Showing posts with label flawed study. Show all posts
Showing posts with label flawed study. Show all posts

Thursday, November 11, 2010

No Matter How You Slice It, It is Still Baloney: Reflections on the Flawed Lesbian Family Study

Source: http://www.law.ucla.edu/williamsinstitute/pdf/Gartrell-Bos-Goldberg-2010.pdf

Introduction

On Facebook someone passed another link from the Huffington Post  claiming there is 0% instances of physical or sexual abuse among the children of lesbian couples.  The conclusion being drawn is that there are no dangers to children for being the child of a homosexual couple.  Suspecting that this claim was highly dubious (similar to the 100% support claimed by dictators), I looked up the actual report (linked above).

This report claims (page 6) that:

A key finding in the current study was that none of the NLLFS adolescents reported physical or sexual abuse by a parent or other caregiver. This finding contradicts the notion, offered in opposition to parenting by gay and lesbian people, that same-sex parents are likely to abuse their offspring sexually. (Arkansas Department of Human Services, 2010;Falk, 1989; Ford, 2010; Golombok & Tasker, 1994; Patterson, 1992)

They contrast this to other studies saying:

In the 17-year-old weighted subsample of the U.S. Office of Juvenile Justice and Delinquency Prevention’s National Survey of Children’s Exposure to Violence (NATSCEV) (Finkelhor et al., 2009a, b, c), it was found that the lifetime rates of victimization by a parent or other caregiver were: 26.1% of adolescents had been physically abused and 8.3% sexually assaulted (D. Finkelhor, personal communication, March 13, 2010)

The inference drawn by supporters is that a child living with a lesbian couple is safer than those with heterosexual parents.  Therefore there is no basis for those people to oppose homosexuals raising children on grounds of abuse.

The Logical Fallacies

There are several logical fallacies with such a conclusion, and I will focus on three: Weak Analogy, Hasty Generalization, and Begging the Question.

Weak Analogy

In an analogy or comparison, we have two different  situations which are compared.  If we say that conditions [A], [B] and [C] are the same in both Situation 1 and 2, we might think that the conditions are acceptable to compare and contrast.  However, if relevant conditions [X], [Y] and [Z] are different, this may show the comparison is not accurate.

In this case, we have a false comparison comparing the abuse suffered by juvenile delinquents compared to the adolescents in this survey.  This is a comparison of apples and oranges.  We are comparing one sample of delinquents to one sample of lesbian couples with children.

A proper comparison would be to compare what percentage of those juvenile delinquents abused came from heterosexual homes compared to heterosexual families as a whole, and then compare this to those juvenile delinquents who were abused and came from homosexual homes to homosexual families as a whole.

Therefore, this study does not have equal conditions for comparison.

There is another problem with their comparison.  It is relevant is that all the couples made use of artificial insemination.  However, this report (see page 2) is being used to discuss homosexual couple adoptions and the objections to them:

Another claim about the origins of sexual orientation that has been put forth in litigation and public discourse by opponents of equality in marriage, adoption, and foster care for same-sex couples is that lesbian and gay parents are more likely to abuse their children sexually.

Quite simply, there is a difference between Artificial Insemination to have a child and to adopt a child.  You can't use the experience of one [artificial insemination] to argue in favor of the other [same sex couple adoption] without studying the experience of the other.

In short, the relevant issues which are different make any general conclusions drawn by this conclusion a weak analogy.

The Hasty Generalization

There is another problem with such a study .  The youth interviewed were in fact a total of 78 youths approximately 17 years of age (half male youth, half female youth) from 77 families which completed all stages of this study.  78 youth are in fact far too small a number to gauge the proper state of affairs of lesbian couples with children.  This is especially notable when the report describes the statistical breakdown of the geography of the couples:

Family region of residence (U.S.)
Northeast 47%
Midwest 1%
South 9%
West 43%

We should point out that 1% of 77 couples would be 0.77 families and 9% of the 77 couples would be 6.93 families.  In other words, in the entire Midwest, we have 1 family represented.  Sure we can assume this group rounded off, but one family in the entire Midwest?  That's asinine to assume they represent the entire lesbian population of the Midwest.

The footnote describes this situation as:

Between T3 and T5, the NLLFS families resided in large urban communities, midsized towns, and rural areas of California, Georgia, Louisiana, Massachusetts, Maryland, Minnesota, New York, Oregon, South Carolina, Texas, Vermont, Virginia, and Wisconsin

That's a pretty biased sample.  90% of the population is drawn from two regions.  Only 10% of the population from the South and Midwest (Georgia, Louisiana, Texas, Wisconsin and Maryland).

Moreover, to make a blanket assumption (0% of lesbian "families" abuse their children) requires how such couples behave in other regions of the world.  Essentially this is a report on how 77 lesbian families (of which 56% broke up within 7 years of the birth of the child [see page 3]) in the US, with a majority in the West and the Northeast) treated their children.

First, we need to know what number of lesbian couples exist in the United States, especially in each region, then what percentage of them these children make up.  Since this study in fact tells us that this study was made up of 77 families with 78 adolescent children (see page 3 of this report).  Since we don't know from this report the number of lesbian couples with children, we cannot say that such a sample is representative of the whole.

Indeed, on page 8 of this report, the Gartrell-Bos-Goldberg study admits the flaw with the NLLFS study:

Despite these strengths, the NLLFS has several limitations.  First, it is a nonrandom sample. At the time that the NLLFS began in the mid-1980s, due to the long history of discrimination against lesbian and gay people, the prospect of recruiting a representative sample of planned lesbian families was even more remote than it is today (Bos et al., 2007). A second limitation is that the NLLFS and NSFG were neither matched nor controlled for socioeconomic status, race/ethnicity, or region of residence. An analysis of a more economically diverse sample would be an important contribution given that same-sex couples raising children are more likely to live in poverty and have lower household incomes than married, heterosexual couples raising children (Albelda, Badgett, Schneebaum,& Gates, 2009; Julien, Jouvin, Jodoin, l’Archeveque, & Chartrand, 2008). In addition, now that it is possible to obtain more information about sperm donors, future studies might benefit from exploring the association between the offspring’s sexual orientation and that of both parents. Finally, although the NLLFS is the largest, longest-running prospective study of planned lesbian families, the findings would be strengthened by replication in a larger sample.

So, what is flawed is, according to Gartrell, Bos and Goldberg:

  1. The sample is non-random.
  2. The sample is not representative
  3. The socio-economic status was not controlled in sample.
  4. Race/ethnicity was not controlled in sample (adolescents were 87.1% white, 12.9% non-white).
  5. The regions represented was not controlled in sample.
  6. The sample is considered too small.
  7. The study did not account for the sexual orientation of the sperm donor.

When you consider that any or all of these can throw off the actual results of the study, it would be a hasty generalization to assume that lesbian couples do not abuse their children on the basis of this study.

The Fallacy of Begging the Question

In discussing physical abuse, this study performs the Begging the Question fallacy when it comes to what is understood by physical abuse.  This study describes (page 7) the idea of less physical abuse among lesbian couples as:

In addition, corporal punishment is less commonly used by lesbian mothers as a disciplinary
measure than by heterosexual fathers (Gartrell et al., 1999, 2000, 2005, 2006; Golombok et al., 2003)

However, corporal punishment has not been established as physical abuse intrinsically, though there is no doubt that some can turn corporal punishment into physical abuse (the "link" referred to further on in the paragraph cited).  The point is, if parents do spank their children, is this automatically considered to be physical abuse?  Most parents, it would seem, would disagree.  So this is an issue which needs to be proven, not assumed to be true.

[In addition, this is a Weak analogy again.  Comparing Lesbian mothers with Heterosexual fathers misses a crucial comparison.  What is absent is the comparison with Homosexual fathers, as well as comparing lesbian mothers to heterosexual mothers]

Thus to conclude that there is more physical abuse among families with heterosexual male parents, basing it on the claim that corporal punishment is most likely to be done by a heterosexual male parent is to beg the question: that corporal punishment is abuse.

Conclusion

Unfortunately, this report makes use of a flawed sample, a weak comparison and an assumption that corporal punishment leads to physical abuse to argue that there is no basis for fears of homosexual couples adopting children.  Since the report by the NLLFS fails to meet the requirements for a controlled study with a representative sample which correspond to the questions asked, this report by the NLLFS cannot be said to prove its point.

No Matter How You Slice It, It is Still Baloney: Reflections on the Flawed Lesbian Family Study

Source: http://www.law.ucla.edu/williamsinstitute/pdf/Gartrell-Bos-Goldberg-2010.pdf

Introduction

On Facebook someone passed another link from the Huffington Post  claiming there is 0% instances of physical or sexual abuse among the children of lesbian couples.  The conclusion being drawn is that there are no dangers to children for being the child of a homosexual couple.  Suspecting that this claim was highly dubious (similar to the 100% support claimed by dictators), I looked up the actual report (linked above).

This report claims (page 6) that:

A key finding in the current study was that none of the NLLFS adolescents reported physical or sexual abuse by a parent or other caregiver. This finding contradicts the notion, offered in opposition to parenting by gay and lesbian people, that same-sex parents are likely to abuse their offspring sexually. (Arkansas Department of Human Services, 2010;Falk, 1989; Ford, 2010; Golombok & Tasker, 1994; Patterson, 1992)

They contrast this to other studies saying:

In the 17-year-old weighted subsample of the U.S. Office of Juvenile Justice and Delinquency Prevention’s National Survey of Children’s Exposure to Violence (NATSCEV) (Finkelhor et al., 2009a, b, c), it was found that the lifetime rates of victimization by a parent or other caregiver were: 26.1% of adolescents had been physically abused and 8.3% sexually assaulted (D. Finkelhor, personal communication, March 13, 2010)

The inference drawn by supporters is that a child living with a lesbian couple is safer than those with heterosexual parents.  Therefore there is no basis for those people to oppose homosexuals raising children on grounds of abuse.

The Logical Fallacies

There are several logical fallacies with such a conclusion, and I will focus on three: Weak Analogy, Hasty Generalization, and Begging the Question.

Weak Analogy

In an analogy or comparison, we have two different  situations which are compared.  If we say that conditions [A], [B] and [C] are the same in both Situation 1 and 2, we might think that the conditions are acceptable to compare and contrast.  However, if relevant conditions [X], [Y] and [Z] are different, this may show the comparison is not accurate.

In this case, we have a false comparison comparing the abuse suffered by juvenile delinquents compared to the adolescents in this survey.  This is a comparison of apples and oranges.  We are comparing one sample of delinquents to one sample of lesbian couples with children.

A proper comparison would be to compare what percentage of those juvenile delinquents abused came from heterosexual homes compared to heterosexual families as a whole, and then compare this to those juvenile delinquents who were abused and came from homosexual homes to homosexual families as a whole.

Therefore, this study does not have equal conditions for comparison.

There is another problem with their comparison.  It is relevant is that all the couples made use of artificial insemination.  However, this report (see page 2) is being used to discuss homosexual couple adoptions and the objections to them:

Another claim about the origins of sexual orientation that has been put forth in litigation and public discourse by opponents of equality in marriage, adoption, and foster care for same-sex couples is that lesbian and gay parents are more likely to abuse their children sexually.

Quite simply, there is a difference between Artificial Insemination to have a child and to adopt a child.  You can't use the experience of one [artificial insemination] to argue in favor of the other [same sex couple adoption] without studying the experience of the other.

In short, the relevant issues which are different make any general conclusions drawn by this conclusion a weak analogy.

The Hasty Generalization

There is another problem with such a study .  The youth interviewed were in fact a total of 78 youths approximately 17 years of age (half male youth, half female youth) from 77 families which completed all stages of this study.  78 youth are in fact far too small a number to gauge the proper state of affairs of lesbian couples with children.  This is especially notable when the report describes the statistical breakdown of the geography of the couples:

Family region of residence (U.S.)
Northeast 47%
Midwest 1%
South 9%
West 43%

We should point out that 1% of 77 couples would be 0.77 families and 9% of the 77 couples would be 6.93 families.  In other words, in the entire Midwest, we have 1 family represented.  Sure we can assume this group rounded off, but one family in the entire Midwest?  That's asinine to assume they represent the entire lesbian population of the Midwest.

The footnote describes this situation as:

Between T3 and T5, the NLLFS families resided in large urban communities, midsized towns, and rural areas of California, Georgia, Louisiana, Massachusetts, Maryland, Minnesota, New York, Oregon, South Carolina, Texas, Vermont, Virginia, and Wisconsin

That's a pretty biased sample.  90% of the population is drawn from two regions.  Only 10% of the population from the South and Midwest (Georgia, Louisiana, Texas, Wisconsin and Maryland).

Moreover, to make a blanket assumption (0% of lesbian "families" abuse their children) requires how such couples behave in other regions of the world.  Essentially this is a report on how 77 lesbian families (of which 56% broke up within 7 years of the birth of the child [see page 3]) in the US, with a majority in the West and the Northeast) treated their children.

First, we need to know what number of lesbian couples exist in the United States, especially in each region, then what percentage of them these children make up.  Since this study in fact tells us that this study was made up of 77 families with 78 adolescent children (see page 3 of this report).  Since we don't know from this report the number of lesbian couples with children, we cannot say that such a sample is representative of the whole.

Indeed, on page 8 of this report, the Gartrell-Bos-Goldberg study admits the flaw with the NLLFS study:

Despite these strengths, the NLLFS has several limitations.  First, it is a nonrandom sample. At the time that the NLLFS began in the mid-1980s, due to the long history of discrimination against lesbian and gay people, the prospect of recruiting a representative sample of planned lesbian families was even more remote than it is today (Bos et al., 2007). A second limitation is that the NLLFS and NSFG were neither matched nor controlled for socioeconomic status, race/ethnicity, or region of residence. An analysis of a more economically diverse sample would be an important contribution given that same-sex couples raising children are more likely to live in poverty and have lower household incomes than married, heterosexual couples raising children (Albelda, Badgett, Schneebaum,& Gates, 2009; Julien, Jouvin, Jodoin, l’Archeveque, & Chartrand, 2008). In addition, now that it is possible to obtain more information about sperm donors, future studies might benefit from exploring the association between the offspring’s sexual orientation and that of both parents. Finally, although the NLLFS is the largest, longest-running prospective study of planned lesbian families, the findings would be strengthened by replication in a larger sample.

So, what is flawed is, according to Gartrell, Bos and Goldberg:

  1. The sample is non-random.
  2. The sample is not representative
  3. The socio-economic status was not controlled in sample.
  4. Race/ethnicity was not controlled in sample (adolescents were 87.1% white, 12.9% non-white).
  5. The regions represented was not controlled in sample.
  6. The sample is considered too small.
  7. The study did not account for the sexual orientation of the sperm donor.

When you consider that any or all of these can throw off the actual results of the study, it would be a hasty generalization to assume that lesbian couples do not abuse their children on the basis of this study.

The Fallacy of Begging the Question

In discussing physical abuse, this study performs the Begging the Question fallacy when it comes to what is understood by physical abuse.  This study describes (page 7) the idea of less physical abuse among lesbian couples as:

In addition, corporal punishment is less commonly used by lesbian mothers as a disciplinary
measure than by heterosexual fathers (Gartrell et al., 1999, 2000, 2005, 2006; Golombok et al., 2003)

However, corporal punishment has not been established as physical abuse intrinsically, though there is no doubt that some can turn corporal punishment into physical abuse (the "link" referred to further on in the paragraph cited).  The point is, if parents do spank their children, is this automatically considered to be physical abuse?  Most parents, it would seem, would disagree.  So this is an issue which needs to be proven, not assumed to be true.

[In addition, this is a Weak analogy again.  Comparing Lesbian mothers with Heterosexual fathers misses a crucial comparison.  What is absent is the comparison with Homosexual fathers, as well as comparing lesbian mothers to heterosexual mothers]

Thus to conclude that there is more physical abuse among families with heterosexual male parents, basing it on the claim that corporal punishment is most likely to be done by a heterosexual male parent is to beg the question: that corporal punishment is abuse.

Conclusion

Unfortunately, this report makes use of a flawed sample, a weak comparison and an assumption that corporal punishment leads to physical abuse to argue that there is no basis for fears of homosexual couples adopting children.  Since the report by the NLLFS fails to meet the requirements for a controlled study with a representative sample which correspond to the questions asked, this report by the NLLFS cannot be said to prove its point.