, RTFM."

NOTE: My Hotmail account is not monitored regularly.
To send me an e-mail, please use the address shown above.
--
View this message in context: http://old.nabble.com/Level-of-significance-tp28581317p28586558.html
Sent from the SPSSX Discussion mailing list archive at Nabble.com.

=====================
To manage your subscription to SPSSX-L, send a message to
LISTSERV@... (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

Professor Diana Kornbrot
  email: d.e.kornbrot@...
   web:    http://web.mac.com/kornbrot/iweb/KornbrotHome.html
Work
School of Psychology
University of Hertfordshire
College Lane, Hatfield, Hertfordshire AL10 9AB, UK
    voice:     +44 (0) 170 728 4626
    mobile:   +44 (0) 796 890 2102
    fax          +44 (0) 170 728 5073
Home
19 Elmhurst Avenue
London N2 0LT, UK
   landline: +44 (0) 208 883 3657
   mobile:   +44 (0) 796 890 2102
   fax:         +44 (0) 870 706 4997

===================== 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

Kornbrot, Diana

Re: Level of significance

Re: Level of significance No this rationale was in 5th Ed too
It is due to Wilkinson’s review [not rosenthal] – he’s the creator of systat
Wilkinson, L. (1999). Statistical methods in psychology journals - Guidelines and explanations. American Psychologist, 54(8), 594-604. http://psycnet.apa.org/journals/amp/54/8/594.html.

Best
Diana

On 17/05/2010 21:45, "Mike Palij" <mp26@...> wrote:

I believe the requirement for exact p-values is based on the American Psychological
Association (APA) style guide. The following is a quote from page 114 in the 6th ed
of the APA publication manual:

|When reporting p values, report exact p values (e.g., p=.031) to two or three
|decimal places. However, report p values less than .001 as p< .001. The tradition
|of reporting p values in the form p< .10, p< .05, p< .01, and so forth, was appropriate
|in a time when only limited tables of critical values were available. However, in
|table "p<" notation may be necessary for clarity (see section 5.16).

I believe that reporting "exact p values" may be due in part to the work by
Robert Rosenthal that popularize the notion that one could convert p-values
into effect size measures. NOTE: I quote from the first printing of the
6th edition which has a remarkable number of errors in it and one should
consult a later printing to see if this rationale has changes.

-Mike Palij
New York University
mp26@...

----- Original Message -----

From: Art Kendall <[hidden email]>

To: SPSSX-L@...

Sent: Monday, May 17, 2010 4:17 PM

Subject: Re: Level of significance

the reported p value is a number that the obtained p value is less than. p < .001 says that less than 1 tenth of a percent of the total area under the curve is beyond the cut-off(s). The computer cannot understand the substantive nature of your data so the programmers report more digits to cover all the bases.

In these circumstances I suggest reporting "p < .01".   With that kind of N, I would orally report the correlation as .5ish. Offhand I cannot think of practical circumstances where reporting the third digit on a correlation does anything other than make it harder for readers to get a notion of what you found.

It is my belief that for most social and behavioral science applications and for the cognitive capacities of most mortals reporting any more than 3 digits for a p is deceitful.

Art Kendall
Social Research Consultants

On 5/17/2010 3:08 PM, Martin Sherman wrote:



When I open up the number of decimal places within SPSS 18.0 it maxes out = at 32 decimal places. Thus, if I have a large N I might find that the p

value (not the significance level) could go out quite far. For instance, I h= ad a correlation of .528 with an N of 142 and got the following output for = the p value.

.000000000007345309847637664000000000

Would one then go ahead and report it as p < . .000000000008

Martin F. Sherman, Ph.D.

Professor of Psychology

Director of Masters Education: Thesis Track Loyola College of Arts and Sciences

Loyola University Maryland

4501 North Charles Street

222 B Beatty Hall

Baltimore, MD 21210-2601

410-617-2417 office

410-617-5341 fax

msherman@...<username@loyola.edu <[hidden email]> >

MailScanner has detected a possible fraud attempt from "www.loyola.edu www.loyola.edu<http://www.loyola.edu <http://www.loyola.edu%3chttp:/www.loyola.edu>

From: SPSSX(r) Discussion [[hidden email]] On Behalf Of kornbrot
Sent: Monday, May 17, 2010 2:52 PM
To: SPSSX-L@...
Subject: Re: Level of significance

In most versions of SPSS I’ve seen the defualt for p values is 3 sign figures
So 0.000 occurs fairly frequently
I usually look up the actual p in EXCEL and I believe SPSS rounds up or down in conventional way
Thus if SPSS give .000 then p < .0005 as p = .00051 [0r even .0005)would be rounded to p = .001
Similarly p = .00 can be interpreted as p <.005
Current recommended practice is to give EXACT p –value and ALSO always to give effect size
Best
Diana

On 17/05/2010 19:32, "M HOLT" <m861holt@...> wrote:

Thanks, Bruce. I must admit I had thought that this couldn't be SPSS (starting with two zero's, "00.00"), and that Humphrie must be using some freeware....does SPSS really use "00.   " ? (My copy of SPSS expired some time ago, but I like the list.)

Best Wishes
Martin

From: Bruce Weaver <bruce.weaver@...>
To: SPSSX-L@...
Sent: Monday, 17 May, 2010 19:13:11
Subject: Re: Level of significance

Martin Holt wrote:
>
> I don't agree with Ryan, below. If p were 0.009, and truncation were in
> operation, you would get 0.00 yet 0.009Â > 0.005.
>
> Martin
>

Martin, the values in the "Sig." column of SPSS output are rounded, not
truncated. E.g., I just generated some output, and then edited the values
in the "Sig." column.

Actual
Value    Displayed (to 3 decimals, the default)
.00050    .001
.00049     .000
.00090     .001

Cheers,
Bruce

-----
--
Bruce Weaver
bweaver@...
http://sites.google.com/a/lakeheadu.ca/bweaver/
"When all else fails, RTFM."

NOTE: My Hotmail account is not monitored regularly.
To send me an e-mail, please use the address shown above.
--
View this message in context: http://old.nabble.com/Level-of-significance-tp28581317p28586558.html
Sent from the SPSSX Discussion mailing list archive at Nabble.com.

=====================
To manage your subscription to SPSSX-L, send a message to
LISTSERV@... (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

Professor Diana Kornbrot
email: d.e.kornbrot@...
   web:     http://web.mac.com/kornbrot/iweb/KornbrotHome.html
Work
School of Psychology
University of Hertfordshire
College Lane, Hatfield, Hertfordshire AL10 9AB, UK
    voice:      +44 (0) 170 728 4626
    mobile:   +44 (0) 796 890 2102
    fax           +44 (0) 170 728 5073
Home
19 Elmhurst Avenue
London N2 0LT, UK
   landline: +44 (0) 208 883 3657
   mobile:   +44 (0) 796 890 2102
   fax:          +44 (0) 870 706 4997

===================== To manage your subscription to SPSSX-L, send a message to LISTSERV@... (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

Professor Diana Kornbrot
email: d.e.kornbrot@...
web:    http://web.me.com/kornbrot/KornbrotHome.html
Work
School of Psychology
University of Hertfordshire
College Lane, Hatfield, Hertfordshire AL10 9AB, UK
voice:   +44 (0) 170 728 4626
   fax:     +44 (0) 170 728 5073
Home
19 Elmhurst Avenue
London N2 0LT, UK
    voice:   +44 (0) 208 883 3657
    mobile: +44 (0) 796 890 2102
   fax:      +44 (0) 870 706 4997

statisticsdoc

Re: Level of significance

In reply to this post by Mike

Re: Level of significance This might be a tangent to the discussion, but p values should not be used as an index of effect size. P may decrrease as a funtion of sample size even when the effect size remains constant or indeed diminishes.
Best,
Steve Brand

www.StatisticsDoc.com

From: Mike Palij <[hidden email]>

Date: Mon, 17 May 2010 16:45:11 -0400

To: <[hidden email]>

Subject: Re: Level of significance