Jump to content

  • Log in with Facebook Log in with Twitter Log in with Windows Live Log In with Google      Sign In   
  • Create Account

Submit your paper to J Biol Methods today!
Photo
- - - - -

I do not know which statistics to use


  • Please log in to reply
6 replies to this topic

#1 Tene

Tene

    member

  • Members
  • Pip
  • 4 posts
0
Neutral

Posted 29 September 2010 - 02:10 AM

Hello Friends,
I have got data from field trials with jatropha: planting with amendment and the control (without amendment).
I mesure the height of each of the 20 plants every week for 4 months. With which statistical test could i compare these heights: plant height from the control plot and plot heiht from the treatment plot.
Could anyone here help me ?
Thanks in advance
Tene

#2 bob1

bob1

    Thelymitra pulchella

  • Global Moderators
  • PipPipPipPipPipPipPipPipPipPip
  • 6,668 posts
565
Excellent

Posted 29 September 2010 - 04:15 PM

It sounds like you have some normally distibuted samples there (height), but it would pay to test whether it is normally distributed before doing statistics on the samples. Either use Student's t-test (for normally distributed samples) or a non-parametric test such as Mann-Whitney test for non-normally distributed samples to compare individual week's growth. To compare growth overall you may need to use an ANOVA or a Kruskal Wallace test.

#3 hobglobin

hobglobin

    Growing old is mandatory, growing up is optional...

  • Global Moderators
  • PipPipPipPipPipPipPipPipPipPip
  • 5,604 posts
127
Excellent

Posted 29 September 2010 - 10:35 PM

Hopefully you used a (generalised) randomised block design....

A single lie is reproachable; a million lies is a statistic.
D. J. T.


#4 Tene

Tene

    member

  • Members
  • Pip
  • 4 posts
0
Neutral

Posted 30 September 2010 - 01:31 AM

Hi Bob, Thanks
One of my colleague told me that the T-test in my case is not recommendable because the variables are not independent. he said the Height of a plant this week depend on the height of the same plant last week and so on...What do you think?
Ok a will try what you said. Thanks

It sounds like you have some normally distibuted samples there (height), but it would pay to test whether it is normally distributed before doing statistics on the samples. Either use Student's t-test (for normally distributed samples) or a non-parametric test such as Mann-Whitney test for non-normally distributed samples to compare individual week's growth. To compare growth overall you may need to use an ANOVA or a Kruskal Wallace test.



#5 Tene

Tene

    member

  • Members
  • Pip
  • 4 posts
0
Neutral

Posted 30 September 2010 - 01:34 AM

Hi,
No i did not use the RBD because the surface was uniform..
Each plant is consider in this case as a replicate... Any idea how one could handle such a data? Thanks for your comments

Hopefully you used a (generalised) randomised block design....



#6 hobglobin

hobglobin

    Growing old is mandatory, growing up is optional...

  • Global Moderators
  • PipPipPipPipPipPipPipPipPipPip
  • 5,604 posts
127
Excellent

Posted 30 September 2010 - 03:31 AM

Hi,
No i did not use the RBD because the surface was uniform..
Each plant is consider in this case as a replicate... Any idea how one could handle such a data? Thanks for your comments


Light intensity and direction and surroundings (other plants) are also the same then?
Anyway a Repeated Measures ANOVA should do the job, if the data conform more or less with the assumptions for ANOVA. A Friedman test is the non-parametric alternative.

A single lie is reproachable; a million lies is a statistic.
D. J. T.


#7 Tene

Tene

    member

  • Members
  • Pip
  • 4 posts
0
Neutral

Posted 01 October 2010 - 06:30 AM

Thanks my dear.
Regards


Hi,
No i did not use the RBD because the surface was uniform..
Each plant is consider in this case as a replicate... Any idea how one could handle such a data? Thanks for your comments


Light intensity and direction and surroundings (other plants) are also the same then?
Anyway a Repeated Measures ANOVA should do the job, if the data conform more or less with the assumptions for ANOVA. A Friedman test is the non-parametric alternative.






Home - About - Terms of Service - Privacy - Contact Us

©1999-2013 Protocol Online, All rights reserved.