01Jun05b Re-analyzing Rind (Desire)
In the famous meta-analysis of Rind et al. [RTB98], one of the aspects that
is studied are the recalled immediate reactions to child sexual abuse
experiences - were they (at the time) seen as positive, negative or neutral
by the younger partner?
The results from a number of studies is given, and
a weighted average is found of 11% positive, 18% neutral, 72% negative for
girls and 37% positive, 29% neutral, 33% negative for boys.
However, the data from the various studies differ widely - positive reactions
2% to 28% for girls and 8% to 69% for boys, negative from 52% to 84% for
girls and from 7% to 54% for boys.
I have looked at the various definitions used for CSA, to see whether this
might explain the large differences between studies for these data. The
definitions differed in various ways:
The maximum age of the child
The minimum age difference
Whether all contacts or only unwanted contacts were counted
Whether all sex or only contact sex was counted
How would these factors influence the percentages of positive, neutral and
First, counting only unwanted contacts would clearly increase the percentage
of negative and decrease the percentage of positive contacts by a significant
amount. In my opinion, studies that a priori distinguish between wanted and
unwanted sexual contacts are unfit to be used in this part of the analysis.
Two studies are invalidated this way: The one of Finkelhor [Fin79] for girls,
and the one of Fishman [Fis91a] for boys. Luckily, removing these studies
does not have a major effect on the numbers that Rind c.s. got.
The percentages for girls remain the same (11% positive, 18% neutral, 72%
negative), for the boys both positive and negative effects go up 1% (38%
positive, 27% neutral, 34% negative). Of course the number of cases on which
the numbers are based now also changes, and becomes 1302 for girls and 576
The effects of the other factors are less certain. My own ideas are that:
The younger the age of the child, the more often experiences will be
negative, and less often positive. The younger the child, the less likely
(s)he is to like adult sexual stimuli. As such, there will be less
positive sexual contacts, or at least positive contacts that are later
regarded sexual. Also, younger children might have forgotten about certain
experiences. This seems to be most likely with neutral experiences, and
least likely at strongly negative ones. As such, sexual experiences would
be even extra tilted towards negative and especially away from neutral when
young children are involved.
Experiences with peers are less likely to be unwanted, and when unwanted
less likely to be based on extreme methods of force, than those with adults.
However, it is not clear how 'older youths' (between 5 and 10 years older)
would fit in. They might well be even more likely to go on against the will
of the child. Thus, the effect of changing age difference limits is unclear.
Finally, the difference between studies where only contact sex would be
included and studies that also include non-contact experiences (voyeurism
and exhibitionism) is also not completely clear. There will be children who
are not interested in contact sex acts, but would like these acts, which
might seem to make it likely that contact sex would be seen more negative.
However, it also seems likely that adults or older youths who want to be
sexual with a child will start with non-contact sexuality, and then go on
only when the child reacts positively. As such, many unwanted sexual
experiences might end in the non-contact stage, while wanted sexual
experiences evolve into contact sex. Finally, non-contact sexuality would
seem more likely to either not be remembered by the child or not be
labeled as 'sexual' by them. This might cause a lower number of neutral
cases. However, the latter effect, even if present, would undoubtedly be
lost in the roughness of the data.
When we look at the various studies Rind c.s. study, dropping [Fin79] and
[Fis91a], we find that the various factors mentioned above are highly
confounded. We see two groups of studies with similar definitions, and a few
that do not fit in with any.
First, there are the studies that include only contact sex. These are two
by Brubaker [Bru91,Bru94] on girls, which both use a maximum age of 16 and
a maximum age difference of 5 years, as well as two different studies on
boys, those by Condy et al. [CTBV87], with a maximum age of 16 and a maximum
age difference of 5 years, but also including cases where the older partner
was less than 5 years older provided (s)he was over 16 him/herself, and
that of Urquiza [Urq89], using a maximum age of 18 with a maximum age
difference of 5 years.
The second large group consists of 5 studies all using a maximum age of
16 years, including both contact and non-contact sex, and setting a minimum
age difference of 5 years when the child is under 13, 10 years otherwise.
Included here are [FB89], [Fro84], [GG88], [LJ93] and [ONe91]. In total,
these are 4 studies for women and 3 for men.
Three studies remain: The large one of Landis [Lan56], using the vague age
limit of 'through adolescence' and no specified minimum age difference (only
that it was an 'older partner'), the one of Schultze and Jones [SJ83],
having a maximum age of 12, while using a minimum age of 16 of the older
partner to specify the age difference, and the one of West and Woodhouse
[WW93], looking at boys only, using a maximum age of 16 and a miniimum age
of the older partner of 16 when the boy was under 11, 18 otherwise. All
three studies include both contact and non-contact sex.
I will give the percentages when looking at only the studies from one group.
(percentages do not always add up to 100 because of rounding)
Comparing the two results in this table we find that:
1. The difference between the groups are not very large
2. The results are more positive than those from [RTB98], especially for boys.
The cause of the second is that the Rind study is quite heavily influenced
by the study of Landis [Lan56] (35% of the female and 30% of the male data),
which has much more negative numbers than the other studies. If we remove
that study from the full data set, we get results comparable to the ones
above (women 15-18-66, men 50-25-24). This should not surprise us, since
after removal of Landis, group 1 and 2 together account for 74% of the female
and 63% of the male data.
Given that finding differences between these two groups is rather useless,
we will instead look at the various single studies. Regarding age maximum,
most studies use an age maximum of 16. Exceptions are [Lan56], [SJ83] and
[Urq89]. Of these, Landis does not use a clear age maximum at all, so it
should be left out of the analysis. From the above, one would expect that
[SJ83], having an age maximum of 12, would be more negative than average.
[Urq89], having an age maximum of 18, might be more positive, but this is
Strangely, the opposite seems to be the case: [SJ83] actually
finds the lowest amount of negative and the highest amount of
experiences of all studies, for either gender. [Urq89] on the other hand
is either comparable to, or more negative than, the other male studies.
As far as these data are concerned, it seems clear that the hypothesis that
younger children are more likely to consider sexual experiences with
significantly older persons as negative, is to be rejected.
Remains for us to discuss the [Lan56] study. The definition of CSA used, as
specified in [RTB98], is "any sex (including non-contact or wanted) through
adolescence with an older partner". As such, I think the positive numbers
much too low to be believable. In any contact where a youth is involved at
least one of them will have an older partner, and as such neither an
occurrence of 'just' 30-35%, nor a positivity of around 5% seems credible.
Rather, someone has misunderstood a definition - whether it is me
misunderstanding Rind c.s., them misunderstanding Landis or Landee's
interviewees misunderstanding him I do not know. I would also like to notice
that the fact that the Landis study is from the 1950s and the others are
from the 1980s or 1990s may be a factor - there might be a change in sexual
climate or an improvement in research methodology that (partly) explains why
Landis's results are so much more negative than the others.
Concluding we can say that the differences between the various studies in
table 7 of [RTB98] cannot be explained by their difference in definition of
CSA. In particular, the numbers seem to contradict the hypothesis that
younger children are more likely to experience sexual experiences with
significantly older people as negative. Rather, the differences seem to be
either caused by statistical variation, or by differences in the way the
studies were conducted other than the definition of CSA.
The data from [Lan56] considerably influence the averages got by Rind c.s.,
and is the
most negative of them all. One might argue that it could be a wrong data
point, invalidated because of changes in the sexual climate since the 1950s
and/or through improvements in research methodology. The results of [Fin79]
and [Fis91] certainly should not be included in this table, because these
studies include some unwanted experiences that would not have been included
had they not been unwanted, thus biasing the result.
|Group 1 only
|Group 2 only