graphic website title banner

The Use and Abuse of Research

by Dr. Merton Strommen

Date Unknown, 2004

The purpose of this paper is to discuss how research, an important ally in securing information, is currently being used and sometimes abused as it relates to the subject of homosexuality.

My entry into research related to homosexuality did not begin until five years ago. Hence, I am still in the process of becoming oriented to this highly complex field of research. While becoming familiar with the research in this contrasting field of investigation, I have become conscious of the extent to which certain convictions about what research has found have filtered into the thinking of leaders in our church.

For instance, I noticed in a book written by a Biblical scholar (whose work I have admired) a statement that if the apostle Paul had known the new evidence that is now available on the phenomenon of homosexuality, he would not have written what he did. Curious, I wrote to him asking about this new evidence. His answering letter admitted that he could not tell me what that might be. Later, in a synod conference I heard a leader in our church use evidence from a study to show that reorientation therapy is not successful but only harmful. I wrote asking for a citation on the study. Because I was not aware that any such study had been carried out. He answered that it really was not a study but an informal survey written up as part of an article. Yet he had presented the information as though it was authoritative. And the pastors assumed the findings were the result of a scientific study.

I have no objection to information based on personal observation. Most of what is believed about homosexuality seems to be based on anecdotes or personal knowledge of a gay or lesbian. But, it should be noted, anecdotes represent the least reliable form of information. For every anecdote it is possible to find another that gives opposite information.

The purpose of a research study is to increase the reliability of the information gained on a certain subject. But even the research studies reported in a professional journal vary in terms of the reliability of their information. Given below is an informal listing that shows how research reports on reorientation therapy reported in journals during the years between 1950-1990 vary from the least reliable to the most reliable. They form a continuum.

  1. Therapist Report — (Conclusions based solely on the therapist’s judgment)
  2. Clients Report — (Various levels of dependability in what is reported)
  3. Follow-up Studies — (Interviews years after change has taken place)
  4. Follow-up Using questionnaires — (Anonymous self-reports following change)
  5. Self-Report with an Objective Assessment — (Scientific measurement)
  6. Experimental Assessment Using Control Groups — (Fully scientific study)

The medical sciences commonly grade the strength of evidence gained from various studies: Grade I-good evidence; Grade II-fair evidence; Grade III-limited evidence; Grade IV-personal opinion. The stronger the evidence the greater the support one has for the hypothesis being tested.

Careful attention to the reliability of research information is always an important consideration –especially when the issue under study is as important as the one now facing the Evangelical Lutheran Church. The issue is not simply one of deciding whether or not to ordain practicing homo-sexuals and bless same-sex marriages. Underlying that decision is the issue of deciding whether or not to consider homosexuality the equivalent of heterosexuality. The two issues are inexorably linked. It is reasonable to say that if the ELCA Assembly in 2005 should decide to ordain practicing homosexuals and bless same-sex marriages, it will be seen as equating homosexuality with heterosexuality. It will be interpreted in the minds of people as a decision of the church that homosexuality is an orientation also created by God with no inherent dangers. It will be assumed that the orientation should therefore be viewed as normal and lauded as good. But are the two sexual orientations equivalents?

The vast array of people in the fields of mental health say they are. This position first became public in 1973 when a group of gay activists stormed the convention business session of the American Psychiatric Association. Their actions proved successful in removing homosexuality from the classification of being a mental disorder. Their primary claim that there is no difference between the orientations centered in a study by psychologist Hooker.

Her study was based on a sample she recruited of 36 highly functional and socially integrated homosexual men whom she matched with 36 heterosexuals who served as a control group. The results of a Rorschach test that had been administered to all 72 participants were then blindly rated for psychodynamic features thought to characterize homosexual men. Because the judges were unable, on the basis of Rorschach responses, to predict sexual orientation of these men, it was assumed the two groups should be viewed as the same. Though this research was convincing to many, it only demonstrated that one can find gays whose scores on measures of psychological adjustment will equal those of heterosexuals. When a sample such as this one is not a random sample of the population about whom generalizations are to be made, one cannot conclude that there are no differences between the two populations.

Since 1973, the issue of equivalency has been a contentious one. Current articles and books denounce heterosexism because it values heterosexual marriage and devalues homosexual behavior. Sustained efforts are being made in courts, legislatures, schools, and denominations to establish the same rights and privileges for gays, lesbians and transsexuals as are given heterosexuals. A primary resource that is being used are the findings of research studies. They are being used to show there are no differences between gays and heterosexuals and that, therefore, equal treatment should be accorded gays. Here we become aware of the use and abuse of research. Many of the studies being quoted are of poor quality and many of the conclusions based on these studies represent politicized interpretations.

An illustration of low quality studies and politicized interpretations can be seen in the research carried out in a relatively new field - studies of children raised in same-sex families. The question being addressed is this: Are there differences in the adjustment of children raised by lesbigay children to those raised by heterosexual parents?

I. Problem One: Quality Limitations in Studies Used

Most of the studies being used to buttress a position many are advocating fail to meet the stringent requirements of a full scale scientific study. For instance, when psychologists Robert Lerner and Althea Nagai, professionals in the field of quantitative analysis made a critical evaluation of 49 empirical studies on same-sex parenting, they concluded that no generalizations can be made reliably that are based on any of these studies. In their book, NO BASIS: What the studies DON’T tell us about same-sex parenting” they insist that none provide a basis for good science or good policy. They then proceed to tell why this is true by identifying problems they discovered in all 49 studies. Every one of the studies, they report, can be faulted on one of the following requirements for a scientific study

Unmet Scientific Requirements

  • Unclear hypotheses and research designs
  • Missing or inadequate comparisons groups
  • Self-constructed, unreliable, or invalid measurements
  • Non-random samples, including participants who recruit other participants.
  • Samples too small to yield meaningful results.
  • Missing or inadequate statistical analysis.

An example of what they fault in most of the studies relates to the use of small samples. All but two of the 49 studies base their conclusions on samples of less than 100 cases. The difficulty this poses centers in the fact that the smaller the sample the less statistical power there is for the test for significance. Statistical tests such as a T test or ANOVA usually need sample sizes of at least 300 for an adequate test of significance .A Chi-Square or Pearson Correlation Coefficient usually need even larger samples. When the sample is small (less than 100), one increases the probability of a Type II error namely, the error of concluding there are no differences between the samples when in fact there may be.

To illustrate, one of the important studies they examined, one by Tasker and Golombok, involved a sample of 46 British young adult children of lesbian and heterosexual mothers. These children were interviewed 14 years after the first assessments had been made of the original sample of 73 children (average age of 10). They used a T-test to compute the significance of differences between the children in this sample of 46. The statistical power of this test (as computed by Lerner and Nagai) was actually .10 with a Type II error probability of .90, this means that there is a 90% probability that the authors had failed to reject the null hypothesis when in effect there might be, significant differences between the two groups. This type of error becomes significant when the study reports that no statistical differences were found between the children of lesbian and heterosexual mothers. One is lead to conclude that the adjustment of children in the two groups is the same.

It is important to note that studies such as the one identified above are being used by courts to make important legal decisions and by medical and mental health organizations to make policy decisions which affect the lives of people. In 1993, a New Jersey court made a ruling based on a finding of one of the studies that “children raised by homosexual parents will turn out normal.” A similar use of one or more of these studies was made by the. New Hampshire Supreme court and the Hawaii Supreme court. Ignored in their decision is the truism that generalizations can only be applied to a given population when based on samples randomly selected from that population.

Studies such as those examined above have also been used to establish the point of view reflected by mental health organizations in the amicus briefs they filed in courts cases. For instance, the American Psychological Association joined the National Association of Social Workers in a amici curiae brief in Bottoms v. Bottoms, November 15, 1993. In this brief which presented their professional judgment, they make the following assertions: ”The belief that a child raised in a household with a lesbian or gay parent is more apt to become lesbian or gay is without any bases in fact”; “the research suggests that lesbian mothers have parenting skills that are equivalent to or better than those of heterosexual parents”. In other words, these mental health professionals were insisting that science has established that children with two parents of the same gender are as well adjusted as children with one of each kind.

This position is also being asserted as fact by social workers, lawyers, policymakers, and the media. But are the statements given in the court case accurate? They might be but not because scientific studies have established them as fact.

Clearly, studies are being used which lack the quality needed for making broad generalizations and providing the authoritative basis for policy and legal decisions. Though they are scientific in part, they are not full scale studies. All that can be said for most of them is this: they are useful in opening up knowledge in a frontier area and fostering a degree of understanding about a new and difficult subject. Though they lack the characteristics of a scientific study they still quality as exploratory research. Their findings can be used in hypothesis building, in identifying new directions for research, and in helping to formulate large well-designed studies. But they are not adequate to support the kinds of statements that were made in the court case identified above.

II. Problem Two: Politicized Interpretations

A second major obstacle to accurate generalizations, centers in the politicization of authors who carry out the studies. Theirs is a tendency to report what they want the study to establish, namely, that there are no differences between children raised by same-sex parents and those raised by a father and mother. Lerner and Nagai who reviewed 49 studies published in professional journals or books made this observation: “With one exception, the authors of these studies wish to influence public policy to support same-sex marriage and the adoption of children by homosexual couples”.(p.7)

A similarly provocative article was published by two sociologists at the University of California, Judith Stacey and Timothy J. Biblarz, under the title, “(How) Does the Sexual Orientation of Parents Matter?” Published in the American Sociological Review, April, 2001, it reflects the same point of view. The authors, who affirm a pro-gay stance, nevertheless observe that researchers frequently downplay findings that indicate differences regarding children’s gender and sexual preferences and behavior as a result. “Virtually all of the published research claims to find no differences in the sexuality of children reared by lesbigay parents and those raised by nongay parents” (p.163)

The basis for the comments of Stacey and Biblarz center in a critical review of 21 of the most scientific studies of which 18 were of advantaged lesbian parents and 3 of gay male parents, studies using matched samples of heterosexual parents. The authors observed that none of the studies they examined were based on random samples of families, most relied on small-scale, snowball and convenience samples drawn primarily from personal and community networks of agencies. Their scrutiny of the findings suggests that the sexual orientation of these parents matters somewhat more for their children than the researchers have claimed.

They illustrate their observation by providing a close look at two of the studies. One is the study by Tasker and Golombok that followed 27 heterosexual single mothers and 39 of their children (average age of 10) as well as 27 lesbians mothers and 39 of their children (also average age of 10). Follow-up interviews with 46 of the original children were conducted 14 years later. What was found?

A significantly greater proportion of young adult children raised by lesbian mothers (than those raised by heterosexual mothers) had a homoerotic relationship. The difference was 24 percent compared to 0 percent. Also, the young adult children reared by lesbian mothers were significantly more likely to report having thought they might experience homoerotic attraction or relationship. Here the difference is: 64 percent compared with 17 percent. (p.170)

Additional findings reported in the article have to do with the number of sexual partners children report having had between puberty and young adulthood. Relative to their counterparts with heterosexual parents, the adolescent and young adult girls raised by lesbian mothers appear ( in the words of the authors) “to have been more sexually adventurous and less chaste”. (p.171)

In spite of differences such as these, the authors Tasker and Golombok give the following conclusion in their report. “The commonly held assumption that children brought up by lesbian mothers will themselves grow up to be lesbian or gay is not supported by the findings of the study and, there was no statistical difference between the young adults from lesbian and homosexual family backgrounds with respect to sexual orientation.” (p.171) To arrive at this conclusion they used a Pearson test of significance on their sample of 46 young adults. According to Lerner and Nagai there was a 90% probability of having made a Type II error in their conclusion, namely, a 90% probability that significant differences might actually exist.

The sociologists Stacey and Biblarz illustrate their point with another study, this one by R. Green >et al In examining this report they find at least 15 statistically significant differences in gender behavior and preferences among the children (4 among boys and 11 among girls) raised in lesbian and heterosexual single-mother homes. Yet the study’s abstract summarizes the study in this way. “Two types of single-parent households (lesbian and heterosexual single-mothers) and their effects on children (3-11 years) were compared… No significant differences were found between the two types of households for boys and girls and few significant differences for girls” (p.170)

The Search for Objective Truth

Sociologists Stacey and Biblarz who differ sharply with statistical analysts Lerner and Nagai on the merits and morals of lesbigay parenthood, do agree that contemporary scholarship on the effects of parental sexual orientation on children’s development is rarely critical of lesbigay parenthood. They admit that this reticence compromises the development of knowledge not only in child development and psychology but also within the sociology of sexuality, gender, and family.

They candidly acknowledged the political dangers they may experience in pointing out that recent studies show that a higher proportion of children with lesbigay parents are themselves apt to engage in homosexual activity. However, they add,” it is neither intellectually honest nor politically wise to base a claim on grounds that may prove falsifiable empirically.”(p.178) They conclude their article by recognizing that “they have challenged the predominant claim that the sexual orientation of parents does not matter at all. Despite limitations of the studies, there is suggestive evidence and good reason, they say, to believe that contemporary children and young adults with lesbian or gay parents do differ in modest and interesting ways from children with heterosexual parents” (p.177) It should be added that even though children of lesbigay parents appear to express a significant increase in homoeroticism, the majority of all children nevertheless identify as heterosexual.

Why do the authors, Stacey and Biblarz venture such a position in a publication as prestigious as the American Sociological Review? It is because they as theorists have noticed that a diverse array of gender theories (social learning theory, materialist, symbolic interaction) point in this direction. There is reason to believe, they say, that parental orientation is positively associated with the probability that children will more likely attain a similar orientation, and they add, theory and common sense support such a view. Children raised in religious homes tend to be religious and children raised in homes where parents smoke are more likely to smoke. The authors contend that children raised by lesbian co-parents should and do seem to grow up more open to homoerotic relationships.(p.178)

Support for the point of view of these two authors comes also from the only study of children of lesbigay parents that is based on a large sample (5,182 adults) drawn randomly from six major U.S. cities. A comparison of lesbigay children in this study with those raised by heterosexual parents gives support to the position of Stacey and Biblarz. The comparisons made by the authors, Paul Cameron and Kirk Cameron, show that when children of same-sex parents are compared to those raised by heterosexual parents significant differences occur on each of the following dimensions: homoerotic inclination, victimization, and social-psychological disturbance, The authors summarize their findings by saying: “Our results suggest that the sexual preference or orientation of the parent influences the preference of the child and that homosexual parents are associated disproportionately with homosexual children” (p.9)


The illustration that I have given of research in one specific area of research relating to homosexuality, underscores the need for careful research before making revolutionary generalizations. Systematic efforts ought to be made by the church to seek out quality research that provides the objective information that is needed when deciding as a church whether or not the two orientations are equivalents. Such research findings can be supplemented by what the church might originate.

Here I make reference to the fact that it is possible for people in the church to design a national study that addresses important questions regarding the issue of homosexuality. The study can be designed so that persons on both sides of the argument are satisfied with its objectivity. Both sides can agree on: the statement of hypothesis, the sampling frame, the instruments to be used, the survey procedures, the methods of analyses, and the persons to be involved in the interpretation of the data. In other words an entirely objective study can be designed and carried out which will yield data that either confirms or rejects the various hypotheses being tested. Or, its findings may turn out to be inconclusive.

We as a church have the people with the needed research and analytic skills—we have demonstrated this through major studies carried out by Search Institute for the government. We have sources that could provide the funds for carrying out studies authorized by the church. A major study could take a positive approach to an issue that is short on proposed solutions. The results, presented in the context of a full review of extant research, would provide an excellent basis for discussion and greater understanding.