OSCAR F. HUDSON, APPELLANT,

v.

STATE OF FLORIDA, APPELLEE

PLEUS, J.

Hudson appeals his convictions for sexual battery upon a person under 12 years of age. He argues that the trial court erred in denying his motion for mistrial after the prosecutor referred to him in closing argument as -a pedophile. He also argues that the trial court erred in admitting the testimony of the state’s DNA expert regarding population frequencies. This court has jurisdiction pursuant to Florida Rule of Appellate Procedure 9.140. We remand, but only for a limited evidentiary hearing consistent with Miles v. State, 694 So. 2d 151 (Fla. 4th DCA 1997).

Hudson argues that the trial court erred in failing to grant his motion for mistrial based on the prosecutor’s reférence to him as a pedophile in closing argument. The prosecutor in her closing argument argued that “pedophiles who snatch children generally are not stupid enough to do it in a location where they are likely to be caught. They wait until the opportunity — .” Hudson asserts this comment was unfairly prejudicial because' it could be interpreted as suggesting that he had a history of criminal sexual activity with children.

A trial-court’s ruling on a motion for mistrial is within its sound discretion and should only be granted when necessary to ensure that the defendant receives a fair trial. Gore v. State, 784 So. 2d 418 (Fla.2001). It is well established, however, that the use of “pedophile profile” testimony as substantive evidence of guilt constitutes error. Flanagan v. State, 625 So. 2d 827 (Fla.1993). In the instant case, while the prosecutor did not attempt to introduce pedophile profile testimony into evidence, her comment in closing argument was nonetheless prejudicial-because, in addition to suggesting Hudson had committed prior illegal sexual acts involving children, it also improperly suggested a profile-type argument, namely, that if Hudson has certain traits which fit the offender profile, he must have abused the victim. See Flanagan at 828.

As in Flanagan, however, we find the error in this case harmless. Initially, we note that the prosecutor did not repeatedly refer to Hudson as a pedophile. Rather, she made a single reference to the word pedophile in her closing argument. See Flanagan at 830. Additionally, the state presented overwhelming evidence of Hudson’s guilt. The 11 year old victim testified in detail about how Hudson digitally penetrated her and choked her. At the first opportunity, she reported this crime to her grandfather. When her grandfather confronted the defendant, the defendant told him that he had “just used my finger to her.” The physical examination of the victim revealed bruising on the victim’s neck and abrasions and lacerations to her vaginal and anal areas. This physical evidence corroborated the victim’s testimony. Moreover, Hudson admitted to police that the victim was inside his house on the day of the incident. When the detective told him he needed fingernail scrapings, Hudson began trying to clean his fingernails. Despite Hudson’s efforts, the fingernail scrapings revealed DNA consistent with the victim. In sum, we conclude that there was no reasonable possibility that the prosecutor’s sole reference to the word pedophile in closing argument affected the verdict.

Hudson also argues that the trial court erréd in allowing the state’s DNA expert to testify about population frequencies because the expert was not qualified to testify in the field of population genetics. In Murray v. State, 692 So. 2d 157 (Fla.1997), the supreme court explained that the DNA testing process consists of two distinct steps — a methodology for determining DNA profiles, as well as the statistical calculations used to report the test results — both of which are subject to the Frye test.1 Murray at 161. When testifying about the statistical analysis, the “expert must, at the very least, demonstrate a sufficient knowledge of the database grounded in the study of authoritative sources.” Id. at 163. However, it is “not absolutely necessary for an expert witness to demonstrate practical experience in the field in which he will testify” or is it required that the expert “helped to assemble the database.” Id -The trial judge has the sole responsibility to make this determination. Id.

In the instant case, defense counsel stipulated that the state’s DNA analyst, Emily Booth, was an expert in serology and DNA analysis, but when she began testifying about population frequencies, the defense objected on the basis-that she wasn’t qualified as an expert in population genetics. The trial court overruled Hudson’s objection without inquiring into Booth’s expertise in this area.

The state makes three arguments in support of the trial court’s ruling. First, the state argues that based on Booth’s experience, including a B.S. in forensic science, nine months of training in conventional serology, ten months of training in DNA analysis, completing over 400 DNA profiles and testifying as an expert 11 times, it is “common sense that this includes not only the forensic serology component of establishing a DNA.profile, but applying that information to an underlying database.” While the state’s argument may be true, the trial court must specifically determine if the expert can “demon strate a sufficient knowledge of the database grounded in the study of authoritative sources.” Booth did not do this before she was allowed to testify.

Second, the state argues that Booth’s testimony demonstrates that she was “well grounded in how the statistics were developed and applied.” On de novo review, there is some testimony to support this argument. Specifically, Booth testified as follows:

A. Yes. Thirteen markers, when you compile the information of - all the different frequencies that they occur in the population and multiply them together, you get a total frequency of occurrence or the frequency of occurrence of this profile. And it’s done on a computer in our laboratory, that’s been checked out. * * *

Q. You were explaining to the jury about the frequency of the profile.

A. Yes. I then entered this profile into a computer for statistical analysis, and it generates the frequency of occurrence of this profile among three common ethnic groups.

Q. And what are those ethnic groups?

A. Caucasians, Blacks, and southeast Hispanics.

Q. You didn’t develop this computer program that does this, correct?

A. No, I did not.

Q. But the people that did use known samples in their research from these various populations, Hispanic and the like, correct?

A. Yes.

Q. You know how many people they used?

A. For Florida?

Q. Uh-huh. For the program you use.

A. For the program we use-they used 800 Caucasians, approximately-350 blacks, and I believe 250 southeast His- . panics.

Q. How many blacks are there in the state of Florida, African Americans?

A. I’m not-sure.

Q. Certainly more than 350?

A. Yes, however—

Q. So it. is based upon the testing of 350 blacks in the state of Florida that we’re able to reach an opinion on the population, genetics that go out to quadrillions, that’s all I’m asking?

A. Yes. And it’s been approved by Bruce Weir, who’s a known statistician in the field and respected.

In determining whether this testimony was sufficient to meet the test, we look to Miles v. State, 694 So. 2d 151 (Fla. 4th DCA 1997). In that case, the state’s expert testified as follows:

So, what we do is we have, we would go out' and test 100 or 200 individuals and find out how often a specific type occurs like the one type of DQ Alpha 1.2,2. We would just, go out and test and see how often it occurs. Each one would do -that. We can multiply that together because they are inherited separately, they’re not linked, and when you multiply that number together you find out what the frequency is.

Miles at 152. The Fourth District noted that:

In this case, the trial court did not determine that the statistical methodology employed by Mr. Ritzline was generally accepted. Nor does the record clearly reveal just what statistical technique was applied. While Mr. Ritzline briefly described how the statistical data was procured, this limited explanation, absent some speculation, does not identify the statistical methodology, nor does it address its general acceptance.... We are also unable to discern from the record whether Mr. Ritzline was properly qualified to report population frequency statistics.

Id. (footnote omitted). The court remanded for a limited evidentiary hearing, stating:

Because the record does not reveal the statistical methodology employed in this case, or Mr. Ritzline’s qualifications to present the statistical evidence, we remand for a limited evidentiary hearing similar to the one ordered in Brim. On remand, the trial court, is to (1) assess Mr. Ritzline’s competence to present the statistical evidence; and (2) clarify the exact methods used in calculating the DNA statistics and then conduct a Frye hearing to determine the general acceptance of the employed statistical techniques. If Mr. Ritzline is adequately qualified,- and if the statistical methodology orighially employed passes the Frye test, appellant’s conviction is to stand. Otherwise, appellant must be afforded a new trial.

Id. 'at 153 (footnote omitted).

Similarly, in the instant case, the record does not reveal the statistical methodology employed, beyond simple multiplication. There was no testimony regarding how the population frequencies of the genetic markers were determined. There was no testimony demonstrating that the methodology used is generally accepted in the scientific community. Additionally, although Booth did provide some limited information about the database, it was not sufficient to demonstrate “sufficient knowledge of the database grounded in the study of authoritative sources.”

Third, the state argues that even if there was an insufficient predicate for the statistical testimony, it was harmless error. We disagree. Booth testified that the DNA found under Hudson’s fingernails was consistent with the victim’s DNA and the frequency of that match occurring was “one in four quadrillion blacks.” That testimony provided a crucial piece of physical evidence to corroborate the victim’s testimony and refute Hudson’s testimony. Without it, undoubtedly the state’s case would have been weaker. Consequently, we remand this case to the trial court for a limited evidentiary hearing consistent with the instructions in Miles.

REMANDED FOR A LÍMITED EVI-DENTIARY HEARING.

SHARP, W., J., concurs.

THOMPSON, C.J., concurs in result only.

. Under Frye v. U.S., 293 F. 1013, 1014 (App. D.C.1923), before admitting into evidence the testimony of an expert witness concerning a new scientific principle, a trial court must determine (1) whether such expert testimony would assist the jury in understanding the evidence or in,deciding a fact in issue; (2) whether such testimony is based on a scientific principle which has gained general acceptance in that particular scientific community; and (3) whether the expert witness is sufficiently qualified to render an opinion on the subject.