Sponsored by: Discount Supplements - For an extra 6% off Absolutely Everything - use code: SEPT6

Stats Help (JK2!)

Author
Ak_88
Pro-Member
  • Total Posts : 19512
  • Reward points: 9522
  • Joined: 2005/11/06 21:35:22
  • Location: Chelmsford, Essex
  • Status: offline
2008/05/15 20:38:32 (permalink)

Stats Help (JK2!)

I'll get my head around it all eventually, but until then i'll have to bask in your glory and soak up your knowledge.

When running a paired sample t-test (because the two sets of data i have are manipulated by a condition/related) the SPSS output gives the 'T' value.

Aside from significance and correlation - does this value signify anything alone or is it arbitrary without the significance value?

The T-value in question is -1.121, and the significance (P) value is 0.266 (so obviously non-significant).

Ta
#1

13 Replies Related Threads

    JK2
    Pro-Member
    • Total Posts : 14598
    • Reward points: 4963
    • Joined: 2003/07/14 15:07:10
    • Location: London United Kingdom
    • Status: offline
    RE: Stats Help (JK2!) 2008/05/15 21:32:19 (permalink)
    The t-value is a calculation based on the means and standard deviations of the two samples.

    In fact, it's this:

    In the results section of your report you need to state the t-value, the degrees of freedom (df) and the p-value.

    #2
    Ak_88
    Pro-Member
    • Total Posts : 19512
    • Reward points: 9522
    • Joined: 2005/11/06 21:35:22
    • Location: Chelmsford, Essex
    • Status: offline
    RE: Stats Help (JK2!) 2008/05/15 21:36:09 (permalink)
    Yup - thats what i thought.

    So the T value does not signify anything if reported alone? I.e it is not indicative of anything unless combined with degrees of freedom and P value?
    #3
    Red Man
    Pro-Member
    • Total Posts : 10217
    • Reward points: 6562
    • Joined: 2006/10/11 12:01:49
    • Location: North West
    • Status: offline
    RE: Stats Help (JK2!) 2008/05/15 21:40:41 (permalink)

    ORIGINAL: JohnKerr2

    The t-value is a calculation based on the means and standard deviations of the two samples.

    In fact, it's this:

    In the results section of your report you need to state the t-value, the degrees of freedom (df) and the p-value.


    Sorry to butt in here but what do you do for a living JK to know this sort of stuff?
    Baffles me guys. (which isnt hard tbh).

    "You're not drunk if you can lie on the floor without holding on." Dean Martin

    #4
    JK2
    Pro-Member
    • Total Posts : 14598
    • Reward points: 4963
    • Joined: 2003/07/14 15:07:10
    • Location: London United Kingdom
    • Status: offline
    RE: Stats Help (JK2!) 2008/05/15 21:41:00 (permalink)
    It is the most important statistic since it's the measure of how different your two samples are. But unless you want your reader/report-marker to look up the p-value for themselves then I'd recommend sticking that in too.

    #5
    JK2
    Pro-Member
    • Total Posts : 14598
    • Reward points: 4963
    • Joined: 2003/07/14 15:07:10
    • Location: London United Kingdom
    • Status: offline
    RE: Stats Help (JK2!) 2008/05/15 21:42:26 (permalink)
    ORIGINAL: Red Man

    Sorry to butt in here but what do you do for a living JK to know this sort of stuff?
    Baffles me guys. (which isnt hard tbh).


    About a hundred years ago I did a degree in psychology - which is essentially all about psychological testing and experimentation and thus statistics and statistical analysis and interpretation.

    #6
    Ak_88
    Pro-Member
    • Total Posts : 19512
    • Reward points: 9522
    • Joined: 2005/11/06 21:35:22
    • Location: Chelmsford, Essex
    • Status: offline
    RE: Stats Help (JK2!) 2008/05/15 21:44:41 (permalink)

    ORIGINAL: JohnKerr2

    It is the most important statistic since it's the measure of how different your two samples are. But unless you want your reader/report-marker to look up the p-value for themselves then I'd recommend sticking that in too.


    Yeah - i've got all the stuff in here but i was just wondering if a T-value can explain anything without the other values - since i have no idea what a 'normal' T value would be.
    #7
    JK2
    Pro-Member
    • Total Posts : 14598
    • Reward points: 4963
    • Joined: 2003/07/14 15:07:10
    • Location: London United Kingdom
    • Status: offline
    RE: Stats Help (JK2!) 2008/05/15 21:49:29 (permalink)
    It's a simple kind of test reliant on the idea of a threshold beyond which you can choose to decide that something more than chance is involved. I wouldn't get too bothered about how the t-value is calculated (for that's what you'd have to do if you wanted to 'understand' it). Just let SPSS do its magic and report the output values.
    post edited by JohnKerr2 - 2008/05/15 21:50:03

    #8
    Ak_88
    Pro-Member
    • Total Posts : 19512
    • Reward points: 9522
    • Joined: 2005/11/06 21:35:22
    • Location: Chelmsford, Essex
    • Status: offline
    RE: Stats Help (JK2!) 2008/05/15 21:51:55 (permalink)
    Ta very much mr Kerr2, appreciated as always.
    #9
    JK2
    Pro-Member
    • Total Posts : 14598
    • Reward points: 4963
    • Joined: 2003/07/14 15:07:10
    • Location: London United Kingdom
    • Status: offline
    RE: Stats Help (JK2!) 2008/05/15 21:52:39 (permalink)
    Very welcome.

    #10
    carl19
    Pro-Member
    • Total Posts : 2955
    • Reward points: 4770
    • Joined: 2006/09/24 21:44:02
    • Status: offline
    RE: Stats Help (JK2!) 2008/05/15 22:12:28 (permalink)
    i think jk2 has it covered


    btw have you done a normality test on the data or are you just going straight into the T test?
    #11
    Ak_88
    Pro-Member
    • Total Posts : 19512
    • Reward points: 9522
    • Joined: 2005/11/06 21:35:22
    • Location: Chelmsford, Essex
    • Status: offline
    RE: Stats Help (JK2!) 2008/05/15 22:15:59 (permalink)
    Thats another point i probably should've covered Carl - i've done a descriptives test but i've only reported the means and standard deviations. Is there any need to report Skewness/Kurtosis or run anything else?
    #12
    carl19
    Pro-Member
    • Total Posts : 2955
    • Reward points: 4770
    • Joined: 2006/09/24 21:44:02
    • Status: offline
    RE: Stats Help (JK2!) 2008/05/15 22:21:56 (permalink)
    we are taught to run a normality test on the data like a shapiro-wilk test or some do a levenes i think

    if the data is normaly distributed, (that is NOT Significantly Different from Normal) on the shapiro-wilk then we would do a t test however if the data was not normally distributed we would do the non parametric version which in your case would be the wilcoxon test i believe
    #13
    Ak_88
    Pro-Member
    • Total Posts : 19512
    • Reward points: 9522
    • Joined: 2005/11/06 21:35:22
    • Location: Chelmsford, Essex
    • Status: offline
    RE: Stats Help (JK2!) 2008/05/15 22:24:24 (permalink)
    Cheers bud - i'll run that now
    #14
    Jump to:
    ©2017 All content is copyright of MuscleTalk.co.uk and its use elsewhere is prohibited. (posting guidelines | privacy | advertise | contact us | supported by)
    © 2017 APG vNext Commercial Version 5.5