** Please visit our website: http://www.africanassociation.org **

Salifou,

Thanks for the update.  After reading Earl's e-mail,
I'm still amazed that anyone could think that abortion
could significantly contribute to reduction of the
crime rate.  As Earl indicated, the study supporting
this belief is "dubious."  What I know about
regression analysis indicates that the r-square value
has to be pretty high in order to state with any
certainty that there's a strong correlation between
two phenomena or parameters.  Otherwise, it is
irresponsible with a weak correlation (as it seems in
this case) to advance the thesis that abortion has a
positive effect on reduction of crime rates.

On the "political" side of things, I'm unlike Earl, in
that I'm more to the far right of the abortion debate.
 I shudder to think that people would, or are willing
to, give license to "women" to kill their babies at
"any time during pregnancy."  The only time I would
think it appropriate to allow abortion is when a
medical decision must be made between saving the life
of the mother or the child.  Any other, with the
possible exception of a pregnancy due to rape, is not
justifiable.  (Note:  In the case of rape, I would
agonizingly leave it to the woman and her God, if she
believes in God.)

Take care,

- Wilmot


--- salifou issoufou <[log in to unmask]> wrote:

> Wilmot and all,
> below is the responses I got from my classmate
> (there are two responses and they are quoted: one
> explaining what his study was about and the other
> explaining his political views):
>
> "Salifou,
>
> Sorry, I just got this message... I don't really
> have internet access at
> home.
>
> Anyway, I assume this controversy was raised by
> Bennett's comments?  I'd
> note that he was attempting a reductio ad absurdum
> argument, in the context
> of saying that abortion is wrong and that (abortion
> foes) shouldn't argue
> over its drawbacks because that opens the door for
> pro-choice folks to argue
> over its benefits.  Taking his argument to the
> logical extreme, aborting
> everyone in the US for a year would probably slash
> crime rates...
>
> In any case: crime, in America, has a strong racial
> component.  It's fair to
> say that, viewed on a per capita WRT race basis,
> crime is strongly black.
> For data, see the Uniform Crime Reports published
> annually by the FBI or the
> enhanced Supplemental Homicide Reports.
>
> Further, it is well documented in the literature
> that propensity for
> criminality is strongly age correlated.  Violent
> crime -- murder/assault,
> home invasion, burglary, rape, etc... -- depending
> on which crime we are
> discussing exactly -- starts to peak at roughly 16
> and falls off after 21.
>
> The final point that is well documented in the
> literature is children do
> better on a wide range of indicators when the
> children are wanted and the
> older their parents are at the age they have their
> first child.  The obvious
> reasons seem to come into play -- the parents tend
> to be better educated,
> have more money, spend more time and make more human
> capital investments in
> their children, are more likely to have access to
> health care, etc.
>
> So, the idea, at least as put forth in Donohue and
> Levitt, is this:
> Abortion is viewed as a birth control of last
> resort.  That is to say, women
> use birth control to prevent "marginal" children.
> These marginal children,
> in turn, had they been born are unwanted and hence
> are likely to receive
> fewer investments of human capital from their
> parents; this makes them more
> likely to become criminals.
>
> There are two competing speculated causative
> methods:
> 1 - the Easterlin Hypothesis (which is a subject of
> controversy among social
> scientists) -- ie that decreases in birth cohort
> size lead to improvements
> in the cohort
> 2 - a further hypothesis by D&L that decreases in
> birth cohort size via
> abortion will lead to improvements in the birth
> cohort above and beyond
> those postulated by Easterlin because it would be
> precisely the children
> most likely to commit crime that wouldn't be born
>
> Then one can look at abortion rates which starting
> in 1973 peaked seven
> years later and at crime rates which similarly
> peaked in the early 90s and
> have since then experienced a steady decline which
> many criminologists have
> difficulty explaining.  Given the time difference
> between these two events
> it is not unreasonable to examine them further.
>
> The problem, of course, is that this is entirely
> observational data, making
> it difficult to postulate causation.  D&L don't even
> attempt to show that
> abortion is a Granger Cause -- one of the methods
> which could be used to
> attempt to show causation with non-random data.  In
> any case, they show,
> with some large problems, that lagged abortions and
> crime are correlated.
>
> D&L conclude that abortion did cause a large portion
> of the decline in crime
> but there are numerous methodological and logical
> problems with their paper.
> They have been challenged by Joyce; in fact, two
> friends and I worked on a
> paper this summer further challenging D&L.
>
> Joyce, in his first reply, does something really
> smart -- using the 5 states
> that legalized abortion 3 years before the rest of
> the country as a natural
> experiment.  If abortion truly caused the decline in
> crime, then it should
> have caused the decline in crime 3 years earlier in
> the early legalization
> states.  He concludes that it doesn't.
>
> Some of the potential problems with D&L's method
> are:
> 1 - D&L assume there were 0 abortions pre-73.  This
> is demonstrably wrong.
> The fact that many legal abortions replaced an
> unknown number of illegal
> abortions is a problem they skirt.
> 2 - crack cocaine became a huge issue during the 80s
> and early 90s and will
> massively distort the crime data
> 3 - abortions are often used by wealthier women as a
> method of controlling
> family size.  Even though these children may have
> been undesirable, they
> seem likely to have, on average, not been
> disadvantaged had they been born.
> 4 - time series data has demographic problems /
> implications that D&L
> ignored, specifically race & cohort issues
>
> Rereading your email, you asked for the method.  You
> would use regression
> analysis along with the above data sources.  Of
> course, there are issues
> with time series data that have to be taken into
> account, particularly how
> time series data lines up with the demographics.  An
> alternative approach,
> used by Joyce and us, is a difference-in-differences
> approach.  We conclude,
> agreeing with Joyce, that the link between abortion
> and crime is extremely
> dubious.  There may be one but nowhere near the
> magnitude postulated by D&L.
> In fact, it's worth pointing out that other papers
> have found a positive
> relationship -- I can't remember the name off the
> top of my head but I have
> it in a binder at home if you want me to find it.
>
> In any case, I hope this helps.
>
> earl"
>
>
> "Salifou,
>
> I just replied to your earlier email; sorry for the
> delay.  If you would
> like a simple explanation of regression analysis to
> share, I think I have
> one somewhere.
>
> Oh, and if this is a political debate and my
> politics matter, I'm on the far
> left of the pro-choice crowd: I support abortion at
> any time during
> pregnancy for any reason and think that government
> would do well to offer
> abortions for free to any woman who wants one.
>
> earl"
>
>
>
> Have a nice weekend
>
> Salifou
>
> "Wilmot B. Valhmu" <[log in to unmask]> wrote:
> ** Please visit our website:
> http://www.africanassociation.org **
>
> Vera,
>
> What scares me is that there are actually people out
> there who believe that abortion can cure crime
> problems.
>
> But then again, I'm not so surprised: in a climate
> of
>
=== message truncated ===

----------------------------------------------------------------------------
To unsubscribe/subscribe or view archives of postings, visit:

        http://maelstrom.stjohns.edu/archives/aam.html

AAM Website:  http://www.africanassociation.org
----------------------------------------------------------------------------