How do I apply the right statistic test?

[Zeraus]

Hello everybody,

 

I have a problem with statistics. I don't get it.

 

I have realized a experiment to test the effects of three proteins in five different concentrations. I had three repetition of each condition, that is, for each protein and concentration. At the end I had 45 cell cultures. I want then to compare the results and effects of these three proteins on the cell. Which kind of statistically test should I use? I repeated the whole experiment one more time, so what if I want to use the results of both experiments?

 

Thank you very much.

 

Zeraus

[John Cuthber]

I'd look at this

http://en.wikipedia.org/wiki/Analysis_of_variance

but it depends on having some stats package to do all the arithmetic for you.

[Zeraus]

I'd look at this

http://en.wikipedia.org/wiki/Analysis_of_variance

but it depends on having some stats package to do all the arithmetic for you.

Thanks John,

 

I will read it.

[studiot]

You need to develop a hypothesis to test.

 

For instance that the measured effect is independent of concentration, normally distributed over some range of concentration or whatever.

 

That is what John's analysis of variance will do for you.

 

Better advice will come if you can state your hypothesis or hypotheses.

 

Better still, next time, formulate your hypotheses before carrying out the measurements.

[Zeraus]

You need to develop a hypothesis to test.

 

For instance that the measured effect is independent of concentration, normally distributed over some range of concentration or whatever.

 

That is what John's analysis of variance will do for you.

 

Better advice will come if you can state your hypothesis or hypotheses.

 

Better still, next time, formulate your hypotheses before carrying out the measurements.

My hyphotesis would be if one of the proteins has a stronger effect than the other one.

 

I really don't know much about statistic. I have heard one need many experiment to do a T-test, what is normally used. I have two experiments. In each one three repetitions of each condition were made. Is it enough?

[Dislayer]

To me it sounds like you have repeated the experiment in each condition 6 times (3 repetitions done twice) is that right? If so you have an n of 6 which is a good sized number to start with. All of the values of what ever you are measuring will need to be averaged and you will need to calculate a standard deviation.

 

To compare the difference between the five different concentrations in one protein (asking the question how does the increasing concentration of a given protein affect the cell) you would need to use a one-way ANOVA. To test the difference between a given concentration of protein between the other two proteins (how does the maximum concentration of protein A affect the cell compared to protein B and C?), a one way ANOVA would be used too. The one way ANOVA would give you differences between individual concentrations or proteins but it would also let you know if the general trend is significant (ie. does the proliferation of the cell increase as the concentration of the protein increases? In a dose dependent manner)

 

You still need to identify what variable you are testing though. Is it the effect of the proteins on cell proliferation? Growth? Secretion of another protein? Cell death? Once you identify what variable you want to test you will need an appropriate control (treatment with a vehicle) to determine significance against.

 

Generally you can download a stats program (something like Prism Graphpad or Sigmaplot). Don't try to use excel stats, they are not correct often and are very limited in their use.