M.C., the father of B.C., appeals the trial court’s order adjudicating B.C. dependent as to him. We are asked to determine whether the trial court erred in finding prospective harm to one child based upon the sexual abuse of another child. We conclude there is a lack of competent, substantial evidence to support a finding of dependency and reverse the trial court’s adjudication of dependency.
M.C. is married to T.C. They are the natural parents of B.C., a three-year-old boy.1 T.C. is the natural mother of A.M., a daughter by a previous marriage. Both children resided together with M.C. and T.C. until the events leading to this proceeding. On June 15, 2004, when A.M. was about eleven years old, the Department of Children and Family Services (DCF) filed a petition seeking emergency *765shelter for the children based on allegations by A.M. that M.C. had sexually molested her. The shelter petition was granted.
On June 30, 2004, DCF filed its petition for dependency of both children. The petition alleged that both M.C. and T.C. were offending parents as to B.C. The petition alleged that M.C. sexually abused his stepdaughter, thus placing B.C. “under substantial risk or [sic] imminent threat of harm, abuse or neglect, within the meaning and intent of Chapter 39, Florida Statutes.” 2 There were no allegations of actual abuse of B.C. T.C. consented to the adjudication of dependency as to both of her children. M.C. denied the allegations of abuse.
The trial court conducted hearings in September 2004 and June 2005 and rendered an order on July 12, 2005, adjudicating B.C. dependent as to his father, M.C. The order made the following finding:
Given there was no corroborating evidence to support these allegations, the Court must weigh the credibility of [A.M.] against [M.C.’s] credibility in assessing whether [M.C.] sexually molested his stepdaughter. In weighing this evidence, the Court finds that it is more likely than not that M.C. sexually abused [A.M.] in some manner.
The trial court heard the testimony of Dr. Robert Wright, a psychologist. Dr. Wright testified for DCF as an expert in the area of assessment and treatment of sex offenders. Dr. Wright had tested and evaluated M.C. and testified at length concerning the results of that evaluation. From his testimony, the trial court made the following findings to support its conclusion that DCF proved a nexus between the abuse of A.M. and the prospective abuse of B.C.:
Dr. Wright, the sexual abuse expert, testified that, assuming the father sexually abused [A.M.], there is substantial risk of imminent harm to [B.C.], based in large part, on the impulsive nature of pedophilia. In addition, the father has not sought any treatment in response to the alleged sexual abuse (admittedly because he denied it occurred).
The Court finds that the father’s conduct, and accompanying mental state, make it highly probable that, without intervention, the father will abuse or neglect another child.... Moreover, the father’s conduct, without any subsequent treatment, suggests behavior beyond his control and likely to continue.
The collective nexus findings suggest that M.C. has mental conditions that place B.C. at substantial risk of imminent sexual abuse unless M.C. receives sex offender treatment intervention. The trial court found additional support for the nexus in M.C.’s failure to seek treatment in response to the allegations of sexual abuse.
A determination of dependency is a mixed question of law and fact. See In re M.F., 770 So.2d 1189, 1192 (Fla.2000). Appellate courts must determine whether the trial court applied the correct law in making its decision and whether the decision is supported by competent, substantial evidence.
The law pertaining to prospective harm to one child by a parent based upon abuse, abandonment, or neglect committed against another child is well settled. In In re M.F., the Florida Supreme Court eschewed a per se rule that conviction of a sexual offense against one child was by *766itself sufficient evidence to support an adjudication of dependency as to other children. Instead, the court required additional proof that would establish a nexus between the prior abuse and prospective abuse of siblings. Id. at 1194. It said:
We conclude that the flexible approach ... rather than the per se rule is more consistent with the plain language of the Act.... While the commission of such an act may be highly relevant, it is not automatically dispositive of the issue of dependency. A court should instead focus on all circumstances surrounding the petition in each case.
Id.
Going further, the court, in a footnote, said:
For instance, in cases involving a parent’s prior sexual act on a different child, a court may consider the following: any similarity between the prior act and the pending case; the temporal proximity of the prior act to the pending case; any treatment received by the parent following the act; and the testimony (if appropriate) of professionals and experts.
Id. at 1194 n. 13.
In In re C.M., 844 So.2d 765, 766 (Fla. 2d DCA 2003), this court stated:
Children who have not been abused may be adjudicated dependent based on abuse inflicted upon their siblings; however, the evidence must demonstrate a nexus between the abuse and any prospective abuse of another sibling.... Generally, this nexus is established when the parent has a mental or emotional condition that will continue, such as mental illness, drug addiction, or pedophilia, and which will make it highly probable that in the future the parent will abuse or neglect another child.
The facts of this case and the testimony of Dr. Wright presented the trial court with troubling and perplexing questions. The issue of whether the sexual abuse actually occurred was particularly difficult. It was a nightmarish “she said, he said” contest between two witnesses of questionable credibility over an issue of fact that carries extremely dangerous or adverse consequences should the fact finder believe the untruthful witness. The record supports a conclusion that A.M. had a motive to fabricate a false tale and had a poor reputation for veracity. Psychological tests performed by Dr. Wright suggested that M.C. had a tendency to externalize blame. The trial judge, without being specific,3 found that “it is more likely than not that [M.C.] sexually abused [A.M.] in some manner.”4 Although our task on this appeal is not to second guess the trial court’s finding of sexual abuse, the circumstances surrounding that decision play into the flexible approach to finding a nexus supportable by substantial, competent evidence.
In determining a nexus to B.C.’s dependency, the trial court did not look to or *767rely upon the evidence related to the sexual abuse events. Instead, the trial court based its findings entirely upon Dr. Wright’s testimony. Dr. Wright repeatedly said he did not rely on anyone’s version of what happened between A.M. and M.C. Furthermore, he had no opinion on whether or not the sexual abuse actually occurred. His opinions and recommendations were based entirely upon his evaluation of M.C. and family background information.
Upon close examination of Dr. Wright’s testimony, we conclude that his opinion, based on his evaluation, was that M.C. was not the kind of person who would commit a sex offense on a child. Dr. Wright’s psychological testing of M.C. disclosed no significant clinical psychopathy or mental disorders.
Dr. Wright testified that there are three pathways by which adults who sexually abuse a child may recidivate against another child: pedophilia, sociopathic behaviors, and impulsivity. He testified that he did not have any data to support a finding that M.C. is either a pedophile or a sociopath. On the matter of impulsivity, he found that most of the risk factors5 for child sexual abuse by impulsive act did not apply to M.C. (We note that many of the risk factors Dr. Wright mentioned were the same as those discussed in In re M.F. and In re C.M.) Dr. Wright mentioned that some risk factors were present, specifically M.C.’s tendency to externalize blame and his wife’s interest in “swinging.”6 However, Dr. Wright testified that swinging is an indicator of sexual promiscuity rather than sexual deviance.
Dr. Wright offered three opinions on the nexus issue, all based solely upon his evaluation. The first opinion was that M.C. did not have an emotional or mental health problem that would make it “highly probable” that he would sexually abuse his son. Second, the risk of sexual harm to B.C. was minimal. Third, whatever the risk was to B.C., it was not imminent,.7 Given the trial court’s total reliance upon Dr. Wright in making nexus findings, these opinions do not provide the competent, substantial evidence required to establish the abuse-dependency nexus.
What makes this case troubling at the appellate level is another opinion Dr. Wright offered. He was asked:
By Mr. Williams [attorney for DCF]: [I]f the Court makes a finding that [A.M.] was, in fact, a victim of sexual abuse by [M.C.], you believe that there is a substantial risk that if untreated, [M.C.] could commit an act on [sic] of sexual abuse against his own child, correct?
Dr. Wright: Correct.
This is a hypothetical question that asked Dr. Wright to assume that the sexual abuse actually happened. Given Dr. Wright’s statements, findings, and opinions from his evaluation, and given his repeated refusal to render an opinion on whether *768M.C. actually committed the offense, this expert opinion evidence is nothing more than a restatement of the per se rule our supreme court refused to accept as dispos-itive of the question of the dependency of one child based on the abuse of another. The mere fact that it surfaces as an expert opinion rather than a legal presumption does not change its legal effect in this case: B.C. is being adjudicated dependent because, and only because, the trial court concluded that B.C.’s father sexually abused B.C.’s half-sister.
The trial court’s finding of substantial risk of imminent harm to B.C. by M.C. based in large part on the impulsive nature of pedophilia was not supported by competent, substantial evidence. The trial court’s finding that M.C.’s conduct, and accompanying mental state, make it highly probable that, without intervention, the father will abuse or neglect another child is also not supported by competent, substantial evidence. All that is left standing is the per se rule disguised in expert testimony that sexual abuse of one child creates a substantial risk of abuse to another child. The findings in this case did not move the nexus-finding process in a linear fashion from abuse of A.M. to dependency of B.C. Instead, it went in a circle and wound up back at the beginning.
The trial court also found that M.C. failed to seek treatment. The trial court cited this “failure” in support of the adjudication of dependency. The failure to seek treatment while a disputed sexual abuse allegation is in litigation cannot be a nexus consideration because it puts the accused parent in a “no-win” situation at trial. If the parent seeks treatment, that fact may be used as an admission of culpability. If the parent refuses, one might conclude, as the trial court did here, that he or she poses a greater risk of imminent harm to other children. Having or not having sex offender treatment does not prove a nexus. What sex offender treatment professionals and their records say about the person treated and his or her potential or probability of recidivating is where the proof lies.
Finally, we note that Dr. Wright did not even recommend sexual offender treatment for M.C. Instead, Dr. Wright recommended a minimal intervention in the form of education about limits, values, and consequences. Therefore, the trial court’s finding that M.C. poses substantial risk to B.C. without sex offender treatment is unsupportable.
Accordingly, we conclude that the trial court’s finding of a nexus between the sexual abuse of A.M. and the adjudication of B.C. as a dependent child is not supported by competent, substantial evidence in the record. Absent the requisite nexus, we are required to reverse the trial court’s order of dependency.
Reversed.
SALCINES and WALLACE, JJ., Concur.. B.C. was two years, nine months old when he was removed from his parents’ custody. He was almost four years old when he was adjudicated dependent. He is described as a three-year-old for purposes of this opinion because that is the birthday nearest the date he left the custody of his parents.
. The definition of a "child who is found to be dependent” in section 39.01(14)(f), Florida Statutes (2003), includes children found “[t]o be at substantial risk of imminent abuse, abandonment, or neglect by the parent or parents or legal custodians.”
. Florida Rule of Juvenile Procedure 8.260(a) requires all orders to be reduced to writing and they "shall contain specific findings of fact and conclusions of law.” The trial court’s order specifically recites what A.M. testified about but does not indicate whether the fact finder believed all or any part of it. The finding of fact merely states that sexual abuse occurred “in some manner.”
. Such a finding suggests that the trial judge resolved this disputed issue of fact based upon the force and effect of the evidence as a whole and by a narrow preponderance. It does not suggest the findings were sustained by clear and convincing evidence. In pertinent part, Florida Rule of Juvenile Procedure 8.330(a) provides that "[i]f the court is of the opinion that the allegations have been sustained by clear and convincing evidence, it may enter an order so stating.”
. E.g., low mental function, depression, anxiety, drug and alcohol addiction or abuse, and criminal history.
. There was no evidence that M.C. and his wife engaged in swinging but that they had expressed an interest in looking into it. Incidentally, there was evidence that M.C. had watched pornography on the Internet, but that was not mentioned as an impulse-driving factor.
. Dr. Wright's opinion about imminent risk came up in two contexts. First he opined that risk was not imminent because of the absence of risk factors. The second time it came up was in a response to a question by the court to which he answered by stating, "There was not enough information to determine something like [imminent risk].”