This is a paternity case in which the appellee, Tammy Clements, claims that the appellant, Edwin Boyles, is the father of her child, Michelle Pugh, who was born out of wedlock on December 6, 1985. Clements introduced the results of a blood test report from National Paternity Laboratories, Inc. into evidence over Boyles’s objection. The trial court found that Boyles was Michelle’s father.
On appeal, Boyles claims 1) that the trial court erred in refusing to dismiss the cause of action when Clements concluded her case and when he concluded his case, and 2) that the trial court erred in permitting the introduction of the results of the blood tests.
We agree that the trial court erred in permitting the introduction of the results of the blood tests and reverse and remand; consequently, we need not address Boyles’s first point of error.
Arkansas Code Ann. § 9-10-108 (Supp. 1989) addresses blood tests in a paternity action and provides in pertinent part as follows:
(a) At the request of either party in a paternity action, the trial court shall direct that the defendant, complainant, and child submit in one (1) or more blood tests or other scientific examinations or tests ... to determine whether or not the defendant can be excluded as being the father of the child and to establish the probability of paternity if the test does not exclude the defendant.
(b) The tests shall be made by a duly qualified physician or physicians, or by another duly qualified person or persons, not to exceed three (3), to be appointed by the court. *577(c)(1) The results of the tests shall be receivable in evidence.
(c)(2)(A) A written report of the test results by the duly qualified expert performing the test, or by a fully qualified expert under whose supervision and direction the test and analysis have been performed, certified by an affidavit duly subscribed and sworn to by him before a notary public, may be introduced in evidence in illegitimacy actions without calling the expert as a witness. If either party shall desire to question the expert certifying the results, the party shall have the expert subpoenaed within a reasonable time prior to trial.
(c)(2)(B) If the results of the paternity tests establish a ninety-five percent (-95 %) or more probability of inclusion that the defendant is the natural father of the child and after corroborating testimony of the mother in regard to access during the probable period of conception, such shall constitute a prima facie case of establishment of paternity and the burden of proof shall shift to the defendant to rebut such proof.
* * * *
Boyles claims that the blood test report did not comply with the requirements of section 9-10-108, thereby precluding the introduction of the results of the report into evidence, because the report did not have the proper jurat, was not in affidavit form, did not state who performed the test or give the qualifications of such person, and did not show who was sworn and to what they had subscribed.
Although the chancery court has broad discretion in determining whether such reports should be admitted into evidence, we hold that the chancellor abused his discretion in this case.
The court of appeals addressed the necessity of compliance with section 9-10-108 in the admission of reports containing the results of blood tests in Ross v. Moore, 30 Ark. App. 207, 785 S.W.2d 243 (1990). In that case, the appellee totally failed to establish the prerequisite statutory foundation for the admission into evidence of the results of a blood test report because there was nothing in the report to indicate the identity of the person who *578performed the test or whether the person who performed the test was a duly qualified expert. Although the report was signed by a Dr. Smith and stated that Dr. Smith was the laboratory director, there was nothing in the report to indicate that Dr. Smith was the person who performed the test, or that he was a qualified expert.
We adopt the rationale and conclusion of the court in Ross v. Moore, supra, when it discussed the admissibility of a blood test report and requirement for strict compliance as follows:
Prior to the adoption of Ark. Code Ann. § 9-10-108, this report would have been considered inadmissible hearsay, and in order to be admissible and fall into one of the exceptions to the hearsay rule, certain foundational requirements must have been met. . . .
The purpose of § 9-10-108 is to relax these foundational requirements and make it less difficult to introduce paternity testing results into evidence. However, to insure the reliability of this type of testing, the foundational prerequisites in the statute must be met. In light of the fact that recently developed genetic testing can, with a high degree of certainty, identify the father of a child and, thus, be viewed as conclusive by the fact-finder in paternity suits, we do not think that strict adherence to the statutory prerequisites is unreasonable.
(Citations omitted.)
In this case, the report from National Paternity Laboratories, Inc. stated:
On April 7, 1988 blood specimens were received from the Poinsett County Child Support Enforcement Agency on Edwin Boyles (alleged father), Tammy Clements (presumed mother) and Michelle Pugh (child) for the purpose of paternity exclusion studies. Completed identification forms of all parties are in our files.
* * * *
The HLA data listed above indicates that Edwin Boyles cannot be excluded as a possible father of Michelle Pugh.
* * * *
*579The Red Cell data listed above indicates that Edwin Boyles cannot be excluded as a possible father of Michelle Pugh.
Utilizing the HLA and Red Cell data, two calculations have been performed to determine the likelihood of paternity. The first calculation is the Probability of Paternity which computes the chances that a single sperm carrying all the necessary genes could be produced by the alleged father in contrast to the frequency in which a sperm might be produced by a random man. This is expressed as the percentage certainty. 100% would be equivalent to proof of paternity, 0 % would be equivalent to non paternity. A value of 50% would mean that the alleged father and the random man have an equal chance of being the biological father. In this case, the Probability of Paternity is 95.44 %.
The next calculation, the Paternity Index, is a means of expressing the likelihood of paternity as a ratio. In this case, the Paternity index is 21 to 1.
The likelihood of paternity based on the calculations according to verbal predicates assigned by Hummel is considered very likely.
The report was signed by Dr. Randall A. Smith, Ph.D., Laboratory Director, and Robert W. Gutendorf, M.S., M.T. (ASCP) SBB, Scientific Director, and their signatures were notarized.
As in Ross v. Moore, supra, there is nothing in the report to indicate the identity of the person who performed the test or whether the person who performed the test was a duly qualified expert. Although the report is signed by Dr. Smith and Mr. Gutendorf and states their positions to be Laboratory Director and Scientific Director respectively, there is nothing in the report to indicate that these two men performed the test or that they are qualified experts.
As a result, Clements has failed to establish the prerequisite statutory foundation for the admission of this evidence, and on this record we cannot say that a proper foundation was laid. See also Newton v. Clark, 266 Ark. 237, 582 S.W.2d 955 (1979); Simolin v. Wilson, 253 Ark. 545, 487 S.W.2d 603 (1972).
*580 On appeal, we consider the evidence in the light most favorable to the appellee. McGuire v. Bell, 297 Ark. 282, 761 S.W.2d 904 (1988) (citing Constant v. Hodges, 292 Ark. 439, 730 S.W.2d 892 (1987)). Although we try chancery cases de novo on the record, we do not reverse a finding of fact by the chancellor unless it is clearly erroneous. McGuire v. Bell, supra (citing Milligan v. General Oil Co., 293 Ark. 401, 738 S.W.2d 404 (1987)).
In reviewing Clements’s testimony, we note that, at best, the evidence is in conflict as to the date of termination of her relationship with Boyles. She affirmatively stated during the course of the trial that:
Q Okay. Let me ask you this. In 1985, uh, did you know Mr. Edwin Boyles?
A Yes.
Q Did y’all have any type of relationship in 1985?
A Uh, we broke up on June, uh, the 18th of the year before she was born.
Q Okay. What do you mean, “broke up?”
A Uh, we never seen each other any day after that. And the reason I know that date so well, it was my Mom’s birthday.
Q Now, what year are you talking about? Was it after Michelle was born?
A No, it was before.
* * * *
Q Okay. Uh, do you remember about when this baby was conceived?
A It was late March or April — or the first of April of ’85 — uh—’84. She was born nine months later.
Q But she was born December, 1985?
A Yes.
* * * *
*581Q Okay. Now, when did you start seeing Mr. Boyles? A I don’t know, exactly.
Q Do you remember the year?
A Uh, no.
Q When did you stop seeing him?
A June the 18th, of ’84 — was the last . . .
* * * *
Q Okay. We start with: “The child was born in December, ’85.” That was my first question. Then I asked you: “During 1985, did you have sexual relations with Edwin Boyles in 1985?”
A Uh, (shaking head.)
Let the record show she is shaking her head in the negative . . .
Yet, Clements also testified that she was pregnant and still going out with Boyles and that Boyles took her to the Health Office for a doctor’s appointment one day.
It is not at all clear on what basis the trial court made its judgment, as the chancellor found that Boyles “by virtue of the testimony, evidence submitted herein, and for other reasons, is hereby found to be the biological father of Michelle Lee Pugh.” We cannot tell what weight, if any, the trial court gave to Clements’s testimony, especially since the trial court also admitted and considered the paternity report, which we now exclude. In this matter, the credibility of the witnesses is a critical factor, and the chancellor obviously is in a superior position to decide such issues.
For these reasons, we are unable to decide this case on de novo review. Therefore, we reverse and remand for proceedings consistent with this opinion.
Dudley and Hays, JJ., would affirm.
*582Price, J., would reverse and dismiss.