NO. 07-96-0140-CR
IN THE COURT OF APPEALS
FOR THE SEVENTH DISTRICT OF TEXAS
AT AMARILLO
PANEL B
JUNE 9, 1998
_________________________________
RUSSELL ALAN GRIFFITH, APPELLANT
V.
THE STATE OF TEXAS, APPELLEE
__________________________________
FROM THE 99TH JUDICIAL DISTRICT COURT OF LUBBOCK COUNTY;
NO. 91-414023; HONORABLE MACKEY K. HANCOCK, JUDGE
_________________________________
Before BOYD, C.J., DODSON & QUINN, JJ.
In a jury trial, appellant Russell Alan Griffith was convicted
of sexual assault. The jury assessed his punishment at confinement
for twenty years in the Texas Department of Criminal Justice,
Institutional Division. By three points of error, appellant
contends the trial court erred in admitting State’s evidence
regarding DNA testing involving a probability of paternity
statistic using Bayes’ Theorem as violating the presumption of
innocence, or in the alternative, the trial court erred in
admitting such evidence without testimony on the mathematical
applications of the test results, and that the court erred in
overruling his motion to set aside the verdict and judgment where
the prosecution introduced inadmissible evidence clearly calculated
to inflame the minds of the jury. Affirmed.
On July 17, 1989, the staff of the Lubbock State School (the
School) had a female patient, T.S., examined because of abdominal
swelling. T.S. was a profoundly retarded female client in her
early thirties. With an I.Q. of 11, T.S. had the mental capacity
of a two year old child, and had very diminished communication
skills. She was therefore unable to tell anyone that she had been
assaulted. An x-ray revealed that T.S. was pregnant. Further
diagnosis placed the date of conception between February 7, 1989
and March 27, 1989. A child was born on December 7, 1989.
After approximately one year, School officials notified the
police when they began to suspect that an employee may have been
the father. Prior to that time, the School believed that the
father was probably one of the male clients at the School.
Appellant started work at the School as a direct care worker
in August, 1988. Appellant worked in the restricted access
dormitories on the night shift from 10 p.m. to 6 a.m. After
reviewing sign-in logs, police determined that five male direct-
care workers, including appellant, had access to T.S. between the
dates of the estimated conception. Blood samples from T.S., the
-2-
baby, and the five male suspects were sent to the University of
North Texas Health Science Center in Fort Worth for DNA testing.
Dr. Arthur J. Eisenberg, the administrator of the lab,
testified as a State’s witness at trial. The DNA test results
excluded four of the five male direct-care workers from being the
father of the child. Appellant was not excluded. Dr. Eisenberg
testified that three different statistical values were generated
from appellant’s DNA test results. One of those statistics, the
probability of paternity, was challenged by the defense. A hearing
on a motion to suppress this evidence was had, and the trial court
overruled the motion. The evidence was then admitted before the
jury which convicted him of sexual assault and sentenced him to
twenty years in the Texas Department of Criminal Justice,
Institutional Division. Appellant timely filed a motion for new
trial, which was denied. This appeal followed.
Appellant’s First Point of Error
In his first point of error, appellant complains that the
trial court erred in admitting testimony regarding DNA testing,
specifically the probability of paternity statistic based on Bayes’
Theorem, because the calculation was based on a presumption of
guilt. Under this point of error, appellant contends that the use
of Bayes’ Theorem to calculate the probability of paternity
statistic permitted the State to convict him without meeting its
-3-
burden of proof. Specifically, he says that the use of the Bayes’
Theorem to calculate the probability of paternity statistic assumes
a fact for which there is no independent proof — i.e., that he had
sex with the complainant. Appellant limits his challenge to the
DNA evidence admitted at trial to the probability of paternity
statistic calculated by the use of Bayes’ Theorem. The remaining
DNA evidence in the record is unchallenged.
The record shows that there are two possible results from a
DNA paternity test. Either a potential father is excluded, meaning
he is shown to not be the father, or he is included. If the male
is excluded from paternity by the test, no statistics are
generated. If the male is included, the results are not absolutely
conclusive that he is the father and there remains a chance or
possibility that he is not the father, even though that possibility
in some instances may be very de minimis. This possibility is
stated statistically. Nevertheless, only the biological father’s
test results will match the child’s test results. When the male is
included, as appellant in this instance, the test results are
reduced to statistical figures derived from all frequencies
assigned to each chromosome region tested (i.e., six in this
instance). The statistical values are reported in three ways: the
paternity index, the probability of exclusion, and the probability
of paternity.
-4-
The paternity index is a value reflecting the likelihood that
a tested man is the father of the child as opposed to an untested
man of the same race. It is expressed in a number. If a paternity
index can be assigned to a man, it means that he is that many more
times likely to be the father than any other randomly selected male
of his race. Paternity index is determined by multiplying together
all of the allele frequencies (rate of occurrence) for each region
tested.
The probability of exclusion considers the DNA of the mother
and the child. This number is a percentage. Since half of a
child’s DNA comes from each parent, by comparing the DNA of the
mother and the child, then excluding the DNA that matches, the
remaining DNA of the child necessarily belongs to the father. This
number reflects the strength of the DNA test, by showing the
percentage of the male population that would have been excluded by
the test.
Finally, DNA test results can be expressed as a probability of
paternity. This number is also a percentage. This statistic is
calculated using Bayes’ Theorem, a mathematical formula in which
probabilities are associated with individual events and not merely
with random sequences of events. Webster’s New Collegiate
Dictionary 95 (1981). Bayes’ Theorem is necessary to convert
probabilities into percentages. The formula is stated as follows:
-5-
or
See M. v. Marvin S., 656 N.Y.S.2d 802, 806 n.4 (Fam.Ct. 1997);
State v. Skipper, 637 A.2d 1101, 1104 (Conn. 1994). The resulting
percentage reflects the percent likelihood that the tested male is
actually the father of the child. The formula requires the use of
a prior probability of an event occurring.
The Test and the Results
After the police collected blood samples from the mother, the
child, and the five male suspects, the samples were sent to the
University of North Texas Health Science Center at Fort Worth where
DNA tests were run. Dr. Eisenberg testified about the results of
the tests and the resulting statistical analysis. Initially, four
DNA regions, or loci, were tested. Three men did not match at any
tested region. One man matched at only one region. Accordingly,
these four men were excluded from paternity. The fifth man,
appellant, matched at all four tested regions. Dr. Eisenberg
testified that two additional genetic regions were tested.
-6-
Appellant matched in both, bringing the total to six matches.
Since the other four men were excluded, no statistics were
generated on them.
Statistics as to appellant’s results were generated.
Appellant’s paternity index was 14,961 (indicating he was 14,961
times more likely to be the father than a randomly selected male of
his race). The probability of exclusion was “in excess” of 99.99%
(the test would have excluded more than 9,999 men of every 10,000
tested). The probability of paternity was “in excess” of 99.99%
(the likelihood that appellant was the father of the child was
higher than 99.99%). It is this third statistical figure that
appellant challenges.
Admissibility of the Challenged Evidence
We are persuaded the admissibility of the challenged evidence
is controlled by the Court of Criminal Appeals’ determination in
Kelly v. State, 824 S.W.2d 568 (Tex.Cr.App. 1992). In Kelly, the
court delineated the standard for the admissibility of novel
scientific evidence. Before admitting such evidence, a trial court
must make the “threshold determination” as to whether the testimony
will help the fact trier understand the evidence or determine a
fact in issue. Thus, when the trial court is faced with a proffer
of expert testimony or a scientific topic unfamiliar to lay jurors,
the trial court’s first task is to determine whether the testimony
-7-
is sufficiently reliable and relevant to help the jury in reaching
accurate results. Id. at 572. If the trial court determines that
the proffered expert testimony is reliable (i.e., probative and
relevant), the trial court must next determine whether the
proffered testimony might nevertheless be unhelpful to the fact
triers for other reasons, such as if it is merely cumulative or
would confuse or mislead the jury, or would consume an inordinate
amount of trial time. In essence, if the trial court determines
that the proffered expert testimony is reliable and relevant, the
court must still determine whether the probative value of the
evidence is outweighed by one or more of the factors in Rule 403 of
the Texas Rules of Evidence.1 Id.
The Court of Criminal Appeals further explained how the
reliable prong of the test of admissibility should be met. For
scientific evidence to be considered reliable, it must satisfy
three criteria. First, the underlying scientific theory must be
valid. Next, the technique applying the theory must be valid.
Finally, the technique must have been properly applied on the
occasion in question. Id. at 573.
1
“Although relevant, evidence may be excluded if its probative
value is substantially outweighed by the danger of unfair
prejudice, confusion of the issues, or misleading the jury, or by
considerations of undue delay, or needless presentation of
cumulative evidence.” Tex. R. Evid. 403.
-8-
These three criteria must be shown by clear and convincing
evidence outside the presence of the jury. Id. Factors that could
affect the trial court’s determination include, but are not limited
to the following: the extent to which the underlying scientific
theory and technique are accepted as valid by the relevant
scientific community, the qualifications of the expert testifying,
the existence of literature supporting or rejecting the underlying
scientific theory and technique, the potential rate of error in the
technique, the availability of other experts to test and evaluate
the technique, the clarity with which the underlying scientific
theory and technique can be explained to the court, and the
experience and skill of the persons who applied the technique on
the occasion in question. Id.
The Kelly court summarized its determination as follows:
To summarize, under Rule 702 the proponent of novel
scientific evidence must prove to the trial court, by
clear and convincing evidence and outside the presence of
the jury, that the proffered evidence is relevant. If
the trial court is so persuaded, then the evidence should
be admitted for the jury’s consideration, unless the
trial court determines that the probative value of the
evidence is outweighed by some factor identified in Rule
403. (Emphasis added.)
When the admission of such evidence is challenged on appeal, the
question is whether the trial court abused its discretion by
admitting the evidence.
-9-
In the case before us, appellant does not challenge the
admissibility of the DNA testing, nor does he attack two of the
three statistics generated from the test results. We note that in
Kelly, the Court of Criminal Appeals addressed for the first time
whether RFLP (restriction fragment length polymorphism) DNA testing
was admissible in a criminal trial. Applying the newly announced
rule, the Court concluded such testing was admissible. Id. at
574.
In this instance, appellant challenges the probability of
paternity statistic calculated from the DNA test results. We
conclude that the probability of paternity statistic meets the
Kelly admissibility requirements and that the trial court did not
abuse its discretion in admitting the challenged evidence.
The trial court conducted a hearing outside the presence of
the jury to determine the admissibility of the State’s DNA
evidence. The State’s expert, Dr. Arthur J. Eisenberg, testified
about the DNA evidence generally and the probability of paternity
statistic in particular. Dr. Eisenberg has a Bachelor’s of Science
in Biology, a Master’s of Science in molecular biology, and a Ph.D.
in molecular biology. He listed a number of organizations he
belongs to involved in DNA testing or research including the
American Association of Blood Banks, the U.S. DNA Advisory Board,
and the Parentage Testing Committee. Further, he testified that he
had been involved in the field of DNA testing since its inception.
-10-
Eisenberg set up and manages the DNA laboratory at the University
of North Texas at Fort Worth where the testing in this case was
performed.
Eisenberg testified that in this case he conducted a paternity
test on five males, T.S., and the baby. Appellant was one of those
five, and he was the only one not excluded from paternity by the
DNA testing. Eisenberg stated unequivocally that the methodologies
used for statistical analysis of the test results were “standard
methods” employed in over 200,000 parentage tests performed
nationwide annually.
Eisenberg explained each of the three statistics in turn. The
probability of paternity was calculated by using Bayes’ Theorem.
Bayes’ Theorem, according to Eisenberg, states that prior to the
testing, there is a prior probability of paternity. He stated that
courts in the United States typically use a .5 or 50% prior
probability because it is a neutral probability. The .5 prior
probability indicates that the tested male either is or is not the
father. Eisenberg further testified that this calculation was a
generally accepted principle, and was standard methodology in
parentage testing, having been used for twenty or thirty years.2
2
Although parentage testing based on DNA analysis has only
been on the scene since the mid to late 1980's, a number of
methods, including Human Leukocyte Antigen (HLA) tests, have been
previously employed in paternity matters. HLA testing invokes the
same statistical calculations, including the probability of
paternity and Bayes’ Theorem. Dr. Eisenberg testified that in the
-11-
Eisenberg further explained the theory and methodology
involved in DNA testing generally. After explaining how DNA
functions and how the tests are conducted, he discussed the
specific results in this case. Eisenberg stated that using the .5
prior probability, which was the standard prior probability
reported in parentage tests, that appellant’s probability of
paternity was 99.99%. At this point, the State passed Eisenberg as
a witness, and defense counsel cross-examined him.
On cross, Eisenberg reiterated that the prior probability of
.5 was a neutral prior probability which did not presume appellant
was guilty of the crime or more likely than not guilty. He
emphasized that he had personally testified in both civil and
criminal paternity matters using the same statistic invoking a .5
prior probability. Eisenberg stated that he had testified in over
a dozen Texas criminal cases involving paternity issues where he
used the .5 prior probability.
Most notably, Dr. Eisenberg was asked point blank whether he
saw any problem using the .5 prior probability in a criminal case,
even assuming the defendant as presumed to be innocent.
Eisenberg’s answer, twice, was “[a]bsolutely not.” He testified
that the .5 prior probability did not unfairly skew the probability
past several years, nearly a million paternity tests in the U.S.
were conducted using DNA or HLA methods, each using the .5 prior
probability calculation.
-12-
of paternity statistic. Moreover, Eisenberg stated that if a lower
prior probability number had been used, like .1, then the
probability of paternity statistic would have been lower, though it
would still be representative of the fact that the appellant had
matched at six genetic test sites.3 According to Eisenberg, if a
prior probability that reflected true parentage testing had been
used, it would have been something higher than .5 and the
probability of paternity would have been even higher than 99.99%.
Based on Dr. Eisenberg’s testimony, the trial court was
required to determine whether the State had shown by clear and
convincing evidence that the probability of paternity statistic
would be helpful to the trier of fact and that it was sufficiently
reliable and relevant to help the jury in reaching accurate
results. Looking to the factors outlined in Kelly, we note that
Eisenberg testified that hundreds of thousands of DNA tests, and
millions of HLA and DNA tests around the nation reported paternity
results using Bayes’ Theorem and the probability of paternity
invoking a .5 prior probability. These tests were conducted by
accredited testing facilities, and the statistical calculation was
“standard.”
3
Ultimately, there was testimony before the jury that the use
of a .01 (1%) prior probability would still generate a probability
of paternity of over 99.3% in this case.
-13-
Eisenberg testified about his qualifications in DNA paternity
testing and reported that he was involved in the field from its
inception. He testified that the statistical calculation was
employed for twenty to thirty years in paternity tests based on HLA
blood typing and later DNA analysis. Eisenberg commented that
there were over fifty other laboratories in the country using the
same techniques and reporting the same statistics. He also stated
that the calculations employed in this particular test were the
“standard” method of reporting paternity results around the
country. Finally, he testified about the techniques involved in
DNA testing, his qualifications in conducting those tests, and his
experience in reporting statistics, which were “co-related” to the
DNA testing.
Based on Eisenberg’s testimony, the trial court clearly
recognized that the Bayes’ Theorem calculation was commonly used in
reporting DNA paternity results. Moreover, it is clear that the
probability of paternity statistic is accepted in the scientific
community of molecular biology in reporting paternity results.
Eisenberg stated that he used the same calculation used in
thousands of other tests, indicating that he properly invoked the
reporting method. Likewise, there was no challenge that he did the
math improperly. We conclude that this evidence was clear and
convincing in showing that the probability of paternity statistic
was valid, the technique applying the statistic was valid, and that
-14-
it was properly applied in this case. Thus, the trial court
properly concluded that the statistic was reliable and relevant to
helping the jury reach accurate results.
As to the second prong of the Kelly test, there was no
challenge to the evidence as being time consuming, cumulative,
confusing or misleading, or otherwise more prejudicial than
probative. The only challenge raised by appellant was his
assertion that the statistic violates the presumption of innocence.
The Presumption of Innocence
The presumption of innocence does not appear in the U.S. or
the Texas Constitutions. However, courts have recognized that the
presumption of innocence is part of the 14th Amendment Due Process
and 6th Amendment right to fair trial. Randle v. State, 826 S.W.2d
943, 945 n. 3 (Tex.Cr.App. 1992); Rogers v. State, 846 S.W.2d 883,
885 (Tex.App.--Beaumont 1993, no pet.). Also, the Legislature has
codified the presumption of innocence in the Texas Penal Code and
the Code of Criminal Procedure. See Tex. Penal Code Ann. § 2.01
(Vernon 1994); Tex. Code Crim. Proc. Ann. art. 38.03 (Vernon Supp.
1998).
It is stated that the presumption of innocence is not a true
presumption. Normally, a presumption is an assumption of fact that
the law requires to be made from another fact or group of facts
-15-
found or otherwise established in the action, which may be
rebuttable or conclusive. Black’s Law Dictionary, 1185 (6th ed.
1990). A presumption acts as a burden shifting device. Id.
By contrast, the presumption of innocence is perhaps better
phrased the “assumption of innocence.” McCormick on Evidence § 342
at 579-80 4th ed. (1992). It merely describes the fact that the
burden of persuasion and production in a criminal matter are on the
prosecution. Id. It cautions the jury to reach their conclusion
solely from the evidence adduced, and not from the fact of arrest
or indictment. Id. citing 9 Wigmore Evidence § 2511 at 407
(Chadbourn rev. 1981).
The presumption of innocence is not a true presumption because
the defendant is not required to come forward with proof of
innocence once evidence of guilt is introduced so as to avoid a
directed verdict of guilty. Black’s Law Dictionary, 1186 (6th ed.
1990). Typically, cases finding violations of the presumption of
innocence involve situations where the defendant is placed before
the jury, dressed in shackles or jail clothes, or where the State
offers evidence that the defendant has been indicted in other
crimes. See Randle, 826 S.W.2d at 946; Lafayette v. State, 835
S.W.2d 131, 135 (Tex.App.--Texarkana 1992, no pet.). Clearly
neither of those situations exist here.
-16-
In the case before us, testimony was elicited from Dr.
Eisenberg about all three statistics. Dr. Eisenberg testified on
direct about probability of paternity based on a .5 prior
probability. On cross, he testified about how the probability
number would change based on different prior probability values.
We conclude that the use of a probability of paternity statistic
based on Bayes’ Theorem in a criminal proceeding does not violate
the presumption of innocence. The use of a prior probability of
.5 is a neutral assumption. The statistic merely reflects the
application of a scientifically accepted mathematical theorem which
in turn is an expression of the expert’s opinion testimony. It is
subject to the same conditions applied to all other expert
testimony. The jury is free to disregard it. It can be weakened
on cross and in argument. The statistic does nothing to shift the
burden of persuasion or production in a criminal matter.
Appellant asserts that his specific challenge is a matter of
first impression in Texas criminal cases. Consequently, he relies
on two cases from other jurisdictions where the courts exclude the
probability of paternity calculation as a violation of the
presumption of innocence. While we do find cases that have
admitted DNA testing and the probability of paternity statistic, we
-17-
have found no Texas criminal case in which the presumption of
innocence challenge was made or addressed.4
The two primary cases the appellant relies on to support this
alleged violation of the presumption of innocence challenges are
State v. Hartman, 426 N.W.2d 320 (Wis. 1988) and State v. Skipper,
637 A.2d 1101 (Conn. 1994). The rationale in Hartman and Skipper
is that the probability of paternity statistic violates the
presumption of innocence because it assumes that the putative
father had sexual intercourse with the mother; stated another way,
it assumes the crime was committed by him in order to prove that
the crime was committed by him. Hartman, 426 N.W.2d at 326;
Skipper, 637 A.2d at 1106 (citing Hartman). Both of these cases
come to this conclusion, at least in part, by relying on Peterson,
A Few Things You Should Know About Paternity Tests (But Were Afraid
To Ask), 22 Santa Clara L.Rev. 667 (1982).
4
In Lagrone v. State, 942 S.W.2d 602, 608 (Tex.Cr.App. 1997),
the Court of Criminal Appeals mentioned Dr. Eisenberg’s opinion on
the probability of paternity statistic without passing on the issue
before us. We note that the probability of paternity statistic has
been admitted in a number of jurisdictions in criminal trials prior
to this case. However, in those cases, the statistic was not
challenged as violating the presumption of innocence. See State v.
Foster, 949 S.W.2d 215, 217 (Mo.App.E.D. 1997); State v. Pierre,
606 So.2d 816, 817-20 (La.App. 3 Cir. 1992); People v. Taylor, 460
N.W.2d 582, 585 (Mich.App. 1990); Martinez v. State, 549 So.2d 694,
696-97 (Fla.App. 5 Dist. 1989); Holley v. State, 523 So.2d 688, 689
(Fla.App. 1 Dist. 1988); State v. Smith, 735 S.W.2d 831, 833-35
(Tenn.Cr.App. 1987); State v. Thompson, 503 A.2d 689, 690-93 (Me.
1986); Bridgeman v. Commonwealth, 351 S.E.2d 598, 602-03 (Va.App.
1986); People v. Alzoubi, 479 N.E.2d 1208, 1209 (Ill.App. 3 Dist.
1985).
-18-
Additionally, the Hartman court bases its conclusion on a
single statement it made just one month earlier in In Re Paternity
of M.J.B., 425 N.W.2d 404 (Wis. 1988). In Hartman, the court said
the assumption underlying the probability of paternity statistic
was “that the mother and the putative father have engaged in sexual
intercourse at least once during the possible conception.”
Hartman, 426 N.W.2d at 326 (quoting M.J.B., 425 N.W.2d at 409, in
turn citing Peterson, 22 Santa Clara L. Rev. at 685). For reasons
we shall explain, we do not agree that the basic assumption that
intercourse occurred is implicit in the statistic.
Peterson’s Santa Clara Law Review article seems to be at the
root of the Hartman and Skipper decisions. That article discusses
the use of blood tests in paternity cases, including HLA testing.
HLA testing reports the same three statistics reported in DNA
testing, and in particular in the case before us. In that article,
Peterson criticizes the value of Bayes’ Theorem. He states that
Bayes’ Theorem accurately reflects the odds that the accused is the
father only if one assumes “that the defendant had intercourse with
the mother and that a random man . . . also had intercourse with
her.” Peterson, Santa Clara L.Rev. at 685. We note that the
author of the article was himself not a statistician or geneticist,
but an attorney and professor. We further note that the author
does not cite direct authority (either legal or scientific) to
support his statement. We disagree with this conclusion.
-19-
Logically, the prior probability assumes intercourse could have
occurred and thus the putative father could be the actual father,
but the statistic does not necessarily assume intercourse did
occur.
As Dr. Eisenberg testified at the suppression hearing, the .5
prior probability is “a neutral prior probability” that indicates
“[e]ither [the putative father] is or is not the father.” There
was no testimony from Eisenberg or Koehler, the defense expert,
indicating that the prior probability assumes intercourse
necessarily occurred. The prior assumption could invoke any number
of possible conditions or permutations, as Peterson points out,
including time of intercourse, frequency, fertility, and the like.
However, by making the prior assumption .5 (i.e., - equally
weighted), Bayes’ Theorem also allows that intercourse may not have
occurred at all.
Hartman and Skipper rely heavily on the conclusion in
Peterson’s article which we consider questionable. Moreover, it is
important to note that the Hartman court, while it quotes M.J.B. in
part, does not follow M.J.B.’s rationale. In M.J.B., the Wisconsin
Supreme Court also stated that “the probability of paternity
statistic is conditionally relevant evidence; only after competent
evidence is offered to show that sexual intercourse between the
mother and alleged father occurred during the conceptive period may
-20-
evidence of the probability of paternity statistic be received.”
In Re Paternity of M.J.B., 425 N.W.2d at 409. However, the
Wisconsin Supreme Court further stated:
This foundational evidence [of intercourse] may be
supplied by the mother herself . . . . However, we note
that this threshold evidence is not limited to direct
testimony by the mother that she engaged in sexual
intercourse with the alleged father. Evidence that the
defendant has access to the mother during the conceptive
period may be offered by any individual knowledgeable of
the facts of their association. By ‘access’ we mean that
the mother and putative father were together at a time,
under circumstances and in a location which would lead a
reasonable person to believe that the sexual intercourse
took place between them.
Id. (emphasis added).5
In the case before us, there was testimony from Lubbock police
that appellant was one of the male care workers who had access to
T.S.’s dormitory. Moreover, there was evidence that appellant
worked the late night shift, from 10:00 p.m. to 6:00 a.m. Both the
police, via the restricted access dormitory log sheets, and
appellant himself, provided evidence that appellant had the
opportunity to be alone in the dorm with T.S. and other patients
during the conceptive period; that is, he had opportunity to be
with the patients without another worker present. Finally, it is
5
We note that M.J.B. is a civil paternity case. The Wisconsin
Court allowed the probability of paternity statistic primarily due
to a state statute allowing such evidence in civil paternity cases.
Nevertheless, the Hartman decision seems to depart from the
rationale in M.J.B. while relying on some of that case’s language.
-21-
important to note that in this case before us, due to T.S.’s
impaired mental facility, there could not be any direct testimony
from her regarding who assaulted her.
Three justices (of seven) dissented in Hartman. Justice
Steinmertz commented in his dissent on the presumption of innocence
issue. “The 50 percent prior chance assumption does not require
shifting the burden of proof to the defendant and is not an
impermissible assumption; rather, it is part of a scientific theory
and the jury should be so told.” Id. at 327. He noted that the
assumption was not made in a vacuum, but was admitted only after
evidence serving as the basis for the statistic was already
admitted. Id. The probability of paternity statistic, Justice
Steinmertz reasoned, is truly neutral. It equally assumes the
defendant is not the putative father, no matter how damning the
evidence in the case. Id. at 328.
We agree with Justice Steinmertz’s evaluation of the
statistic. In the case before us, there was evidence that
appellant had access and opportunity to have intercourse with T.S.
The DNA test itself indicated appellant was the father of the
child. Dr. Eisenberg testified in no uncertain terms that the
theory was used as the standard method of reporting paternity
tests. On cross, he testified about the effect of lower prior
probabilities on the probability of paternity. As with any other
expert testimony, the jury was free to disregard it entirely.
-22-
Nothing about the statistic shifts the burden of persuasion to the
defendant.
In contrast to Hartman, Skipper represents the strongest
denunciation by a court of the probability of paternity statistic
as violating the presumption of innocence. 637 A.2d 1101 (Conn.
1994). There, the defendant was convicted of second degree sexual
assault. The Connecticut Supreme Court stated “[t]he assumption
that sexual intercourse had occurred was not predicated on the
evidence in the case, but was simply an assumption made by the
expert.” Id. at 1106. Since Bayes’ Theorem cannot be invoked
without assuming a prior probability of paternity, the court
reasoned that its use was inconsistent with the presumption of
innocence. Id. at 1107. The Connecticut Court further reasoned
that if a value presuming innocence was entered into the equation,
the value being zero, then Bayes’ Theorem would produce a 0%
probability of paternity. Id. at 1108. Beyond that fact that this
decision rests on Peterson’s questionable conclusion, we simply do
not agree with the Connecticut Court’s rationale.
In this instance, five individuals were determined to have
access to T.S. during the period the child was conceived.
Initially, there was no presumption assigned to any of these men’s
paternity. Only after the men with access were tested, and all but
one excluded, was a prior probability employed. At that point,
appellant was the only actual man included, and the statistic
-23-
presumes either he or a random man could have been the father.
Thus, the .5 prior probability accurately represents that he either
is or is not the father.
Moreover, the presumption of innocence cannot require us to
enter a prior probability of zero into Bayes’ Theorem as suggested
by the Connecticut Court. A zero prior probability does not simply
presume a defendant is innocent. Rather, a zero probability, in
fact presumes that it was impossible for the defendant to be the
father.6 When a zero prior probability is plugged into Bayes’
Theorem (the formula), naturally the probability of paternity
results becomes 0%. The presumption of innocence does not require
a jury to assume it was impossible for a defendant to commit the
crime charged. Rather, it requires the jury to assume as a
starting proposition that the defendant did not commit the crime,
until proven otherwise. The probability of paternity, as Dr.
Eisenberg testified, is merely a way of expressing and interpreting
the actual DNA test results. Thus, the statistic itself does
nothing to shift the burden of going ahead to the defendant.
Finally, appellant cites a third case, State v. Spann, 617
A.2d 247 (N.J. 1993). There the New Jersey Supreme Court held that
where the clear impression was given to the jury that the 50% prior
probability was a scientific assumption, the admission of the
6
Likewise, a prior probability of 1 (or 100%) would assume
that no one else but the accused could have been the father.
-24-
probability of paternity statistic was reversible error.7 Id. at
253. In Spann, there was no explanation to the jury about how the
evidence in the case might affect the prior probability, and how
that would in turn affect the probability of paternity statistic.
The court reasoned that a jury should use its own estimate of the
prior probability of paternity, and not rely on the expert’s
assumption of the defendant’s access to the woman. Id. at 254.
We note that the New Jersey Court did not conclude that the
probability of paternity statistic violated the presumption of
innocence. In fact, the court discussed a number of issues to help
guide attorneys and courts in deciding whether the statistic would
be admissible in any given case. Id. at 257-60. The court
referred to concepts of general acceptance, reliability, and
usefulness for the jury. Id. at 258. Ultimately, for future
cases, the New Jersey Court left the determination of admissibility
of the probability of paternity statistic to the trial court,
implying that they found no interference with the presumption of
innocence. Moreover, the Spann Court expressly rejected the
suggestion that the Wisconsin Supreme Court arrived at in M.J.B.,
i.e., that intercourse must be proven before the probability of
paternity statistic can be admitted. Id. at 261. The New Jersey
Supreme Court stated that “[t]he calculation - Bayes’ Theorem - if
7
This case involved Human Leukocyte Antigen (HLA) testing
rather than DNA testing, but Bayes’ Theorem is used to calculate
probability of paternity in both tests.
-25-
valid, does not depend on any particular degree of confidence in
the fact of intercourse.” Id.
The presumption of innocence places the burden on the State to
move forward and prove that the defendant committed all the
elements of the crime beyond a reasonable doubt. In a sexual
assault case, one element the State must show is that the defendant
caused “the penetration of the . . . female sexual organ . . .” of
the victim. Tex. Penal Code Ann. § 22.011(a)(1)(A)(Vernon Supp.
1998). While it is true that the probability of paternity
statistic presumes that the defendant could have had intercourse
with the mother of the child, it does not assume that he did have
intercourse. As Dr. Eisenberg testified, a prior probability of .5
assumes that the defendant is just as not likely the father of the
child as it assumes he is the father. Moreover, even if the prior
probability was .9, strongly presuming that he was the father, it
still does not conclusively establish, or presume or assume he had
intercourse with the woman. This is a matter for the jury based on
all the evidence in the case, which could include no access,
impotence, vasectomy and other similar matters.
The Indiana Court of Appeals, over an objection that Bayes’
Theorem violated the presumption of innocence, expressly concluded
that the probability of paternity statistic was admissible in a
criminal trial. In Davis v. State, a husband and wife were
-26-
convicted of neglect of a dependant. Their baby was abandoned on
the side of a gravel road within hours of its birth. Using HLA
testing and Bayes’ Theorem, the State showed that the Davis’s were
the parents of the abandoned child. On appeal, the parents
contended that Bayes’ Theorem violated the presumption of
innocence.
In Davis, one element the State had to prove was that the
abandoned child belonged to the defendants. Using parentage tests,
the State was able to link the defendants to the child in order to
prove that they had committed the crime charged. In the case
before us, the State has also used parentage tests to link the
defendant with the crime charged. The issue in Davis was whether
Bayes’ Theorem could be used in a criminal case to show parentage.
The Indiana appellate court determined that the .5 probability
invoked in Bayes’ Theorem was a neutral consideration and that the
probability of parentage statistic was admissible. Id. at 138.
In this instance, we conclude that probability of parentage
statistic is admissible under Kelly v. State, supra, and that its
admissibility under Kelly does not violate the appellant’s
presumption of innocence. Appellant’s first point of error is
overruled.
Appellant’s Second Point of Error
-27-
In the alternative to his first point of error, appellant
claims in his second point that the trial court erred by admitting
the probability of paternity statistic because there was no
testimony regarding the mathematical applications of the test
results of the probability of paternity testing using Bayes’
Theorem. Under the point, appellant, in essence claims that as a
condition of admissibility, the State is required to call a
mathematical expert to comment on the possible interpretations of
the statistical evidence. We disagree. Rule 702 and Kelly make no
such requirement for the admission of the scientific evidence in
question.
To support his position, appellant points out that where
Bayes’ Theorem has been permitted, some courts require certain
precautionary conditions be met before allowing the evidence.
Particularly, he points to Spann v. New Jersey, 617 A.2d at 264.
While the New Jersey Supreme Court indicated that it might be
necessary to have expert testimony from a geneticist and a
mathematician in order to allow Bayes’ Theorem evidence at trial,
we note that the court was reviewing admissibility of evidence
under its own state standard. As we have previously discussed
above, in Texas the admissibility of scientific evidence is
governed by Rule 702 of the Texas Rules of Evidence and the
-28-
standard laid out in Kelly.8 Again, we are convinced that the
statistical evidence presented in this case satisfied that test.
The record contains testimony from Dr. Eisenberg addressing
the relevance and reliability of the probability of paternity
statistic. In the hearing on the motion to suppress, he testified
about his extensive credentials and expertise in the field of
molecular biology as applied to genetic testing. He testified that
the methodologies employed in the DNA testing were standard,
including the statistical calculations that were used to interpret
the test results. Specifically, he testified that use of the .5
prior probability was standard in parentage testing, and that it
was a neutral factor since it did not “give any weight to either
side” on the issue of paternity. Dr. Eisenberg testified before
the jury that if the prior probability in the calculation were
reduced to .01 (1%), reflecting a very low assumption that
appellant was the father, the probability of paternity was still
“in excess of 99 percent.” Finally, he testified that the tests
run in this case were run twice in order to verify the results and
rule out the possibility of errors. In light of this testimony,
8
At the time of trial, the Texas Rules of Criminal Evidence
and Texas Rules of Civil Evidence were still separate. As of March
1, 1998, these rules have been consolidated. While the new rules
technically do not apply to this matter, we note that the current
Rule 702 is identical to the old Rule 702 under the Criminal Rules.
-29-
the trial court was within its discretion to admit the probability
of paternity statistic under the Kelly test.9
Even assuming arguendo that the probability of paternity
statistic was improperly admitted, we conclude that such error was
harmless. The defense had the opportunity to cross examine Dr.
Eisenberg on the use of the prior probability. By cross, the
defense pointed out to the jury the nature of the probability of
paternity statistic and how it could be misleading. The defense
did not question the other two statistics at all. Based on other
evidence that appellant had access to T.S., that he had opportunity
to be alone with her, that he knew she could not consent to sexual
intercourse, that appellant matched on all six regions of DNA loci
tested, that the test included him while excluding 99.99% of the
male population of his race, and that his paternity index made him
nearly 15,000 times more likely than the random man to be the
father of T.S.’s child, we conclude beyond a reasonable doubt that
the admission of the probability of paternity, even if error, made
9
Although we conclude that the statistical evidence was
properly admitted, it is worth noting that the statistics merely
reinforce the truly damning evidence in this case - the DNA test
itself. Eisenberg testified that only the biological father or his
identical twin would match the child’s DNA at every site tested.
Appellant himself testified that he did not have an identical twin.
Eisenberg stated that based on the DNA test, it was his opinion
that appellant was the father of the child, barring a first order
relative (i.e. brothers or father) or an identical twin. As
between first order relatives, appellant was 64 times more likely
to be the father. Here, the test results speak for themselves.
Appellant matched at all six genetic sites tested.
-30-
no contribution to the conviction.10 Appellant’s second point of
error is overruled.
Appellant’s Third Point of Error
In his third point of error, appellant claims the trial court
erred by overruling his motion to set aside the verdict and
judgment rendered against him and grant him a new trial because the
prosecution knowingly introduced inadmissible evidence clearly
calculated to inflame the minds of the jurors against him. We
disagree.
The complained of statements came from Janice Robinson,
another state school employee. The State’s attorney asked Robinson
if she was aware of a statement made by the appellant that the
female clients of the State School were “easy” or that they “wanted
sex.” Robinson answered the question affirmatively before defense
counsel objected. Upon objection, the court held a hearing outside
the presence of the jury. The court denied the defense’s motion
for mistrial based on prosecutorial misconduct, then sustained the
objection. The jury was brought back in, and the court ordered the
jury to disregard the question and the answer. In essence,
10
Appellant waived his right to remain silent, and took the
stand voluntarily. On cross, he conceded that there was a
“possibility” that he had time alone with T.S., he knew T.S. was
“very retarded” and she “probably” couldn’t understand the nature
of sexual contact or activity, and that he could not explain why
the DNA test results came out as they did.
-31-
appellant contends that in offering the statement, the prosecution
committed prosecutorial misconduct which constitutes reversible
error. Again, we reiterate our disagreement.
The decision to grant or deny a motion for new trial is within
the discretion of the trial court, and appellate courts will not
reverse such decisions absent an abuse of discretion. State v.
Gonzalez, 855 S.W.2d 692, 696 (Tex.Crim.App. 1993). Moreover,
error in asking an improper question or admitting improper
testimony may generally be cured by an instruction to disregard.
Livingston v. State, 739 S.W.2d 311, 335 (Tex.Cr.App. 1987). An
exception to this rule exists where it appears that the question or
answer is clearly calculated to inflame the minds of the jurors and
is of such a character as to suggest the impossibility of
withdrawing the impression produced on their minds. Kemp v. State,
846 S.W.2d 289, 308 (Tex.Cr.App. 1992). The issue is whether the
jury was so affected by the question that they were unable to
disregard it as instructed. Huffman v. State, 746 S.W.2d 212, 218
(Tex.Cr.App. 1988).
Even if we concluded the question was calculated to inflame
the minds of the jury, we cannot conclude that the question or
answer was of such a character as to suggest the impossibility of
withdrawing the impression produced on the jurors’ minds. At
worst, the question placed before the jury the idea that the
appellant may have made some statement indicating he thought the
-32-
female clients at the state school were seductive or sexually
aggressive. There was nothing in the offered statement indicating
appellant actually had sexual intercourse with the female clients.
The question and answer did not suggest that appellant had
confessed guilt where the appellant was denying guilt at trial.
See Ladd v. State, 629 S.W.2d 139 (Tex.App.--Dallas 1982, pet.
ref’d).
Assuming without deciding that the evidence offered was
inadmissible, we conclude beyond a reasonable doubt that any error
was cured and otherwise rendered harmless by the trial court’s
instruction to disregard. The trial court did not err in
overruling appellant’s motion for new trial. Accordingly, we
overrule appellant’s third point of error.
In conclusion, we overrule appellant’s three points of error
and affirm the judgment of the trial court.
Carlton B. Dodson
Justice
Quinn, J., concurring
Publish. Tex. R. App. 47.4.
-33-
Griffith, Russell Alan v. State
Combined Opinion