EPD's What Do They Mean?

[El_Putzo]

On EPDs, it's possible to have a negative number, correct? Say you've got a bull that's got a bw EPD of -1.5. Would he be better to use on a heifer than a bull that's got a bw EPD of +2.0?

Theoretically yes. That is with all other things being equal as in accuracies.

 

[dun]

On EPDs, it's possible to have a negative number, correct? Say you've got a bull that's got a bw EPD of -1.5. Would he be better to use on a heifer than a bull that's got a bw EPD of +2.0?

That's just one piece of the pie. If calving ease EPDs are available they're a better indicator.
If you look through the semen catalogs you'll find bulls of the same breed that have low BW and poor calving ease and higher BW with better calving ease. That's where the calf shape starts to come into play.

dun

 

[lakading]

Cathy said:

We are letting our desire for mathmatical absolutes outway reality.

And here you show that you don't understand EPDs. Expected Progeny Differences, no mathmatical absolutes.

Until you understand how EPDs work and how they should be used, you'll not be able to use them as they were created to be used. That you choose to not understand them doesn't make them wrong.

Right on Frankie. Also, EPDs are meant to be a selection tool when comparing 2 specific animals and how they would perform in random matings to the same type of cattle, managed the same way, on the same farm.

 

[lakading]

Let's also not forget that a lot of the offspring from "Proven" A.I. Stud bulls and donor dams are pampered so as to sell them as seedstock. High accuracies are not the end point, but rather the beginning. You must then sample these sires and evaluate their progeny's performance for your own management style (whether it be range conditions, pampering, or anything inbetween). That's what makes the cattle business so dang fun. So many variables in our pursuit of great cattle.

Remember, it isn't the destination, but the journey that counts.

Place your emphasis on continually striving to improve your own herd and with a little luck and a lot of hard work, the rest will fall into place.

 

[jaydill]

jaydill wrote:
On EPDs, it's possible to have a negative number, correct? Say you've got a bull that's got a bw EPD of -1.5. Would he be better to use on a heifer than a bull that's got a bw EPD of +2.0?


Theoretically yes. That is with all other things being equal as in accuracies.

That's just one piece of the pie. If calving ease EPDs are available they're a better indicator.
If you look through the semen catalogs you'll find bulls of the same breed that have low BW and poor calving ease and higher BW with better calving ease. That's where the calf shape starts to come into play.

dun

Thanks

 

[ollie']

Another consideration Cathy is where the largest percent of the data is collected. For example if 65% of all the data on Angus bulls is collected in the plain states of Ks, Ne, Sd,Nd,Wy, and Mt, (which I would say is a fair estimate) then that means that the data will be skewed being more accurate in climates and environments like those states and less accurate in North east Tx on fescue even if the accuracies are at .99.

Ollie, I understand what you are saying but EPD's are not meant to tell anyone how a bull will do in a different environment.

In a perfect world, Bull A is breed to 40 cows and Bull B is bred to 40 cows. These cows and their subsequent calves will be kept in the same contemporary group. WHEREVER THEY ARE, be it Texas or Montana, makes no difference, the difference in traits passed on by either bull could be calculated to some degree of accuracy.

Those "DIFFERENCES" should be the same (again, with some degree of accuracy) wherever the bull is used.

Mike , I either don't understand or disagree. If a bull works well in Iowa with a high concentrate diet compared to another bull and you send those bulls to western Tx, I don't think the data would be accurate. For example bull A might be a high growth terminal type sire with a long side and big cannon bone and a 9 frame . Bull B might be a compact, deeper, thicker, more easy keeping bull. Both of these bulls theoretically could have +50 epd's when compared to Bull C who is a high proven sire from a stud. Their averages would stay at +50 as long as the two bulls A and B were used fairly equally across these two environments. If the west Texas cattleman used the gutless, high growth bull in his environment those lanky calves might would dry up and blow away compared to the moderate easy fleshing bull. The high growth bull would blow away our moderated bull on unlimited feed. Their epd's could be the same but they won't work the same on different farms all the time . You (as I know you are aware of) need to select bulls that work for your operation.

 

[Jeanne - Simme Valley]

Ollie & Mike are BOTH right. EPD's have environment taken out of the equation because they indicate DIFFERENCES between bulls. If Bull A is bred to a herd of cows AND NO OTHER BULL has been used in the herd, his offspring within that herd has little or no affect (I think none) on the database because there is no other bull's contemporaries to COMPARE to.
BUT, you have to take into consideration the LIVE animal to fit environments. So Bull A with +50 and Bull B with +50, have to be the right phenotype for the area that he (as a live animal) or his offspring are to be raised.
EPD's are only PART OF THE PICTURE.

 

[Jeanne - Simme Valley]

The Red Angus association has a table that shows what the expected variation is for trait based on the accuracy. While there can alwasy be rogue aberations from them I would expect them to be a pretty reliable source. I would think that all of the associations have a similar resource

dun

I also think they all do. Simmental does.

 

[ollie']

The Red Angus association has a table that shows what the expected variation is for trait based on the accuracy. While there can alwasy be rogue aberations from them I would expect them to be a pretty reliable source. I would think that all of the associations have a similar resource

dun

I also think they all do. Simmental does.

Am I on the wrong page again or are ya'll. I don't think Cathy is commenting on the percent deviation compared to the accuracy of the epd but rather the deviation based on each individuals repeatability (or lack of ). Which is it Cathy?

 

[norriscathy]

The Red Angus association has a table that shows what the expected variation is for trait based on the accuracy. While there can alwasy be rogue aberations from them I would expect them to be a pretty reliable source. I would think that all of the associations have a similar resource

dun

I also think they all do. Simmental does.

Am I on the wrong page again or are ya'll. I don't think Cathy is commenting on the percent deviation compared to the accuracy of the epd but rather the deviation based on each individuals repeatability (or lack of ). Which is it Cathy?

IMO, the deviation between the values of BW, WW, etc is a lot more important than the accuracy range. Can't get excited in any event with 20%, 30%, 40% or even 50% accuracy. Even 50% would seem to mean that you are outside the 4 standard deviation range as often as not. My main question is: What is 1 standard deviation for each value? Not sure if the Associations make this information available or not. With 90% to 95%, I start believing EPD's, but usually by then the bull has long since gone to McDonalds.

 

[MikeC]

The Red Angus association has a table that shows what the expected variation is for trait based on the accuracy. While there can alwasy be rogue aberations from them I would expect them to be a pretty reliable source. I would think that all of the associations have a similar resource

dun

I also think they all do. Simmental does.

Am I on the wrong page again or are ya'll. I don't think Cathy is commenting on the percent deviation compared to the accuracy of the epd but rather the deviation based on each individuals repeatability (or lack of ). Which is it Cathy?

IMO, the deviation between the values of BW, WW, etc is a lot more important than the accuracy range.

But Cathy, as the accuracy goes up the deviation possibility goes down. But it is not linear. I don't think?

 

[lakading]

the possibility of deviation goes down, but the amount of the standard deviation is the same.

 

[MikeC]

the possibility of deviation goes down, but the amount of the standard deviation is the same.

Whatever :roll:

 

[norriscathy]

In my humble and uninformed opinon; no one knows what the EDP numbers mean.

WHAT DOES A +2 BIRTH WEIGHT EPD MEAN?

Don't think it means the bull will actually hit an +2. The statistical odds of hitting one number are probably unbelievably high. More than likely the +2 is the center of a range of numbers. With a standard deviation of 5# our box (range) would be from +22 to a -18 BW (4 standard deviations). If the standard deviation is .5 then our box is from +4 to 0. Which box is it? Don't know; as far as I know that information is not readily available.

WHAT DOES AN "ACCURACY" OF 30% MEAN?

Does this mean they are only 30% confident that their results are correct? I hope not; you'd do better throwing darts at a Ouija Board. I know of no one that what do any kind of statistical work with out a 90% to 95% confidence level that their work was correct. So let's give them the benefit of the doubt on this one. Does the 30% accuracy mean the bull will have a +2 BW EPD within a 30% range; +1.4 to +2.6; kind of similar to standard deviation? I think not; because then a 60% accuracy would mean a range of +.8 to +3.2; actually a worse result than 30%. We all "know" 60% is better than 30%. Therefore is is a reasonable assumption to believe that a 30% "accuracy" is a 30% probability.

A 30% PROBABILITY OF SOMETHING HAPPENING ALSO MEANS THERE IS A 70 % PROBABILITY IT WON"T HAPPEN.

So now we have a +2 BW EPD being anywhere from a -18 to a +22 with a 70% chance it will be nowhere in that range!


Would someone with a little knowledge please help my old and feeble mind understand what these numbers mean!!!!

 

[Frankie]

In my humble and uninformed opinon; no one knows what the EDP numbers mean.

WHAT DOES A +2 BIRTH WEIGHT EPD MEAN?

It means, very simply, if you breed a cow to a bull with a +2 BW EPD, doesn't matter what breed, under what conditions, you would EXPECT her calf would weigh two more pounds at birth than if you bred that same cow to a bull, same breed, with a BW EPD of 0. It's a statistical average, of course, not an exact science, but it's that simple.

Don't think it means the bull will actually hit an +2. The statistical odds of hitting one number are probably unbelievably high. More than likely the +2 is the center of a range of numbers. With a standard deviation of 5# our box (range) would be from +22 to a -18 BW (4 standard deviations). If the standard deviation is .5 then our box is from +4 to 0. Which box is it? Don't know; as far as I know that information is not readily available.

It doesn't matter. The EPD will never, ever tell you what something will weigh. So why are you so concerned about deviations? EPDs only allow you to compare one breeding animal to another.

WHAT DOES AN "ACCURACY" OF 30% MEAN?

It means there's enough information on the EPD database to allow the computer program to assign that accuracy to that animal.

Does this mean they are only 30% confident that their results are correct? I hope not; you'd do better throwing darts at a Ouija Board. I know of no one that what do any kind of statistical work with out a 90% to 95% confidence level that their work was correct. So let's give them the benefit of the doubt on this one. Does the 30% accuracy mean the bull will have a +2 BW EPD within a 30% range; +1.4 to +2.6; kind of similar to standard deviation? I think not; because then a 60% accuracy would mean a range of +.8 to +3.2; actually a worse result than 30%. We all "know" 60% is better than 30%. Therefore is is a reasonable assumption to believe that a 30% "accuracy" is a 30% probability.

Again you're trying to make EPDs much harder than they are. EPDs are built for simple use along with other, older methods of choosing a breeding animal.

A 30% PROBABILITY OF SOMETHING HAPPENING ALSO MEANS THERE IS A 70 % PROBABILITY IT WON"T HAPPEN.

But that's not what a .30 EPD accuracy means. You're simply trying to make them do something they won't do. You can't. You can rail, rant, question, all you want, but EPDs are meant to compare one breeding animal to another. If you breed your cattle to a bull with a BW EPD of 5, you would EXPECT them to produce calves that weigh 5 more pounds at birth than if you bred those same cows, same management, to a bull with a BW EPD of 0. It's that simple.

So now we have a +2 BW EPD being anywhere from a -18 to a +22 with a 70% chance it will be nowhere in that range!


Would someone with a little knowledge please help my old and feeble mind understand what these numbers mean!!!!

In my humble opinion and experience in talking to lots of Angus bull buyers, many, many people know, understand and use EPDs as a selection tool in the cattle business.

 

[Frankie]

How are your math skills. Here's a link to the Model for a maternally influenced trait: http://www.iowabeefcenter.org/pdfs/bif/ ... Evaluation

Do you understand that? I don't. There are similar models for other traits reported as EPDs. Here are some more links:

http://www.iowabeefcenter.org/pdfs/bif/ ... Evaluation

http://www.iowabeefcenter.org/pdfs/bif/ ... 20%20EPDS'

http://www.extension.iastate.edu/Pages/ ... valuation'

Take time to read them and then we can talk some more if you're still confused.

 

[Jeanne - Simme Valley]

Frankie, you are exactly right. EPD's are a simple TOOL to compare bulls to be used on the same cows.
Anyone not wanting to use them can complicate them and make them confusing, but they are not an exact science (although the methods to arrive at them are very complicated in my feeble little mind).
You have a great way of answering questions. Thanks

 

[Frankie]

Frankie, you are exactly right. EPD's are a simple TOOL to compare bulls to be used on the same cows.
Anyone not wanting to use them can complicate them and make them confusing, but they are not an exact science (although the methods to arrive at them are very complicated in my feeble little mind).
You have a great way of answering questions. Thanks

Thank you, Jeanne. I don't spend a lot of time worrying about how EPDs are created. It's too complicated for me, too. EPDs work. Most people I deal with understand them and that they work. I know that a lot of people don't use them, but that's their choice.

 

[norriscathy]

Frankie & Jeanne; I am not questioning how EPD's are calculated, I'm sure the methods used are the finest and most sophisticated available. I question what the answers mean. I agree that the +2 BW is a statistical average; but an average of WHAT? A BW range of 0 to +4 has an average of +2; but so does a BW range of -18 to +22. What BW EPD range can we expect of this bull?

Sorry to burst your bubble, but if something is 30% "accurate" the reverse is also true! It is also 70% "inaccurate".

 

[lakading]

Sorry to burst your bubble, but if something is 30% "accurate" the reverse is also true! It is also 70% "inaccurate".

That's why you have to evaluate the standard deviation and the associated possible change with any given accuracy. Statistics don't lie. EPDs are not an absolute, rather a selection TOOL based on statistical data. EPDs do work if you know how to interpret and use them.