Research in Biology (statistical help)

[TheBigEscape]

Hi!

I am a biology student working on a paper about the effect of heavy metal treatments (3 treatments) on the following variables:
a. root length - before and after treatment
b. shoot length - before and after treatment
c. leaf area
d. chlorophyll content

all of which I have measured already. I want to know if there is any model that I can use to determine if there is a relationship among the variables and at the same time I could use the data taken from the treatments to compare it with the control. I figured that there are so many that I can (ANOVA, etc) do but I am confused on what is the 'best' method that I can use.

And my knowledge on statistics is basic, I have yet to take units of advance stat in the near future.

Thanks a lot!

[TheEcologist]

Hi!

I am a biology student working on a paper about the effect of heavy metal treatments (3 treatments) on the following variables:
a. root length - before and after treatment
b. shoot length - before and after treatment
c. leaf area
d. chlorophyll content


Thanks a lot!

This depends on a variety of factors (variable under study, distribution ect) but the two-way Anova is a likely candidate for what you need (and probably the single most used test in experimental plant science).

Here’s a website that should help you on your way:
http://udel.edu/~mcdonald/statbigchart.html

Nobody here can really help you much without knowing what type of variable you have. So this page is useful to study first if you need to post back:
http://udel.edu/~mcdonald/statvartypes.html

[bugman]

If your variables, shoot length, root length leaf area and chlorophyll are meassured from the same individual they are not independent and the chance of correltaion beteween these variabales is high.

You have two factor: Treatment and time.

I think you should be looking at a two-factor MANOVA for your data set.




Phil

[TheEcologist]

A two-factor MANOVA.
Phil

Indeed, I would agree (If I guess your question correctly).

[TheBigEscape]

Thanks for the link TheEcologist, I guess I have many to review about Stat, (I took basic stats two years ago).

I have been looking to two-way ANOVA since that is what they normally use with this kind of data (1 nominal & 1 measured variable) but I have unequal sample sizes (different n per treatment; since some of the plants I have grown died).

I initially grown 40 seedlings per treatment (control, treatment 1, 2 and 3) and I've measured their root and shoot length before starting the treatments and after 1 month (some of the plants died) I've measured their root and shoot length [after treatment]. From what I remembered there is ANOVA for unequal sample size isn't? but I still have to research on that.

Any tips/suggestions would be greatly appreciated. ^_^

[bugman]

I would still recommend a MANOVA,SPSS copntrols for unequal group size (what are the sample sizes?).

[TheBigEscape]

I would still recommend a MANOVA,SPSS copntrols for unequal group size (what are the sample sizes?).

Thanks bugman for taking the time to reply.

Well, I started with 160 seedlings (before treatment) and randomly grouped them into 4 (1 control, 3 groups of treatments) so there are 40 seedlings per treatment. So after a month of treatment, only:

control - all 40 survived
treatment 1 - out of the initial 40, only 39 survived
treatment 2 - also 39 survived
treatment 3 - 38 survived

I have come across the SPSS site earlier and have decided to read it when I get home.

So far I really appreciate all the replies, and boy! the replies were fast.

[bugman]

I think that the loss of only a couple of seedlings won't effect your analysis. If there were large differences >5 you might need to consider it; but otherwise it looks fine.

There are rules about the sample sizes - but I can't remember off hand.

Phil