Fwd: bug: p-value > 1.0 for paired t-test

---------- Forwarded message ----------
From: William Simpson <address@hidden>
Date: Thu, Nov 12, 2009 at 6:16 PM
Subject: Re: bug: p-value > 1.0 for paired t-test
To: John Darrington <address@hidden>

Thanks very much John!

I grabbed the binary from
http://www.cecaps.ufmg.br/pspp/?page_id=128&lang=en
which, so far as I know, is the place that the PSPP home page directs people to when they want to download a windows binary.

So, maintainer over at cecaps, if you're out there, please fix! Thanks!

Cheers
Bill

On Thu, Nov 12, 2009 at 6:05 PM, John Darrington <address@hidden> wrote:

This bug was fixed some time ago. See
http://savannah.gnu.org/bugs/index.php?26936

Obviously it didn't make it into the copy that you have.

You can work around it by reversing the order of the pair.

J'

On Thu, Nov 12, 2009 at 03:54:36PM +0000, William Simpson wrote:
Windows version 0.7.2
Data file attached

paired t-test results (see tail end for screwy p-value > 1.0):

1.1 T-TEST. Paired Sample Statistics
#==========#======#==#==============#========#
# # Mean | N|Std. Deviation|SE. Mean#
#==========#======#==#==============#========#
#Pair 0 dc #365.84|18| 69.31| 16.34#
# dnc#366.52|18| 77.33| 18.23#
#Pair 1 pc #372.82|18| 66.64| 15.71#
# pnc#362.14|18| 65.03| 15.33#
#==========#======#==#==============#========#

1.2(2) T-TEST. Paired Samples Correlations
#======#========#==#===========#====#
# | # N|Correlation|Sig.#
#======#========#==#===========#====#
#Pair 0|dc & dnc#18| .98| .00#
#Pair 1|pc & pnc#18| .98| .00#
#======#========#==#===========#====#

1.3(1:2) T-TEST. Paired Samples Test
#==============#==========================================#
# # Paired Differences #
# #-----+--------------+---------------+-----#
# # | | | 95% #
# # | | +-----#
# # Mean|Std. Deviation|Std. Error Mean|Lower#
#==============#=====#==============#===============#=====#
#Pair 0dc - dnc# -.68| 18.06| 4.26|-9.66#
#Pair 1pc - pnc#10.68| 13.29| 3.13| 4.07#
#==============#=====#==============#===============#=====#

1.3(2:2) T-TEST. Paired Samples Test
#=====#====#==#===============#
# | | | #
#-----+ | | #
# | | | #
#-----+ | | #
#Upper| t |df|Sig. (2-tailed)#
#=====#====#==#===============#
# 8.30|-.16|17| .87#
#17.29|3.41|17| 2.00#
#=====#====#==#===============#

2.1 T-TEST. Paired Sample Statistics
#==========#======#==#==============#========#
# # Mean | N|Std. Deviation|SE. Mean#
#==========#======#==#==============#========#
#Pair 0 dc #370.20|17| 68.85| 16.70#
# dnc#372.61|17| 75.13| 18.22#
#Pair 1 pc #376.36|17| 66.92| 16.23#
# pnc#365.79|17| 65.10| 15.79#
#==========#======#==#==============#========#

2.2 T-TEST. Paired Samples Correlations
#======#========#==#===========#====#
# | # N|Correlation|Sig.#
#======#========#==#===========#====#
#Pair 0|dc & dnc#17| .98| .00#
#Pair 1|pc & pnc#17| .98| .00#
#======#========#==#===========#====#

2.3(1) T-TEST. Paired Samples Test
#==============#=================================================#====#==#
# # Paired Differences | | #
# #-----+--------------+---------------+------------+ | #
# # | | | 95% | | #
# # | | +------+-----+ | #
# # Mean|Std. Deviation|Std. Error Mean| Lower|Upper| t |df#
#==============#=====#==============#===============#======#=====#====#==#
#Pair 0dc - dnc#-2.41| 17.01| 4.12|-11.16| 6.33|-.59|16#
#Pair 1pc - pnc#10.57| 13.69| 3.32| 3.53|17.61|3.18|16#
#==============#=====#==============#===============#======#=====#====#==#

2.3(2:2) T-TEST. Paired Samples Test
#===============#
# #
# #
# #
# #
#Sig. (2-tailed)#
#===============#
# .57#
# 1.99#
#===============#

3.1 T-TEST. Paired Sample Statistics
#==========#======#==#==============#========#
# # Mean | N|Std. Deviation|SE. Mean#
#==========#======#==#==============#========#
#Pair 0 dc #370.20|17| 68.85| 16.70#
# dnc#372.61|17| 75.13| 18.22#
#Pair 1 pc #376.36|17| 66.92| 16.23#
# pnc#365.79|17| 65.10| 15.79#
#==========#======#==#==============#========#

3.2 T-TEST. Paired Samples Correlations
#======#========#==#===========#====#
# | # N|Correlation|Sig.#
#======#========#==#===========#====#
#Pair 0|dc & dnc#17| .98| .00#
#Pair 1|pc & pnc#17| .98| .00#
#======#========#==#===========#====#

3.3(1) T-TEST. Paired Samples Test
#==============#=================================================#====#==#
# # Paired Differences | | #
# #-----+--------------+---------------+------------+ | #
# # | | | 95% | | #
# # | | +------+-----+ | #
# # Mean|Std. Deviation|Std. Error Mean| Lower|Upper| t |df#
#==============#=====#==============#===============#======#=====#====#==#
#Pair 0dc - dnc#-2.41| 17.01| 4.12|-11.16| 6.33|-.59|16#
#Pair 1pc - pnc#10.57| 13.69| 3.32| 3.53|17.61|3.18|16#
#==============#=====#==============#===============#======#=====#====#==#

3.3(2:2) T-TEST. Paired Samples Test
#===============#
# #
# #
# #
# #
#Sig. (2-tailed)#
#===============#
# .57#
# 1.99#******************** p-value > 1.0!! SPSS gives 0.006
#===============#

Please let me know what's going on here. Is it just a numerical problem with
very small p-values??

Cheers
Bill

_______________________________________________
Pspp-users mailing list
address@hidden
http://lists.gnu.org/mailman/listinfo/pspp-users

--
PGP Public key ID: 1024D/2DE827B3
fingerprint = 8797 A26D 0854 2EAB 0285 A290 8A67 719C 2DE8 27B3
See http://pgp.mit.edu or any PGP keyserver for public key.

-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.9 (GNU/Linux)

iD8DBQFK/E5ximdxnC3oJ7MRAqA+AJ9PiutmymoXhtPdFOgmXAi8Jj5zWACfbx9u
Mu10EUp0vHOs2VqvqcYSrFA=
=uPtN
-----END PGP SIGNATURE-----

From:	William Simpson
Subject:	Fwd: bug: p-value > 1.0 for paired t-test
Date:	Thu, 12 Nov 2009 18:17:37 +0000