Father appeals from a trial court judgment pursuant to ORS 109.119 that awarded custody of his children to grandmother, who is father’s mother. The issue is whether grandmother has proved by a preponderance of the evidence that father cannot or will not provide adequate love and care for the children or that placement of them in his custody will cause an undue risk of physical or psychological harm to them. Encompassed within that burden of proof is a statutory and constitutional presumption in favor of father that grandmother must overcome. See O’Donnell-Lamont and Lamont, 184 Or App 249, 56 P3d 929 (2002). On de novo review, we reverse because grandmother has not overcome the presumption that favors biological parents in custody disputes between parents and nonparents.
Father and mother, his former wife, had three daughters, B, E, and H. The children were 10, 8, and 6 years old in June 2000, the time of the custody hearing. In 1995, father and mother were separated, and a petition to dissolve their marriage was pending. The children were living with mother in Bend, while father was living in Portland. During the dissolution proceedings, grandmother and father learned that mother had exposed the children, particularly B, to sexual and other abuse while they were in her care. With grandmother’s help, father obtained temporary custody of the children and moved them and himself to grandmother’s home in Bend, where they stayed until 1999.1
Before he returned to Bend, father was working at a bar in Portland. He had also engaged in prostitution for money and drugs. In 1993 and 1994, he was drinking heavily, using drugs, and talking about suicide. His family planned an intervention on his behalf, but he avoided participation in the effort. When father first moved back to Bend in 1995, he had only limited involvement with the children. He cooked many of their meals and spent time with them in the afternoons, but he rarely attended parent-teacher conferences or *528other school functions. Father had difficulty controlling his temper with the children, and when he was angry he yelled at them and called them obscene names. He spent most of his time working on the computer, much of it at night, with the result that he slept during a large part of the day. Grandmother had a more significant role in the children’s lives than did father, although she frequently worked well into the evening and got home after the older children were in bed. H was especially attached to grandmother and stayed up late to see her, compensating by taking naps during the day (she was too young for school). Two of father’s sisters also played important roles in caring for the children; one of them, together with her two daughters, lived in the family home for over a year.
Father did not have a steady job while he was in Bend, but he worked on a number of computer projects, which, among other things, helped him to develop his skills in the field. He began with a website and other work for grandmother’s real estate business and moved onto similar work for other Bend businesses. The record does not indicate how much he earned from that work, but it is clear that grandmother provided the primary financial support for both him and the children while they lived with her. In the first years that he lived in Bend, father also used drugs occasionally, and in a few instances he was an escort for older, wealthier men, including accompanying one on a business trip to Japan. Father appears to recognize in his brief that his actions during those years would justify an award of custody to grandmother under any standard.
In late 1997, father met Michael Chism, a truck driver who lived in Roseburg and who is about 15 years older than father. Chism has two children; one was in college and the other, who apparently is about E’s age, visited him every other weekend. Father and Chism developed a relationship and began visiting at each other’s homes. Father occasionally took the children with him to Roseburg. Chism is an alcoholic, and at times father’s relationship with him involved excessive drinking. However, in other ways father saw Chism as a mature role model, particularly with regard to Chism’s responsibility toward his children. In late 1998 or early 1999, father decided to move to Chism’s house in *529Roseburg and to take the children with him. At about the same time, a confrontation with B led him to realize that his yelling and swearing at the children was inappropriate and damaging, and, according to both of them, he stopped that kind of conduct.2
Father moved the children into Chism’s house in Roseburg at the end of May 1999, shortly before the end of the school year, and enrolled them in Roseburg schools. In early June, grandmother filed this petition and obtained a temporary custody order; the police took the children from their Roseburg schools without notice to father and returned them to Bend. Father regained custody in August and continued to exercise custody up to the time of the hearing in June 2000, subject to frequent visitation with grandmother. It is of some significance that at that time the trial judge in this case, acting as a juvenile court judge, returned the children to father rather than placing them with grandmother. After moving to Roseburg, father obtained a job that allowed him flexibility to be with the children and to participate in school activities.
In contrast to his actions during his early years in Bend, father’s parenting of the children in Roseburg in the 10 months before the hearing was exemplary, and the uncontroverted evidence is that the children thrived in their home in Roseburg. Father was actively involved in all aspects of their lives. He got them ready for school every morning and walked the younger girls to their school a few blocks away. B’s school was across the street, and she usually went by herself. He picked up the two older girls at the end of the school day. For much of the year, Chism picked up H when he was in town. However, shortly before trial he went on an alcoholic binge and was incapacitated for several days. Thereafter, he entered a rehabilitation program, and the daycare provider took over the task of picking up H. There is no evidence about what impact, negative or otherwise, Chism’s difficulties had on the children.3
*530During the same time period, father was active in the children’s schools, assisting in their classrooms and accompanying them on field trips. At home, each of the children had appropriate chores. Father helped them with any homework. He was appropriately warm and affectionate with them. In the evening, they read and watched television. Father took care to avoid exposing them to inappropriate programs and videos. The children played at a basketball hoop in the backyard (there is a special bucket on the ground so that H can also participate). The children testified that they enjoyed living with father and that they preferred to stay with him.4
All of the children did well in school while they were in Roseburg. Their teachers uniformly described them as bright and articulate and said that they made friends easily. B began the year almost two grades below grade level in mathematics but finished the year at or above grade level. One teacher rated father’s participation in school activities as being in the ninetieth percentile among parents. No teacher indicated that the children showed signs of being at risk or in need of.counseling. B’s school counselor testified that the best *531thing for her to do to overcome any problems from her previous experiences was to get on with her life, and, in the counselor’s opinion, she was succeeding in that endeavor. Grandmother does not seriously contend that she would be entitled to custody if the only evidence of father’s care of the children related to the period of time that the children lived with him in Roseburg. Despite the above evidence, the dissent appears to give little or no weight to the evidence of father’s care and interaction with the children while they were living with him in Roseburg.
Also, two experts conducted studies related to the custody issue. They had different purposes for their studies, differences that affect the weight to be given to their testimony. Dr. Catherine Bolstad, a psychologist, conducted a custody study in which her focus was on comparing grandmother and father as potential custodial parents. Her primary interest was to determine the best interests of the children, and she did not give any particular weight to father’s status as the biological parent. In contrast, Joseph Mazza, a licensed clinical social worker, conducted a home study and parenting evaluation of father to determine the current conditions and circumstances of the family. He did not attempt to compare father’s home with grandmother’s but, rather, to evaluate it in light of his experience with many other homes and his knowledge of general theory about parenting. Those different purposes affected the nature of the information that the experts gathered and the conclusions that they reached based on that information. Mazza’s focus is closer to the matters that are relevant to the applicable legal standard than is Bolstad’s. As a result, his information and opinions are generally more helpful in deciding this case than are Bolstad’s.5
Bolstad conducted extensive interviews with father, Chism, grandmother, and others, observed the children interacting separately with the two potential custodial families in her office, received information from other sources, and administered psychological tests. She did not visit either *532home, in part because she believes that home visits frequently lead to an artificial environment as those involved put on a show for the observer. Mazza interviewed father and Chism, administered psychological tests, and spent many hours observing the family at home, including over five hours with the children alone. He testified that observing the children in their home environment was crucial to understanding how they and father actually related to each other.
Neither Bolstad nor Mazza found indications of significant psychopathology in either father or grandmother. Although Bolstad was concerned that father might have a tendency toward narcissism and placing his interests ahead of the children’s, Mazza saw no indication of that problem in his observations of the home environment in Roseburg. Bolstad recommended that grandmother receive custody for several reasons: (1) at least in part because of their experiences with their mother, the children need a good parent more than normal children do; (2) the children had lived with grandmother from 1995 to 1999; (3) father has a history of drug and alcohol involvement and had not been involved in the children’s school work and care until recently; (4) father’s taking the children out of school and moving to Roseburg shortly before the end of the school year demonstrated his unwillingness to put the children’s needs ahead of his own; (5) father does not adequately value the children’s relationship with grandmother; (6) father treats the children like adults and does not set limits and boundaries for them; and (7) the children, especially B, need counseling, which, in Bolstad’s opinion, father was less likely to provide than was grandmother.
We turn now to a description of the applicable legal standard in a case involving a custody dispute between a biological parent and a nonparent who has developed a child-parent relationship with the children.6 Because grandmother filed her petition before the effective date of the 1999 version of ORS 109.119, the controlling statute is ORS 109.119 *533(1997). Williamson v. Hunt, 183 Or App 339, 343-44, 51 P3d 694 (2002). ORS 109.119(2)(a) (1997) authorized a court to grant custody to a grandparent or other person who had a child-parent relationship with the child if the court determined by a preponderance of the evidence that it was appropriate and in the best interests of the child to do so.
The statute must be interpreted in the context of the applicable federal constitutional law. In Wilson and Wilson, 184 Or App 212, 55 P3d 1106 (2002), we considered at length the impact of the United States Supreme Court’s decision in Troxel v. Granville, 530 US 57, 120 S Ct 2054, 147 L Ed 2d 49 (2000), on the 1997 version of ORS 109.119. We concluded that, under Troxel, biological parents have a fundamental due process right “ ‘to make decisions concerning the care, custody and control of their children.’ ” Id. at 217 (quoting Troxel, 530 US at 66). After discussing decisions of the Oregon Supreme Court and this court, we held that that fundamental constitutional right supervenes a pure “best interest of the child” analysis in a custody dispute, “replacing it with one in which a fit biological parent will presumptively prevail over a nonparent unless the nonparent presents compelling reasons to overcome that presumption[.]” Id. at 219. Thus, the threshold issue is whether grandmother has proved that father “cannot or will not provide adequate love and care or that the children will face an undue risk of physical or psychological harm in [father’s] custody.” O’Donnell-Lamont, 184 Or App at 256-57.
In light of the above legal standards, we turn back to the evidence. When the children lived with grandmother from 1995 to 1999, they also lived with father, who despite his problems played a significant role in their lives. Although his role as a parent was clearly inadequate, it was not the total nullity that the dissent suggests, and it improved in the later years. Father’s history of drug and alcohol abuse during that period of time is probably the most serious concern to us, particularly in light of Chism’s continuing difficulties. Father insists that he enforces a “zero tolerance” policy for drugs in the home, and Mazza did not find anything to the contrary during his extensive home study. As with other aspects of father’s dramatic change in behavior, we cannot be entirely certain how longlasting those changes will be. Nonetheless, *534the record does not permit us to speculate that the children are presently at risk because of exposure to drugs if father is given custody, particularly in light of the presumption in his favor.
The other issues that concerned Bolstad also do not overcome the presumption. Father might have appropriately delayed the move to Roseburg, but, in light of his overall relationship with the children after the move, we are not persuaded that he placed his interests ahead of the children. Father testified that he does not think that the children need to see grandmother as often as the court ordered, and he clearly resents what he sees as her interference with his role as their parent. Parental decisions about the extent of contact with third parties are the kinds of decisions that are presumptively protected by the constitution. However, father also believes that the children and grandmother should remain close, and he has placed far fewer restrictions on their communication with her while they are in his home than she placed on their communication with him when she had custody.7
Bolstad’s concern that father treats the children like adults appears to be based in part on how B was dressed when she came for the interview that Bolstad conducted with father, compared to how she was dressed when she came with grandmother. Again, we are persuaded that the evidence in this regard is insufficient to overcome the presumption in favor of father. Finally, although Bolstad believed that the children needed counseling, the school counselor disagreed. Also, Mazza did not perceive the level of distress that would ordinarily require counseling, although he did not believe that counseling would be harmful to the children, particularly in light of the relationship between father and grandmother. On this record, father’s decision not to provide counseling unless the children request it does not rise to the level that would persuade us that granting custody to father *535will cause an undue risk of psychological harm to the children, as would be required to overcome the presumption.
Nonetheless, we have two continuing, significant concerns based on the record before us. First, there is the evidence of Chism’s alcoholism and its effect or potential effect on the home environment if father is granted custody. Secondly, and related to our first concern, we cannot be certain that father’s turnaround will continue. Mazza’s testimony that people often make significant life changes between the ages of 28 and 32, which is father’s age group, gives reason for hope, but father’s previous life-style, especially before 1995, weighs against that hope. Mazza suggested in his testimony that father could benefit from counseling or parenting classes, and we agree that those things could help reinforce the other changes that he has made.
What weight to give those concerns depends on to what extent they are predictors of the future, and, ultimately, we must decide whether those concerns suffice to rebut the presumption in favor of father that is part of the constitutional right that the Supreme Court described in Troxel. We are mindful that what we decide in this case about the sufficiency of the evidence in that regard constitutes precedent for purposes of other cases. However, each case must be decided on its own facts. Here, each concern is ameliorated by the fact that father exercised his parenting responsibilities in exemplary fashion for the 10 months that he had custody. That performance tends to render our concerns about the future speculative. Because of the speculative nature of those concerns, we conclude that we cannot give efficacy to the presumption and at the same time award grandmother custody on this record. Our concerns must give way to the law pursuant to Troxel. Consequently, we hold that grandmother has not overcome the presumption that is accorded father because he is the biological parent. It follows that he is entitled to custody of the children.
In reaching a different conclusion, the dissent fails to give adequate consideration to the legal standard that we must apply. Although it states that standard correctly, 185 Or App at 541-42 (Deits, C. J., dissenting), it does not use it in evaluating the evidence. The heart of the standard is that we *536analyze only whether father can provide adequate love and care for the children and whether they are at undue risk of harm in his custody. It is simply irrelevant to that determination whether grandmother might provide better care, for that consideration lapses back into a “best interests” test that is inapplicable. The dissent nevertheless relies heavily on Bolstad’s report and testimony, even though Bolstad’s express purpose was to make a custody determination based on the best interests of the children. In addition, Bolstad refused to visit father and the children at home to assess what was occurring outside the somewhat artificial environment of her office.
Some of Bolstad’s concerns, which the dissent echoes — particularly that the children might be at risk because of what they experienced while they were living with their mother or that they will suffer undue psychological harm if removed from grandmother’s custody — simply do not have any support in the evidence of their current situation. Witnesses who had no interest in the outcome of the case (their teachers and school counselor) uniformly stated that the children were doing well and that they demonstrated no indicators of needing counseling; thus, neither the negative effects that they may have experienced with their mother nor any negative effects from their removal from grandmother’s custody are apparent from their current behavior. Indeed, as the school counselor pointed out, it is often better for children to move on with their lives than to focus on things that happened previously. Because of the nature of Bolstad’s study, including its comparative nature and its focus on the past rather than the present and future, and because of the more current evidence of the children’s behavior and living situation in the 10 months leading up to trial, we do not give Bolstad’s testimony the weight that the dissent does.
In contrast to Bolstad’s testimony, Mazza’s home study focused on precisely the issues that are the most relevant to our decision. He evaluated how the children were doing with father, not how they might do with grandmother, and he concluded that they were doing very well. The dissent *537suggests that Mazza was not sufficiently aware of the situation while father lived in Bend because he spoke only to people whom father suggested.8 However, as the dissent notes, Mazza reviewed Bolstad’s report before conducting his study and thus was aware of her findings. The dissent does not explain why it was inappropriate for Mazza to rely on Bolstad’s interviews of the people that she had interviewed and to spend his time speaking with additional people whom Bolstad had not seen.
We also give weight to Mazza’s experience with the children’s protective services unit of the State Office of Services to Children and Families; he is currently the supervisor of 13 caseworkers in that unit. As a result of that experience, he is both aware of the factors involved in creating an environment that may present a danger to children and of the differences between a family that is at risk and one that is not. Unlike Bolstad, Mazza spent many hours observing father and the children in their home in Roseburg. That time included five hours with the children when father was not present. In light of Mazza’s experience and the time he spent with father and the children, we find his testimony highly probative.
What the dissent does not fully appreciate is that Mazza provided the most probative evidence regarding how father and the children will relate to each other in the home in the future. Bolstad’s opinion focused on the time period before the children lived with father in Roseburg, as did grandmother’s other evidence. While that evidence could be helpful in predicting how father and the children will relate in the future, it does not take into account the time period closest to trial. Only Mazza described how the children related during the most recent time period, which may well *538be the best predictor of how they will relate in the future. The picture that he provided is remarkably favorable to father, and he provided cogent reasons for believing that it will probably continue to be favorable. He recognized — as we recognize — that father’s previous actions raise concerns and that there are no guarantees of future behavior.9 We agree with him, however, that those concerns, by themselves, are insufficient to show by a preponderance of the evidence that father cannot provide adequate love and care to the children or that they will suffer an undue risk of harm in his custody. As a result, father is entitled to custody.10
Reversed and remanded with instructions to award custody to father.
Mother did not appear at the hearing on grandmother’s petition for custody of the children, and the court entered an order of default as to her. So far as the record indicates, she does not have any visitation rights.
There is no evidence that father’s outbursts have continued since that confrontation and his change in behavior.
The dissent refers to an incident in which Chism was incapacitated by alcohol when H came home. It fails to point out that father insisted that Chism get help and made it clear to him that, otherwise, he and Chism could not continue living *530together. As a result, Chism entered the rehabilitation program. Grandmother testified that H told her about the incident two or two and one-half months before the hearing. Father testified that Chism’s relapse occurred about one month or one month and one-half before the hearing. Because of the additional details that he described, we find father’s recollection to be more accurate. We also conclude that there was only one relapse rather than, as the dissent believes, two separate incidents. It is clear that father provided alternative daycare for the children after-wards and no longer relied on Chism for that purpose.
There is some suggestion in the record that B may have organized a united front on that point. There is also evidence that the previous summer, before the hearing that returned temporary custody to him, father encouraged the children to testify in his favor. In reaching its decision, the trial court apparently gave considerable weight to a tape recording of a telephone conversation that he had with the children on that subject. Because the tape was a year old at the time of trial, involved a different hearing, and came before the time period that father had custody, we do not give it the weight that the trial court did or the dissent does. In any event, we give the children’s testimony about their custody preferences little weight in light of their ages. What we find more persuasive is the uncontradicted evidence of father’s healthy and extensive involvement with the children in the 10 months preceding the trial in this case. We note, incidentally, that the reason that the tape exists is that grandmother would not let the children talk with father unless the conversation was recorded or took place on a speaker phone with someone else present. That fact, among other things, undercuts the dissent’s argument that grandmother will encourage the children’s relationship with father.
If we were to find that grandmother had rebutted the presumption, we would then need to decide the best interests of the children. Bolstad’s information and opinions would be more directly relevant to that issue.
As we noted in O’Donnell-Lamont, 184 Or App at 253 n 4, the relevant Oregon cases have all involved a biological parent and a nonbiological nonparent. We do not mean by our references to that fact to suggest that the rights of an adoptive parent would be different from those of a biological parent. See ORS 109.350.
If visitation is unduly restrictive in the future, the court has authority to award appropriate visitation under current ORS 109.119(3)(a) to ensure that the children have adequate contact with grandmother, after taking into account father’s constitutional rights as a biological parent.
The dissent notes the trial court’s criticism that Mazza spoke only with people who were favorable to father. As we point out, he was aware through Bolstad’s report of people who were unfavorable. Mazza’s purpose was to evaluate father and the home that father provided for the children, not to compare it with an alternative home. To some extent, the trial court’s comment reflects its assumption— which the dissent seems unable completely to shed — that its role was to compare the home that father provides with the alternative of grandmother’s home. That, however, is precisely contrary to the standard that Troxel requires us to adopt.
We do not hold, as the dissent suggests, that the time that the children lived with father necessarily ameliorates father’s previous inadequacy. See 185 Or App at 545-46 (Deits, C. J., dissenting). Rather, on the facts of this case we find the more recent period to be more predictive of father’s future actions than the more remote period.
In the final analysis, it appears that we and the dissent disagree about the weight to he given the evidence that father’s parenting was exemplary for the 10-month period of time immediately before trial. That in itself may reflect a disagreement about whether it is possible for father to change. We believe that it is possible. We acknowledge that this is a close case, but it also illustrates the importance of the constitutionally required presumption in favor of a biological parent. If the presumption is to have efficacy in a close case, it must be the presumption that becomes the determining factor. It is in accordance with that understanding that we hold in favor of father.