From

Location

Message

Date Posted

Grittyville, WA

K for the nerds out there, using the 2145 and 2230 as an example, I notice that:

In 2016 the average weight off the 2145 was 1393 lbs. The weight average weight off the 2230 was nearly the same at a hefty 1384 lbs average. I think both were consider to be seeds with top potential for 2016, although considering the Wallace being the more well-known brand at the time, and perhaps the McMullen being more widely available, and the Wallace being the larger pumpkin, it did sell for more. But my guess here is that people felt approximately the same about the potential of both seeds (remember this was in 2015/2016 when neither had been planted yet).

And in that first year out, 2016, the McMullen did (barely) outperform the 2230 Wallace. The McMullen yielded a 1393 lbs average and the wallace yielded a 1384 lb average.

The Wallace was grown 32 times, the McMullen, 35. Which is good evidence that people did feel about the same about both seeds in early 2016.

So, what happened in 2017...

Well, after the 2624.6, (which is included in the 2016 average progeny weight of 1393) the price of the McMullen went way up. The price of the 2230 went down. The McMullen got planted more, and the Wallace got planted less. Now here's whats weird:

In 2017 the average grower got out 1542 lbs out of the McMullen. But average lbs that the average grower got out of the 2230 Wallace decreased to 1181. (This figure includes ten growers and two 1900+ lb pumpkins...)

Its the same seed as it was back in 2016, so what changed?

11/20/2017 6:28:20 PM

Grittyville, WA

title is supposed to read "untangling one of the untangibles"

11/20/2017 6:29:28 PM

Grittyville, WA

In both cases, the seed did not change. Something else changed... Attitude about the seed?

Can you say positive feedback loop?... Echo chamber?.. Self-fulfilling prophesy?

11/20/2017 6:32:17 PM

Grittyville, WA

People, and stats, can be arbitrary and biased so don't read too much into impressive stats. Next year the stats will be even better, but its still the same seed, and for all we know its the same as the 2230.5 Wallace. We became biased and then this bias reinforced itself and it will continue to do so next year. Before our own bias for the 2145, the average progeny weight from this seed was much closer--no different really-- than that of any other reasonably good seed. Keep that in mind.

Basically, if you ignore all my posts about stats you will be safe lol...

11/20/2017 7:01:28 PM

Bloomfield, Iowa

I'm growing both of these seeds next to each other in 2018. That will be a good test.

11/20/2017 7:55:25 PM

Walhalla, ND.

I think the change came from two factors. 1- the more experienced growers purchased the more expensive seed I.E. 2145. The least experienced went with the not so expensive seed I.E. 2230. A quality seed that's in the most experienced and seasoned grower plot will result in a heavier fruit.

11/20/2017 8:20:24 PM

Deer Park WA

Agree with Hobbit 100%

11/20/2017 9:59:20 PM

Grittyville, WA

One other stat: On the whole, depending how it’s calculated, growers improved by 1-3% between 2016 and 2017. The 2145’s average progeny weight increased by 10%. The 2230.5 went down 15%. So maybe around 25% of the performance of a seed can be attributed to these arbitrary, subjective influences like who planted the seed and did it get the best spot in the patch. Also & perhaps worst for the stats, did growers plant more than one and leave their duds at home...

Overall, I’m impressed by the collective knowledge of, and individual & collective intuition of growers. Feel free to email me about pumpkin stats wsu_brandon@hotmail.com

11/20/2017 11:52:02 PM

Lexington, Ky

To make any real sense of the data, you would need to plot grower location for both seasons and factor in the weather as well.

11/21/2017 5:15:00 AM

Western NY

Total overall average can be a dubious marker.

I would run the numbers comparing the average of the top 10, 15, or 20 fruit for each for example.

11/21/2017 6:58:49 AM

Willamette Valley, Oregon

I like the 75th percentile (the top 25%) of offspring weights as a measure of mother-seed quality, because it focuses on the better end of things in a fair and balanced way. (The downside, of this, and other methods is that you'd really want at least 10-15 seeds grown by random growers...and we'll never have that.) In this assessment 2230 ranks 2nd and 2145 ranks 6th.

Analyzing parent quality by different methods tends to give different rankings, and that alone points to the extreme variability and uncertainty in doing these kinds of "proven seed" assessments. We really can't say very much about which seeds are the best even when using sound statistical methods.

11/21/2017 11:35:04 AM

Willamette Valley, Oregon

One trap to be careful of is looking at the largest fruit grown for a seed, or say, the number of fruit over 1800lbs for a given seed. This is a highly biased and generally terrible way to assess seed quality. These types of assessments are highly dependent on the total number of fruit from a given seed making it to the scale, which is essentially a grower's choice popularity contest and not an unbiased seed quality assessment. With 75+ fruit making it to the scale, the 2145 has probability on it's side.

As a quick test, I picked 20 groups of 75 fruit at random from the GPC list and the average maximum weight in each 75-fruit group was just short of 2000 lbs and ranged from 1657 to 2363lbs. That indicates that the 2145 is perhaps only slightly better than we'd expect any random seed to be for that metric.

11/21/2017 11:51:49 AM

Herd mentality ....just ask Joze and there's something to that and it plays out in the auctions.
Those super hot seeds do not come along every year

11/21/2017 3:17:59 PM

Grittyville, WA

Agree with bathabitat (he's a top level grower btw)... For top level growers, it would be foolish to not look at (& focus on) the top level results... 1900+ lbs??? I haven't done this exactly... I've been trying to work towards a method of analysis that could pry into the "what's the best seed for the average grower" question.

I guess I should be asking: is there anything in the data that I don't know, or that goes against an assumption I've made?

11/25/2017 4:03:42 AM

Middleton Ma

I can say I am sorry to say I have grown a 2145 for 2 years in a row . We have made two nice crosses that we will now grow. But I am responsible for keeping this seed from being one of the best seeds ever grown. So to establish a chart you should also now apply a grower ratting to each seed to come up with an average.so maybe i am grower rating minus 60 you now have a great seed 🎃

11/25/2017 10:50:27 PM

Massachusetts

Yes Don and Jeanne. In my case top growers call me every spring begging me not to grow their seed. :=}

11/26/2017 6:14:16 AM

Grittyville, WA

lol no worries i looked at the curves on these datasets and played around with dropping some of the lowest numbers, it doesn't affect the stats too much really though to include or exclude any single entry. Every seed has a very uniform bell curve, almost entirely regardless of who grows them, the only seeds which are really affected by the who-grows-them handicap are the orange ones grown only by people wanting to grow for prettiness. Everyone else is growing for size, and if you are too, and the low stats are actually necessary to give a realistic data curve. Lol though.

11/27/2017 3:32:01 AM

Grittyville, WA

I know Tom K you are not a low stat... you're just not as high as you'd like to be:)

11/27/2017 3:34:25 AM

Willamette Valley, Oregon

Glenomkins - I don't see anything fundamentally wrong with the math on your seed assessments. Averages or medians are a fair assessment, unaffected by sample size.

My comments about the biasing effects of sample size are more of a warning to others who might be trying to assess in ways that aren't sound math.

The real problem is that the observational (non-experimental) nature of the pumpkin database makes any meaningful conclusions very tenuous, even when the math is fair and unbiased. Although it's still fun to do.

11/27/2017 4:05:20 PM

Grittyville, WA

Agreed-- its all with a lot of salt... Looking at data is almost like looking at a pile of random seeds on the table and then you ask, so which is the best one?"

11/28/2017 7:05:15 PM