Real Real Results

[dontknowanything]

http://thriplow-farms.co.uk/2017/04/24/real-real-results/

 

[Mr Charisma]

Looks very interesting ,,,,, how are things looking at the moment ,,,,,, I gather things are fairly dry in the land of milk and honey !!!!!!

 

[Gong Farmer]

Point is BASF or whoever wouldn't get 50 farms top do that, it's very time consuming for the farmer. Keeping it simple may affect the statistical validity but at least it means you get a good number taking part.

 

[dontknowanything]

Point is BASF or whoever wouldn't get 50 farms top do that, it's very time consuming for the farmer. Keeping it simple may affect the statistical validity but at least it means you get a good number taking part.

I think I said pretty much exactly that?!?!

 

[dontknowanything]

Looks very interesting ,,,,, how are things looking at the moment ,,,,,, I gather things are fairly dry in the land of milk and honey !!!!!!

Yes, very dry, but crops are still holding on, no leaf curling etc yet. Things are OK in general, for the time being.

 

[Hampton]

Point is BASF or whoever wouldn't get 50 farms top do that, it's very time consuming for the farmer. Keeping it simple may affect the statistical validity but at least it means you get a good number taking part.

Exactly, and even if the results don't turn out in Basf favour, and they don't recruit any farms under 1000acres, then they have detailed knowledge of 50 farms, farmers and well over 50000 acres!
It's absolutely brilliant from their point of view and yet most farmers see only the free chemical bit.

Incidentally David, I really like your trial idea, way more relevant than the real results nonsense

 

[yellow belly]

That's why I joined the Lincolnshire cereal centre when the original Cotswold cereal centre expanded to the whole country
Now merged with Niab as niab tag independent trials

The trial system used is randomised block system
looking at one trial at one location is always less accurate it does not matter how large the plots are if there are not enough replicates

 

[dontknowanything]

Exactly, and even if the results don't turn out in Basf favour, and they don't recruit any farms under 1000acres, then they have detailed knowledge of 50 farms, farmers and well over 50000 acres!
It's absolutely brilliant from their point of view and yet most farmers see only the free chemical bit.

Incidentally David, I really like your trial idea, way more relevant than the real results nonsense

I don't think this is a totally cynical exercise from BASF, it's also backed by ADAS & YEN so hopefully the data can be used for farmers' gain as well.

 

[Breakthru]

The quality of statistical analysis used in small field trials in UK has been questioned in the past and so the move to larger plots is an improvement IMHO.
I always used tramline sized plots for my own benefit and know lots of others who have done exactly that.

 

[Clive]

I hate small plot trials

To my mind the bigger the plot the lower the potential for error is or at least the more the errors and in field variation are diluted

I can't see any logic behind a smaller plot being more accurate than a big one

 

[dontknowanything]

I hate small plot trials

To my mind the bigger the plot the lower the potential for error is or at least the more the errors and in field variation are diluted

I can't see any logic behind a smaller plot being more accurate than a big one

A 0.1ha plot is not more accurate than a 1ha plot. But ten 0.1ha plots will give you a more accurate result than one 1ha plot.

You can disagree with it it you want, but sample size is totally critical in getting statistically valid results.

 

[dontknowanything]

The quality of statistical analysis used in small field trials in UK has been questioned in the past and so the move to larger plots is an improvement IMHO.
I always used tramline sized plots for my own benefit and know lots of others who have done exactly that.

The use of larger plots in the Real Results trial is great thing, but it doesn't give good results for individual farmers.

Casting aspersions by saying things "have been questioned" is just weasel words, unless you want to back up what you mean by that?

 

[Clive]

A 0.1ha plot is not more accurate than a 1ha plot. But ten 0.1ha plots will give you a more accurate result than one 1ha plot.

You can disagree with it it you want, but sample size is totally critical in getting statistically valid results.

I don't disagree at all - your last paragraph is exactly what I'm saying

So 2 ha is more accurate than 1ha even if the 1ha is made up of 10 x 0.1 ha sub plots

If you don't agree with me ask someone with a degree in statistics, that's what I did and they concurred bigger sample size always = less error or influence of variation over result

You are proposing lots of small plots which is great but the total of those small plots is still smaller than the large plots that real results uses

Why not harvest your entire plot areas after you have taken the small sub plots out and compare the results - that way as well as trialling fungicides you are testing methodology here as well ?? That would maybe be a more interesting trial than the fungicides !

 

[Breakthru]

The use of larger plots in the Real Results trial is great thing, but it doesn't give good results for individual farmers.

Casting aspersions by saying things "have been questioned" is just weasel words, unless you want to back up what you mean by that?

One was a professor from an Ag Uni. Why have so many people done large scale trials that fail to live up to small trial claims.

 

[B'o'B]

Exactly, and even if the results don't turn out in Basf favour, and they don't recruit any farms under 1000acres, then they have detailed knowledge of 50 farms, farmers and well over 50000 acres!
It's absolutely brilliant from their point of view and yet most farmers see only the free chemical bit.

Incidentally David, I really like your trial idea, way more relevant than the real results nonsense

They get details about one field on the farm (17Ha in my case and they have no idea how typical that field is in relation to the rest of the farm). I can also vouch that they have recruited at least one farm under 1000a.
I for one think it is good to get involved with these tramline trials (I'm part of the LearN project). These trials don't happen without an added hassle factor with 4 different mixes and therefore 3 rinse outs, it also makes combining the fields a pain to do it right but I have gained knowledge and confidence from taking part in the LearN project and I hope taking part in this will add to that, maybe it will maybe it won't but that's a risk I'm willing to take.

 

[dontknowanything]

I don't disagree at all - your last paragraph is exactly what I'm saying

So 2 ha is more accurate than 1ha even if the 1ha is made up of 10 x 0.1 ha sub plots

If you don't agree with me ask someone with a degree in statistics, that's what I did and they concurred bigger sample size always = less error or influence of variation over result

You are proposing lots of small plots which is great but the total of those small plots is still smaller than the large plots that real results uses

Why not harvest your entire plot areas after you have taken the small sub plots out and compare the results - that way as well as trialling fungicides you are testing methodology here as well ?? That would maybe be a more interesting trial than the fungicides !

Sample size means the number of samples taken, not the physical size of the plot! That's like saying in a medical trial a result from a fat person weighing 200kg is better than from two people weighing 80kg each.

https://en.m.wikipedia.org/wiki/Sample_size_determination

So your person with a stats degree is talking about having more numbers of plots, not bigger sizes.

As for combining the whole plots, it does mention in my blog that we will be doing that as well as using a plot combine.

 

[Clive]

Sample size means the number of samples taken, not the physical size of the plot! That's like saying in a medical trial a truly from a fat person weighing 200kg is better than from two people weighing 80kg each.

https://en.m.wikipedia.org/wiki/Sample_size_determination

So your person with a stats degree is talking about having more numbers of plots, not bigger sizes.

As for combining the whole plots, it does mention in my blog that we will be doing that as well as using a plot combine.

No - bigger sample size in this case mean more ha compared sinple as that imo (and my mate with the stats degree !)

Glad your going to combine entire plots as well though as it will be a useful comparison to your multiple small plot method

 

[dontknowanything]

One was a professor from an Ag Uni. Why have so many people done large scale trials that fail to live up to small trial claims.

I'm not doing a small plot trial, they are a third of a hectare each. You're getting confused between the statistics of sample size, and the problems with using physically small plots and trying to relate that to farm scale.

 

[dontknowanything]

No - bigger sample size in this case mean more ha compared sinple as that imo (and my mate with the stats degree !)

Glad your going to combine entire plots as well though as it will be a useful comparison to your multiple small plot method

No, sample size is a well defined statistical term, and it most certainly does not mean the physical size of the plots.

Definition: "Sample size determination is the act of choosing the number of observations or replicates to include in a statistical sample."

Id go back to your stats friend and get him to explain it a bit better to you.

@Feldspar

 

[Clive]

No, sample size is a well defined statistical term, and it most certainly does not mean the physical size of the plots.

Definition: "Sample size determination is the act of choosing the number of observations or replicates to include in a statistical sample."

Id go back to your stats friend and get him to explain it a bit better to you.

@Feldspar

I understand what he's saying and it makes sense to me (might try get him in here actually !)

You are honestly saying a trial over a total of 1ha 10 x0.1 ha sub plots is more accurate and less subject to variation than the same trial over a 10ha plot ?

More ha you study the less significant variation becomes, it is that simple

Your over analysing something that's actualy pretty simple imo

We have been brainwashed by small plot trials done by men in white coats justifying their pay packet over the years imo they only do small plots because they can't do bigger onse affordably

What's wrong with a real result at field scale from a farmer under all the constraints and variation that real world situation brings ? That's how we all farm in reality and not in 0.1ha squares !