The phrase “Correlation does not imply causation” has developed in to a Falsehood, as I discuss here. This is in part because people often use the phrase to argue that a particular correlation has no meaning, which is a false argument. It is, of course, true that a correlation does not in and of itself prove a causal link between two things. And, as pointed out in a few places, but I’ll refer you to this Mother Jones piece for background, the relationship between single mothers and homicide and other crime is … well … interesting.

The idea is to blame single mothers for crime. They, being single mothers, would do a lousy job of raising their offspring, who would then grow up and be criminals. There is a racist undertone to this, with a twist. There are those who would like to blame non-white people for all the crime, and this is a version of that. The same racist “science” that underscores the link between melanin in one’s skin and criminal behavior also purports that there is a link between melanin in one’s skin and all the things that lead to single motherhood, from promiscuity to a tendency to not follow social rules to, amazingly, a shorter gestation period which would, I assume, allow single mothers to produce more offspring than might otherwise be possible.

Anyway, there are those who remember the link being made, and here is what the data for violent crime in the US looked like at the time:

The correlation between single motherhood and crime rates is clear. The single mothers must be causing the crime by producing criminal babies!

Nice relationship there. All else being equal one would want to look at these two variables and see if there is anything to the argument. We KNOW the two variables are related, because otherwise, why would anyone put them on the same graph!@!?? And, once they are on that graph they are perfectly correlated. One small thing that would need to work out is the fact that the criminals are probably not infants, but I’m sure that the time lag thing is just a detail.

Anyway, now, time has passed since that correlation seemed to be explanatory of high crime rates. Has anything changed since then? Let’s look at the data, updated:

Single mothers appear to have stopped causing crime at around 1990 or so.

Well, now that we see the whole data set, one might wonder why single motherhood rate was not correlated to crime rate prior to 1970. Maybe that’s the time lag thing. It turns out, in fact, that of the 50 years shown on this graph, the correlation seems to work for only about half the time.

Single mothers, you are off the hook. FOR NOW!

Comments

  1. #1 Philip Cohen
    November 27, 2012

    That graph is for the whole country, FYI. -pnc

  2. #2 Eric Lund
    November 28, 2012

    Well, now that we see the whole data set, one might wonder why single motherhood rate was not correlated to crime rate prior to 1970.

    Because it was the Dirty F—ing Hippies, whose numbers increased rapidly through the 1960s, that’s why. </snark>

  3. #3 Greg Laden
    November 28, 2012

    Phillip: Thanks for pointing that out, you are correct. The ORIGINAL story on which the MJ story was based discussed Washington DC specifically, as that is where the Single Mother – Homicide connection was being made. The MJ graph is better because the data are more robust.

    The DC oriented post is here: http://www.theatlantic.com/sexes/archive/2012/11/single-moms-cant-be-scapegoated-for-the-murder-rate-anymore/265576/

  4. #4 Nathaniel Cooke
    Bristol, UK
    November 30, 2012

    Statistically more broken homes &* single parents occur in poorer communities, where there is also more violent crime. In certain circumstances a lack of suitable role models may well make someone more likely to be a violent criminal, but that could just as easily happen if they had two parents who also happened to be assholes.

  5. […] Correlation and Causation: Single Mothers and Violent Crime [Greg Laden's Blog] (scienceblogs.com) […]

  6. #6 Diaz sebastien
    Paris
    January 3, 2013

    I m studying the structure of violent crimes on the usa and i’m very interested to see the evolution of the sungle mother during the time. My study is on recent information and based on more social statistics. I can say that the structure of the family is one of the best thing to influence the violent crimes activities. Others parameters are the poverty,the immigrant and the education.
    My site:
    http://sebastiendiazcrimesanalysis.blogspot.fr

  7. #7 IrishFarmer
    somewheres
    November 6, 2013

    Your chart, while more robust, is also misleading. You’re looking at crime as a whole, rather than on the level of demographics. The fact that chances of a person committing violent crimes rises if they were raised in a single-mother household doesn’t necessarily mean that as single motherhood increases, the overall violent crime rate will increase as well. A recent article compares this to poverty, where poverty increases the incidence of crime, but during the recession when poverty increased crime did not.

  8. […] Correlation and Causation: Single Mothers and Violent Crime [Greg Laden's Blog] (scienceblogs.com) […]

  9. #9 ScottishFarmer
    SCOTLAND OBVIOUSLY
    October 2, 2014

    Someone needs to give out trophies for worst run on sentences:

    “There is a racist undertone to this, with a twist: There are those who would like to blame non-white people for all the crime, and this is a version of that; the same racist “science” that underscores the link between melanin in one’s skin and criminal behavior also purports that there is a link between melanin in one’s skin and all the things that lead to single mother hood, from promiscuity to a tendency to not follow social rules to, amazingly, a shorter gestation period which would, I assume, allow single mothers to produce more offspring than might otherwise be possible.”

    Horrid!

  10. #10 Greg Laden
    October 2, 2014

    Well, that is not really one sentence. But it is a bit run on. Sorry for ruining your day!

    I did, though, change some of the punctuation to make it seem less like a run on sentence.

  11. #11 Roger
    November 12, 2014

    Irishfarmer’s comment is correct: “You’re looking at crime as a whole, rather than on the level of demographics.”
    Both graphs contain data showing the incidence of single mother households and the rate of violent crime. In the first chart both datasets show these measurements rising and so you have interpreated that this means these two are corerleated. The second graph shows the single mother measure continuing to rise while the crime measure decrease, and so you have interpreated that this means the two are not correlated.
    If you’re wanting to see if there is link between the two measures simply graphing each measure and seeing if they rise and fall at a similar rate over time is not the way to achieve this. What needs to be done to examine this potential link is to examine crime rates in single mother households and compare those rates to crime rates for other types of households (dual-parent married, dual-parent not-married, single father, etc.). A good scientist would adjust for variables like education or income, for examples, as well (so you’re comparing the single mother households in the low income strata with, other households also in the low income strata, and then single mother households in the middle income strata with, other households also in the middle income strata, etc.).
    What you have done with your two graphs proves, frankly, absolutely nothing either way.

  12. […] Correlation and Causation: Single Mothers and Violent Crime [Greg Laden's Blog] (scienceblogs.com) […]

  13. #13 Sally
    March 10, 2015

    But single mothers should be blamed, they made the choice to have sex with a man they didn’t know too well (or a man that they knew wasn’t father-figure material).
    There is no racism about these findings, either. Even when you account for race, economic background, etc. the findings are still the same: that children need both their mother and their father during their upbringing, or else there will be various ill effects. (more specific findings on emotional problems, education, crime and obesity on fatherhood dot org).

  14. #14 Greg Laden
    March 10, 2015

    Sally, I agree that there is a likely link, on average, between single motherhood and various problems. But, you have characterized each and every single motherhood situation very precisely and, there is no cause to do that; you are simply wrong at the very least by oversimplification. And, regardless of any issues with family size/structure, the point made here is still valid about what is and is not correlated. You are trying to support a correlation shown to be numerically invalid based on your strong belief that it must be valid.

  15. #15 Anton Webber
    Midlothian, TX
    October 21, 2015

    Isn’t it possible that “broken windows” policing and “tough on crime” legislation introduced in the 90s can explain the drastic change in violent crime rates that you show in your second graph?

  16. […] social demographicsof the household. It seems since having been given enough time, we see the correlation drop off.Single Moms Can't Be Scapegoated for the Murder Rate Anymore – The Atlantic That's true, […]

  17. #17 A&F
    USA
    February 9, 2016

    You are purposely being disingenuous The chart is misleading (and no signal mothers are not “off the hook”) that sharp drop in crime is the direct result of more aggressive policing and mass incarceration. (Control for that and factor this fact into your chart sir) and you still see the casual relationship to crime and single motherhood.

  18. #18 Greg Laden
    February 9, 2016

    A&F, you have made a very specific argument about causality, a simplistic one, that you have not supported with good evidence or argument.

  19. #19 dean
    February 9, 2016

    ” that sharp drop in crime is the direct result of more aggressive policing and mass incarceration.”

    A simplistic statement that is not directly supported by data.
    – It isn’t obvious why you say “single mothers are not off the hook” – you haven’t given anything to support it
    – The statement that the decline in crime is due to aggressive policing and mass incarceration is not defendable. it ignores the fact that crime was decreasing prior to the increased policing, and ignores the fact that areas that did not step up policing and incarceration saw similar decreases
    “(Control for that and factor this fact into your chart sir) and you still see the casual relationship to crime and single motherhood.” Again, no evidence in support of this: even if true it would not provide evidence of a causal relationship

  20. #20 David Ivancheck
    Florida
    February 18, 2016

    Unfortunately Greg, graph are made from statistics, and statistics can be adjusted up and down, depending on the motives of an individual. I think the most accurate way to do these things is to deal with this, is to have the FBI gather the information themselves, directly from the CORRECTION FACILITIES, themselves. I think when the prisoner first comes into a facility, one of the questions they should be asked is whether they grew up with both mother and father, at home. This should give us a fairly accurate assessment of whether single family homes are a direct cause of crime.

    Personally, I believe that it is THE ENVIRONMENT that the child grows up in; but we have to BE REAL with ourselves, MANY single parent households have problems. For the most part MANY of the mothers get OVERWHELMED with COOKING for the child; CLEANING the child; getting the child to SCHOOL, and then doing homework with the child; it can bring a mother to the point of despair…and when this happens, many times THE HOMEWORK goes out of the window; the child sees that their not doing well in school; they feel inadequate and THEN COMES THE ANGER and ATTITUDE;.and that’s all you need to create A CRIMINAL…ANGER and DISPAIR. The average person LOCKED UP, felt ANGRY about his position in life, and felt there was NO WAY OUT

  21. #21 Greg Laden
    February 18, 2016

    Certainly, one could argue that the research is in adequate and argue for more. Also, I’m sure you are right about the importance of the environment, and certainly, single mothers are likely to be overwhelmed.

    However, the point of this post is that a certain graph (based on data as you note) was misused, and causality misattributed.

  22. #22 dean
    February 18, 2016

    What does research say about people who use randomly placed (and unnecessary) upper case letters?

  23. #23 Brainstorms
    February 18, 2016

    This should give us a fairly accurate assessment of whether single family homes are a direct cause of crime.

    In a word, no. It will give you some assessment of correlation, but it will not establish causation.

    (And its assessment will be cloudy and noisy, as there are many variables at play — and the answers may not even be truthful.)

    Personally, thanks for SHARING your OPINIONS about RAISING CHILDREN, but your GUESSES about the CAUSES of criminality are JUST FANCIFUL musings that MAY OR MAY NOT be accurate.

    Hence, such musings are made from personal viewpoints and agendas, and opinions paraded as facts can be adjusted up & down, depending on the motives of the opinion-giver. I think the most accurate way to do this is to collect data and perform scientifically correct and peer-reviewed analyses.

  24. #24 Stan
    July 16, 2016

    Does anyone actually think that a single parent can raise a child better than two parents can, on average?