EDITING LACTATION RECORDS FOR USDA-DHIA GENETIC EVALUATIONS

DAIRY HERD IMPROVEMENT
1985

WIGGANS, G.R., & WAITE, L.G.
VOLUME: NCDHIP HANDBOOK

Lactation records of cows enrolled in the National Cooperative
Dairy Herd Improvement Program (NCDHIP) testing plans are eligible
for use in the U.S. Department of Agriculture-Dairy Herd
Improvement Association (USDA-DHIA) genetic evaluations. Those
records are submitted to USDA's Animal Improvement Programs
Laboratory (AIPL) by nine dairy records processing centers (DRPC's)
in format 4, a computer format that contains descriptive
information about a lactation. This information includes the cow's
yield for milk and components, identification of the cow and her
relatives, and the cow's reproductive status. Records submitted to
USDA are edited extensively to ensure the accuracy, integrity and
consistency of lactation data and of genetic evaluations for which
the data are used.

A multistage screening process determines which records are
included in USDA-DHIA genetic evaluations. Records are eliminated
completely from the USDA genetic evaluation system for the
following general reasons.

1. Record is of a category not used in calculating genetic
evaluations; for example, the testing plan was not approved for
genetic evaluations, the sire was unidentified or the lactation was
greater than 305 days in milk (DIM).

2. Data values are outside of the limits.

3. Values conflict with those from previously received records.

Of the remaining records, some are not used for genetic
evaluations but do remain on file in the USDA genetic evaluation
system. These records are not used for computing genetic
evaluations because the basis for comparison with other records is
insufficient, for example, too few contemporaries, calving date
before 20-year period from which records are selected,(2) fewer
than 15 days in milk or birth date too close to sire's birth date.
Crossbred animals are also in this group.

Records eliminated because they fall in a category not used for
calculating genetic evaluations (Table 1) are not returned to
DRPC's. Lists of records eliminated because they are outside limits
or conflict with previously received records (1060 errors) are sent
to DRPC's on tape or microfiche. These lists enable DRPC's to
explain to dairy producers why records were eliminated from genetic
evaluations. If a record is not used because of insufficient basis
for comparison, it remains on file and can be used if more
information on contemporaries becomes available. These records are
identified as not usable on bull evaluation and daughter reports
(form 1202).

Records Eliminated Because Data Values Are Outside Limits or
Conflict

Errors that result because data values are outside limits or
conflict with values in previously submitted records are coded and
returned to DRPC's either on computer tape or microfiche. Up to
four of the following code types may be listed:
A,B,C,D,E,F,G,I,J,L,M,N,O,P and S.

Type A errors are identification conflicts with herdbook data
from breed associations or with other lactation records (Table 2).
A conflict of identification data between two or more lactation
records (code A5) is one of the most frequent errors in records
submitted to USDA. Many times the error is really in the cow's
number. What appears to be two conflicting records of one cow is
really records for two different cows.

----------------------------------------------------------------------
PLEASE NOTE: THE FOLLOWING TABLE IS WIDER THAN THE SCREEN. USE THE
RIGHT ARROW KEY TO VIEW THE RIGHT SIDE OF THE TABLE. TO
PRINT THE ENTIRE TABLE YOU MUST FIRST EXPORT THE TABLE
AND USE A WORD PROCESSOR OUTSIDE THE RETRIEVAL SYSTEM.
----------------------------------------------------------------------

TABLE 1. Categories of records eliminated from calculations of USDA-DHIA
genetic evaluations.

Category Edit criteria
Record value unacceptable Invalid character in field.
Identification unacceptable All zeros or contains a blank in
positions 3 through 9 or a character
other than a number in positions 6
through 9.
Cow identified Unequal breed codes for cow and dam.
Sire
Unidentified All zeros or "UNK" in positions 3 to
5.
Number unacceptable Ear tag number for sire of
registered cow.
Breed unacceptable
Registered cow Unequal breed codes for cow and dam;
Unequal breed codes for cow and
sire.
Grade cow Unequal breed codes for cow and sire
if cow not coded as X (crossbred)
unless breed of cow is A, B, G or M
and animal prefix code is 36 or 37
to denote identity-enrollment
status; Breed code X for cow and
equal codes other than X for sire
and dam; Breed code U for sire.
Herd location unacceptable Non-U.S. location code in herd code.
Nonofficial dairy recordkeeping plan not Type-of-test code 40 or higher.
approved for calculating USDA-DHIA
genetic evaluations
Outdated information for record in Calving date more than 15 months
progress before run date; More recent record
received for lactation.
Calving date unacceptable All zeros for month, day or year.
Lactation length unacceptable Lactation of less than 7 days in
milk (40 days in milk for records in
progress) or more than 305 days in
milk.
Record unusable for genetic evaluations Termination code 9.
progress before run date; More recent record
received for lactation.

Calving date unacceptable All zeros for month, day or year.

Lactation length unacceptable Lactation of less than 7 days in
milk (40 days in milk for records in
progress) or more than 305 days in
milk.

Record unusable for genetic evaluations Termination code 9.

Type B, C, D, E, F and G errors involve questionable or invalid
data values for birth date (type B, Table 3), cow identification
number (type C, Table 4), dam identification number (type D, Table
5), breed designation (type E, Table 6), calving date (type F,
Table 7), and prefix code for eartag identification number (type G,
Table 8).

Among birth date errors, code B4 (less than 9 months between
birth dates of 2 daughters of the same cow) often involves cows
that (1) actually are twins but are not coded as such in the
lactation records or breed association data or (2) result from
embryo transfer and are not so coded by the breed association. For
registered cows with successive identification numbers and the same
sire, dam and herd, a twin code is assumed. Another common
situation is an incorrect dam identification number. This usually
applies only to grade animals because the authenticity of
registered information is checked against herdbook data if
available.

Cow identification numbers outside limits for the breed (error
code C1) often involve an incorrect breed code for the cow. USDA
developed a check against high and low registration numbers for the
reported breed to avoid assigning registered cows to the wrong
breed.

Cows rejected because dam identification number is outside limits
for the breed (error code D1) include those with numeric private
dam numbers of grade cows, which usually are only one to four
positions in length. Because the number contains no alphabetic
characters, it is classified as a registration number. All eartag
numbers must include alphabetic characters. Eartag numbers that do
not have a valid prefix code in the first two positions also are
rejected.

For many of the calving date errors (type F), the real problem
may be in some area other than calving date. Although the calving
dates would be correct for these records, another error still would
cause them to be rejected. An age at calving of 14 months or less
as calculated from reported calving and birth dates (error code F4)
may have resulted from an incorrectly reported birth date. If a cow
has two records with calving dates less than 9 months apart (error
code F5), two cows may have been reported with the same eartag or
registration number. Another cause for an F5 error is a record that
was started with an abortion but was reported without the proper
termination code; also, the previous record was not reported as
terminated by an abortion. One apparent cause of two of a cow's
records with identical calving dates but differing yield (error
code F6) is the reporting of corrected yield data without
correction code 1 in position 87 of the lactation record. Records
received within the 6 months preceding a genetic evaluation run are
reviewed, and only the most recently calculated record for each
herd-cow-calving date combination is retained. The sequence of
herds by days in milk is respected. Therefore, a cow could transfer
back to an earlier herd without losing her earlier record. Although
a cow with more than 20 lactation records (error code F9) is
possible, usually more than one cow is involved because of improper
coding. Therefore, only the first 15 lactations of a cow are
included in genetic evaluations.

TABLE 2. Type A errors: identification conflicts with herdbook data from breed
associations or with other lactation records.

Error code Conflict
A1 Two or more identification fields (sire, dam, birth date) in
record conflict with herdbook of breed registry association. 1
A2, A3 Record eliminated because identification information disagrees
with identification previously verified as being correct. 2,3
A4 Conflict of identification data (sire, dam, birth date) between
two lactation records of cow but both previously verified. 3
A5 Conflict of identification data between two or more lactation
records of cow, neither verified nor marked with a correction
code. 4

1 If only one field (sire, dam or birth date) in lactation record conflicts
with herdbook, lactation record is changed to conform with herdbook. Records
adjustable in this manner are not eliminated. Format-20 records are sent to
DRPC with an indication of conflicts with pedigree data. For grade cows with
conflict for dam number in two records of the same cow, both dam numbers
are converted to zeros if sire and birth dates of the two records are
identical and non zero.
2 Original distinction between codes was source of verification, which no
longer is meaningful.
3 Code 2 in position 31 of format 4.
4 Also includes conflicts between incoming record and previously reported
records in USDA master files.

TABLE 3. Type B errors: questionable or invalid birth date.

Error
code Questionable or invalid data
B1 Birth year higher than year in which record was received by USDA.
B2 Birth month either 0 or greater than 12.
B3 Birth day either 0 or greater than 31 for a registered cow.
B4 Less than 9 months between birth dates of two daughters of the
same cow and no indication of multiple birth.

TABLE 4. Type C errors: questionable or invalid cow identification number.

Error code Questionable or invalid data
C1 Registration number out of range for breed. 1
C2 Invalid "new series" eartag number. 2
C3 Registered Brown Swiss or Red Poll coded as foreign. 3,4
C4 Numeric identification with 99 in first two positions.

1 Each cow registration number automatically is checked against an upper and
lower limit appropriate to the breed. The lower limit is chosen sufficiently
small to screen out only the most obvious reporting errors. The upper limit
normally is the highest number issued by the breed association to present
date.
2 Eartag number must include nine positions with a prefix code in the first two
positions, alphabetic characters in the next three positions, and only
numerals in the last four positions.
3 First two positions = 04.
4 No Canadian registry associations for Brown Swiss or Red Poll; no
importations from other countries allowed at present for these breeds.

TABLE 5. Type D errors: questionable or invalid dam identification number.
Error code Questionable or invalid data
D1 Dam identification number out of permissible range for breed. 1
D2 Registered cow reported with either a zero or grade dam
registration number.
D3 Dam's registration number higher than cow's registration number
and no verification code present in records. 2
D4 Dam's eartag number invalid. 3
D5 Registered Brown Swiss or Red Poll cow has dam coded with foreign
registration number. 4,5
D6 Dam identification number is numeric with 99 in first two
positions.
1 Each registered dam identification number compared with upper and lower
limits appropriate to breed.
2 If situation valid, verification code 2 must be entered in position 31 of
format 4, or pedigree information that verifies higher dam registration
number must be received from breed association.
3 Eartag number must include nine positions with only numerals in last four
positions. Prefix code is in first two positions, and any alphabetic
characters must be is three remaining positions.
4 First two positions = 04.
5 No Canadian registry associations for Brown Swiss or Red Poll; no
importations from other countries allowed at present for these breeds.

TABLE 6. Type E errors: questionable or invalid breed designation.

Error code Questionable or invalid data
E2 Registered cow, sire or dam designated
breed X (crossbred) or U (unknown).1

1 Not applicable if cow coded breed A (Ayrshire), G (Guernsey)
or B (Brown Swiss) with a prefix code of 36 or 37.

TABLE 7. Type F errors: questionable or invalid calving date.

Error
code Questionable or invalid data
F1 Calving date later than date that record was received by USDA.
F2 Month of calving coded as 0 or greater than 12.
F3 Day of calving coded as 0 (registered cows) or day of calving
coded as greater than 31 (any cow).
F4 Age at calving calculated from reported calving and birth dates
less than or equal to 14 months. 1
F5 Two records of cow reported with calving dates less than 9 months
apart.
F6 Two records of cow reported with identical calving dates but with
different amounts of milk or fat.
F7 Two lactations of a cow reported with equal yield but different
calving dates. 2
F8 Calving date plus days in milk more recent than date record was
received by USDA.
F9 More than 20 lactation records reported for cow.

1 If record valid, verification code 2 should be entered in position 31 of
format 4.
2 If records are correct, enter verification code 2 in position 87
of format 4.

TABLE 8. Type G errors: questionable or invalid prefix code 1 in an
eartag identification number.

Error code Questionable or invalid data
G1 Illegal prefix code in cow number.
G2 Illegal prefix code in sire number.
G3 Illegal prefix code in dam number.

1 See Fact Sheet H-1 1.

Type I errors (Table 9) occur if nonnumeric characters are
entered in numeric fields of a record. These errors usually
originate at DRPC's.

If the same identification number appears for two or more
animals in the pedigree, a type J error (Table 10) is assigned.

A type L error (Table 11) is assigned to a record if the number
of days milked three times a day exceeds the total number of
reported days in milk.

Records with questionable milk yield are rejected as type M
errors (Table 12). Actual milk yield is checked against upper and
lower limits for each breed.

An N type error (Table 13) is assigned to a record if the number
of reported days in milk was less than 7 or more than 305.

Records with questionable fat yield (type O error, Table 14)
also are checked against upper and lower limits for each breed.

TABLE 9. Type I errors: nonnumeric characters in numeric fields
of the record.

Error code Description
I1 Blank or alphabetic characters in one or more registration
numbers in the pedigree.

TABLE 10. Type J errors: same identification number for two or more animals
in the pedigree.

Error code Description
J1 Identical identification numbers for cow and sire, cow and dam,
or sire and dam.

TABLE 11. Type L errors: number of days milked three times a day (3X)
greater than total number of reported days in milk.

Error code Description
L2 Days milked 3X greater than total days in milk.

Component percentages for editing purposes (type P error, Table
15) are computed from actual reported milk and component yields.
Records with a production verification code or incomplete records
are checked against separate limits for fat percentage. Protein and
solids-not-fat (SNF) limits are not affected by verification.
Although protein and SNF information outside these limits is
deleted, the record may be used for milk and fat evaluations.
Protein and SNF records outside limits are included in the 1060
errors sent to DRPC's.

Type S errors (Table 16) are assigned to records with
questionable sire identification numbers. Sire registration numbers
that are outside breed limits (error code S1) usually involve an
invalid number of some type. For example, digits may be missing, or
an eartag number may have been reported improperly.

Correction of Errors in Lactation Records

A lactation record that has been eliminated from genetic
evaluations can be corrected by the following procedure: 1)
resubmitting the lactation record with the correct information or
2) including a verification code to indicate that the information
is valid although outside the edit limits.

Also, a lactation record that is in the file may contain
incorrect information and need correction. If the calving date is
correct, the lactation record should be resubmitted with a
production correction code in the proper column. If the calving
date is incorrect, a deletion record with the incorrect calving
date should be submitted with exactly the same data as the original
lactation record (that is, the same cow, sire, and dam
identification numbers, the same breed codes for each animal, the
same birth date, and the same calving date). Then, the lactation
record should be resubmitted with the correct calving date. This
correction procedure is necessary only if the previous lactation
record was not a record in progress. If a cow's identification data
remain the same, only one of her lactation records can be deleted
and resubmitted in any given run.

If more than one lactation record for a cow must be deleted and
resubmitted, then all her lactation records must be deleted. A
deletion record should be submitted with the same identification
data as in the original lactation records (that is, the same cow,
sire, and dam identification numbers, the same breed codes for each
animal, and the same birth date) and a calving date of all 9's. If
the cow's identification data have remained the same, her correct
lactation records should be resubmitted for the next run; if her
identification data have changed, the corrected lactation records
can be resubmitted for the same run. If any records in the same run
have identification data the same as in the deletion record, then
those records also will be deleted and not used for genetic
evaluations.

Pedigree Errors

Pedigree information is sent from the breed associations and the
Verified Identification Program to AIPL to be included in its
pedigree file. If only one identification field (sire, dam or birth
date) of the lactation record conflicts with information sent by
the breed association, the lactation record is changed to conform
with the pedigree file, and corrections are reported to DRPC's in
format 20. If more than one field is in conflict, the record is
deleted and included in the 1060 errors sent to DRPC's.

TABLE 12. Type M errors: questionable milk yield.

Error
code Questionable data
M2 Actual milk yield (pounds per day milked) outside
limits for breed
Cow's age
<36 months >36 months
Lower Upper Lower Upper
Breed limit limit limit limit
Ayrshire 5 200 10 200
Brown Swiss 5 200 10 200
Guernsey 5 200 8 200
Holstein 10 200 15 200
Jersey 5 200 8 200
Milking Shorthorn 5 200 8 200
Mixed 5 200 8 200
Red and White 5 200 8 200
Red Dane 5 200 10 200
Red Poll 5 200 8 200
M3 Milk yield not reported.

TABLE 13. Type N errors: number of reported days milked was less
than 7 or more than 305.

Error code Description
N1 Days in milk less than 7 or more than 305.

TABLE 14. The O errors: questionable fat yield.

Error code Questionable data
O2 Actual fat yield (pounds per day milked) outside limits for breed
Cow's age
<36 months >36 months
Lower Upper Lower Upper
Breed limit limit limit limit
Ayrshire .2 10 .3 10
Brown Swiss .2 10 .3 10
Guernsey .2 10 .3 10
Holstein .3 10 .4 10
Jersey .2 10 .3 10
Milking Shorthorn .2 10 .3 10
Mixed .2 10 .3 10
Red and White .2 10 .3 10
Red Dane .2 10 .3 10
Red Poll .2 10 .3 10
O3 Fat yield not reported.

TABLE 15. Type P errors: questionable component percentage.

Error
code Questionable data
P1 Fat percentage outside limits for breed:(1,2)
Lower Upper
Breed limit limit
Ayrshire 2.5 6
Brown Swiss 2.5 6
Guernsey 2.5 7
Holstein 2.0 6
Jersey 3.0 8
Milking Shorthorn 2.3 6
Mixed 2.0 8
Red and White 2.0 7
Red Dane 2.5 6
Red Poll 2.5 6
P4 Protein percentage outside limits for breed:(3,4)
Lower Upper
Breed limit limit
Ayrshire 2.5 4.3
Brown Swiss 2.6 4.6
Guernsey 2.5 4.7
Holstein 2.3 4.2
Jersey 2.8 5.1
Milking Shorthorn 2.4 4.4
Mixed 2.3 5.1
Red and White 2.3 4.7
Red Dane 2.4 4.6
Red Poll 2.6 4.6
P5 Solids-not-fat percentage outside limits for breed:(3,5)
Lower Upper
Breed limit limit
Ayrshire 7.5 10.5
Brown Swiss 8.0 11.0
Guernsey 8.0 11.0
Holstein 7.5 10.0
Jersey 8.0 11.5
Milking Shorthorn 7.5 10.5
Mixed 7.5 11.5
Red and White 7.5 11.0
Red Dane 8.0 11.0
Red Poll 8.0 11.0

1 Records with production verification code 2 in position 87 of format
4 have lower limit of 1.5 percent and upper limit of 9.9 percent
regardless of breed.
2 Incomplete records have lower limit of 2.0 percent and upper limit of
8.0 percent regardless of breed.
3 Records with less than 15 days in milk not included in
genetic evaluations.
4 Records outside limits not used in genetic evaluations but remain in
USDA file if within general limits of 1 to 6 percent.
5 Records outside limits not used in genetic evaluations but remain in
USDA file if within general limits of 6.1 to 11 percent.

TABLE 16. Type S errors: questionable sire identification.

Error code Questionable data
S1 Registration number of sire out of range for breed.
S2 Registered cow with an eartag sire number. 1
S3 Invalid eartag number for sire of grade cow.
S4 Registered Brown Swiss or Red Poll sire codes with a foreign
designation. 2,3
S6 Numeric sire identification number with 99 in first two
positions.

1 Eartag number must include nine positions with only numerals in last
four positions. Prefix code is in first two positions, and any
alphabetic characters must be in three remaining positions.
2 First two positions = 04.
3 No Canadian registry associations for Brown Swiss or Red Poll; no
importations from other countries allowed at present for these breeds.

Summary

Editing procedures for lactation records were established to
ensure the utmost reliability for data that contribute to genetic
evaluations. Many limits are relaxed if the data can be verified
and are coded as verified. Interest in correcting records has
increased because of the wide distribution of Cow Indexes (CI). A
dairy producer knows that a cow without a CI has had her records
eliminated from computations of genetic evaluations. DRPC's are
provided with reports on eliminated records so that questions from
dairy producers can be answered and records can be corrected.

��������������������The National Dairy Database (1992)��������������������
�����������������������\NDB\DAIRY\TEXT\DA102200.TXT�����������������������

%f TITLE;EDITING LACTATION RECORDS FOR USDA-DHIA GENETIC EVALUATIONS
%f COLLECTION;DAIRY HERD IMPROVEMENT
%f ORIGIN;Maryland
%f DATE_INCLUDED;June 1992