|
 Essay - mar/abr 2007
Statistics, appoximations and close truths I
James C. Delouche
Professor Emeritus Mississippi State University
Lessons in a Rural Courtroom
In the mid-1960s I was sitting in the witness room in a county courthouse in the Delta region of Mississippi. It was August, the non-air-conditioned room was hot, and I was nervous. I had been subpoenaed as a witness in civil litigation between a cotton farmer and seed company.
The farmer claimed that cottonseed he purchased from the company did not produce an adequate population of plants which resulted in a total crop failure because it was too late for replanting and conditions were unfavorable. I became involved in the litigation as a witness because as the Director of the State Seed Testing Laboratory my signature was on the report of tests made on an "official inspection sample" of seeds from the seed lot in dispute.
This was not the first time I had been summoned as a witness in civil litigation involving seeds. Usually, however, I was informed a few days before the trial date that the litigants had reached an out-of-court settlement and/or the case was dismissed. But, I realized this was not to be one of those times as the bailiff opened the door of the witness room and called me into the courtroom. The plaintiff's attorney hurried through the formalities of introducing me and establishing my credentials to the court and jury, then declared that I was a hostile witness and thus my responses were to be mostly limited to YES or NO unless otherwise allowed.
The attorney's questions and my responses quickly established that the seeds purchased and planted by the farmer and the seeds in the official inspection sample tested in the State Seed Laboratory were from the same batch (lot) of cottonseed. He then pointed out that the label on the seeds purchased by the farmer stated that the germination was 80% while the germination on the report of the inspection sample was only 73%.and asked why the seed lot was not cited for mislabeling. "You can answer," he said, so I did. "The seed lot was not considered mislabeled because the labeled germination was within the allowed tolerance with the germination found in the official test. "Allowed tolerance, you say, just what is that?" "It's mostly a statistical parameter based on the variance associated with sampling," I responded. He continued his questioning, "If the germination of the official test had been 71% would the lot have been cited for mislabeling." "Yes," I replied, "71% would have been lower than the allowed tolerance." "Professor, I believe you are telling this court then that seeds germinating 73% are OK but seeds germinating 71/% are not OK, that the difference in germination in the first case is allowed but not when germination is just 2% less. Are you really admitting that seed testing is not an exact science? Answer YES or NO." "No, I am not." I protested. But, he went on: "You say NO but we all know that the answer should be YES because if it were an exact science it would not produce such funny results and unreasonable answers." He continued, "Does the 73 or 80% germination, since you say they are the same, of cottonseed assure my client or any farmer planting such seed that an adequate stand (population) of plants for a good crop will be obtained? Answer YES or NO." "No," I answered. Turning to the jury the attorney exclaimed, "If 80% germination does not assure the buyer that he will obtain a good stand, we are left to wonder just what it does assure, what it does mean." Then turning back to me he asked, "Do you agree that the requirements under our state seed law can be considered mainly truth-in-labeling provisions? YES or NO." "Yes," I agreed. He continued, "Let us then go directly to the bottom line of this case. Since you are a learned professor, I want you to respond to this court to a simple multiple choice proposition: To the plaintiff, Farmer XX, the 80% germination on the label of the seeds he purchased and planted that failed to produce more than a few hundred plants per acre represented a) a truth,, b) a close truth, c) a distant truth, d) none of these. You need not respond professor. I withdraw the question. Your Honor, I have no further questions for this witness."
The defendant's attorney said he did not have any questions so I was dismissed and returned to the witness room angry at the plaintiff's attorney for his unfair questions and comments and at the defendant's attorney for not giving me a chance to display my intelligence rather than my stupidity on cross examination and grievously embarrassed by my performance. (Note: the jury decided for the farmer but the verdict was later overturned on appeal on the basis of belately uncovered evidence that several other farmers had purchased and planted seeds from the same lot without problems.)
I spent much time and space on the experience related above because it is typical of the many interesting, unpredictable, exasperating but always learning experiences I have had involving seeds, skilled lawyers, smart farmers, responsible seed producers and companies, as well as some of the various sorts that were not at all trustworthy and even devious. Remembering and reflecting on these experiences provided the themes for the previous and present essays and perhaps the next several ones.
The previous essay dealt with some of the tests used to describe and characterize seeds, i.e., seed lots. This and several follow-up essays will examine the sources of variability that influence the "structure" or composition of seed lots, the approximation procedures used to achieve the desired structure, the statistical methods and parameters used to establish and monitor the quality components of seed lots and some of the close truths involved in describing and characterizing them.
Seeds are Different
Seeds are different from the other inputs for crop production. Seed producers usually do not have the rigorous control over the composition of the seeds produced that is possible for other inputs such as fertilizer, pesticides, irrigation water, even credit. The composition of seed lots is strongly influenced by natural forces and circumstances, i.e., biological, edaphic and climatic, that vary among locations and times and are subject only to a moderate degree of control. Thus, most seed lots consist of a principal population of pure germinable seeds with the desired inheritance that can be and usually is mixed with several to many undesirable sub-populations of other seeds and particles. The main sub-populations are:
1) seeds identical to those in the principal (desired) population except in germination, vigor, disease presence (physiological and pathological categories);
2) seeds similar in appearance but significantly different genetically (seeds of other varieties);
3) seeds different in appearance and physical properties such as size, density, shape (other crop and weed seeds); and
4) debris from seeds, plants and other field materials gathered in the harvest (inert matter).
The proportions of the various sub-populations in seed lots can vary among different production fields and within a seed lot from the same field because of differences in climate, fertility, soil moisture, weed populations, cropping history, the time of planting and in other ways. The type and proportions of undesirable material in seed lots can also be affected by the harvesting equipment used, the time of harvest, and the skill of the operator .There are many instances of soybean seeds severely damage by untimely, incautious harvesting, cottonseed ruined because of improper adjustment of the mechanical harvesters, and peanut seeds transformed mostly into inert matter by improperly set-up and adjusted shellers.
The post-harvest operations for seeds especially aeration, drying and handling, are critical in maintaining the portion of the principal population of pure, germinable and vigorous seeds in the seed lot. Delayed and inadequate aeration or drying of seeds to a safe moisture content decreases the portion of the desirable seeds in the lot and increases the amount of dead and/or low vigor seeds. Likewise, handling and conveying seeds with poorly maintained equipment and/or machines that are unsuitable for seeds can dramatically increase the amounts of the undesirable components in the seed lot, e.g., inert material, dead and damaged seeds.
The procedures and protocols recommended and followed in seed production, harvesting, drying and processing, i.e., to prepare the seeds for marketing, are basically employed to increase and enhance the desirable seeds component in a seed lot. Most of them are not designed or expected to produce exact or certain outcomes but rather to produce results that are approximations which, although inexact by definition, are achievable and adequate. Seed lots are subjected to a series of steps and/or operations that incrementally change the composition of the seed lot in the desired direction to the extent that is technically and/or economically possible, i.e., to the point of approximation. The effectiveness of the several procedures and operations involved in accomplishing this objective and the approximate results achieved by any one or all of the procedures used are monitored by statistically based quality assurance measures and quality tests.
The next essay will look at some of the technical and economic limitations of the protocols and operations used to change the composition of seed lots in the desired direction and the procedures and tests employed to monitor and establish their effectiveness.
|
|
............
|
