TV Does Not Cause Autism
10/17/06 14:25 Filed in: Research
This morning (10/17/06) an article appeared on Slashdot referring to a study that ostensibly claimed that television watching causes autism.
As has become very common in our media-driven society, the concepts and information in the study were significantly simplified for the article, and conclusions were drawn that the study, nor its authors, could possibly support. Unfortunately, when it comes to our children, and particularly disorders as puzzling as autism, everyone dealing with it – parents, professionals, administrators – is desperately hungry for any advance in understanding. This being the case, information presented in this shorthand way quickly becomes accepted fact, despite it’s not being an accurate representation of the available information. Given this, I am very concerned that we will very soon be hearing it stated, as if it were established fact, that television causes autism.
In light of this, let me say up front that this entry probably should be more accurately titled “this article does not show that TV watching causes autism”.
The study, the paper for which can be found here, is a correlational and regressional analysis of the rate of television watching as compared to the rates of autism diagnosis. The word correlation is very important here, as it refers to a relationship between two or more factors. In a positive correlation, one factor increases in relation to another factor. An example of a positive correlation would be the relationship between height and weight – as a general rule, as height increases, weight increases.
A negative correlation is one in which, as one factor increases, the other factor decreases. An example of this would be the average rate of traffic citations as related to age – as your insurance company will tell you, as driver age increases the rate of traffic citations declines.
It is extremely important to understand that, while a correlational analysis shows a relationship, the one thing that it cannot show is cause. It is a mantra that beginning statistics students learn to chant: Correlation does not imply causation. Weight does not cause height, and being young does not cause traffic citations to occur (the parents of adolescents will undoubtedly shake their heads at this statement – what I mean to say is that the age, by itself, does not cause the citations to occur. There is a third factor – impulsivity and poor judgment – which is also associated with youth, that causes the traffic citations to occur). Unfortunately, the author’s of this article, particularly in Section III, further confuse this issue with their liberal and inaccurate use of the word “trigger” to refer to the correlational findings in other studies.
A regressional analysis is a means by which one factor can be used to predict another. If I know your height, for example, I can make predictions about what your weight, within a certain range, is likely to be. It also does not speak to causes.
Another thing that media reports often do not explain well is the quality of the study, or the data upon which it is based. This study was conducted by a pair of economists using existing available data to examine relationships between the rate of television watching in the homes of children under three years of age, and comparing this to the rates of autism diagnosed. As is often the case with existing data, the authors were limited to what was available. Given this, it is vital to understand that the authors had absolutely no information about the television viewing habits of the children. The data that they had available to them was about the television viewing habits of the adults in the home. They then used data about the weather to determine whether it was likely or not that the children would have been spending more, or less time inside during the time the adults were watching television.
As you can probably begin to see, there are an awful lot of “ifs” and “mights” that we must agree to before we begin to agree that there is even any useful information from this report. We must assume that the time adults spend watching television is equivalent to the time the children are watching television. The authors attempt to bolster this by using their weather analysis to make it seem unlikely that the children are playing outside during those times. This may seem a reasonable effort, but all it really tells us is that they are likely to have been inside. During that time they may have been watching television with their parents, but it seems just as likely that they may have been playing on their own, playing with one parent while another was watching television, napping, eating or a myriad of other potential activities. In addition, it must be noted that, for many households it seems likely that some of the adult television watching time occurs after the young children in the household have gone to bed. We must assume, based upon the author’s presentation, that the children are living in the same location at eight years of age that they were in from birth to three years of age. We must agree that the populations living on the west coast of California and Oregon are comparable to the populations in the eastern half of those states (which they are clearly not, the largest population centers in both states being on the west coast). We are supposed to accept the author’s results, even though they do not show a positive correlation in California, the most populous state in the union, and, as such, arguably the most valid population upon which to conduct the analysis. We are supposed to accept their initial results as valid, and then take the leap to believing that the rate of autism has grown because of the growth of cable television. As I noted, an awful lot of “ifs” and “mights”
Finally, even if we agree to the ifs and mights all we have at the end of this study is a correlation between adult television watching and autism. This, despite the unfortunate use of the word “trigger” by the authors. What we can say is that there appears to be a relationship between the amount of time adults in a home watch television, and the incidence rate of autism. Assuming this relationship holds true (and it is the only study of its kind that I am aware of, so there is no guarantee that it will hold true under repeated examination), there are a wide array of variables that could account for this relationship. One obvious example is the fact that, for many children with autism, the symptoms show up very early on. Most children with autism are considerably less socially engaging than typically developing children. In those homes, it may simply be the case that the adults spend less time on social interaction with their child (because the child is not as interested in social interaction) and, as a result, more time watching television. It’s also the case that the increased rates to which the authors refer occur in larger populations centers, which would be expected to have more diverse populations (including a greater variety of people with disabilities), and a far greater array of professionals trained in recognizing autism. Or it may simply be coincidental, with no causative variable at all.
What is certain is that there is no information about this study that implicates anything about the watching of television itself as a cause of autism. There is not one iota of research information in this study that suggests that the lighting, the type of story pattern, the rapid changes in subject matter, or any other factor consistent with television watching, is causative of autism. It is simply not there. Unfortunately, the authors of this study do not report their results in the typical, responsibly cautious manner usually seen in research studies. Rather, they fragrantly refer to television “triggering” autism several times throughout the study, despite the fact that they are not providing any information that actually supports this. In fairness, the authors try to argue, in a very convoluted fashion, that because precipitation results in higher likelihood of TV watching, and TV watching is correlated to autism, and because having autism cannot cause precipitation, it stands to reason that TV watching triggers autism. They are careful to qualify these statements later in the study. However, the study itself is 67 pages long, and written in very convoluted technical language. The paper’s title – “Does Television Cause Autism”, is the clearest, simplest portion of the paper to read, and that title is what has gotten media attention.
If there is irresponsibility here, it is in the release of the information to the general media and allowing it to be summarized in the way it has been. This is certainly not a limitation common only to this study – it’s happening all of the time nowadays.
There has been much discussion about the rising rate of autism in our country, and what this actually means. Some camps hold that there is an epidemic, while others feel that it’s due to a broadening of diagnostic criteria, or better understanding and recognition on the part of the people doing the diagnosing.
Update: We've moved to a new comment system - one that will work better with the website overall, I think. However, as a part of that transition we would have lost a thoughtful and heartfelt comment made by a parent in relation to this article. That comment is printed in it's entirety below - EJW:
Thanks for such an interesting article. However I would pray that parents and professionals who know the slightest thing about Autism, realize this is just a stab at money or fame at seeing their name published with such a crazy article. What kid doesn't sit around and watch the TV? Oh yeah my son with Autism does not watch TV. The sounds and vibrations are very annoying to his nervous system. And how about the article in my local newspaper earlier this year that headlined "Rainfall levels could cause increases in autism rates". It reads: An analysis of data from three states suggests that counties with higher precipitation levels also have higher autism rates......I wonder about those other states as well??? Three states isn't a very big percentage. And of course we can't forget Jenny McCarthy who states her son is "cured" and the biggest one of all....vaccinations! So my own opinion as a parent, is read what you will, but really in the end, its up to us the parents to love and protect and be sure we are reaching every possible avenue to provide for our loved ones the therapies, structure and routine to make their daily living the best possible, even though it isn't exactly perfect for us. Thanks for the great website Dr Wade. Great work! Kim Zera | 04.22.09 - 6:59 pm
As has become very common in our media-driven society, the concepts and information in the study were significantly simplified for the article, and conclusions were drawn that the study, nor its authors, could possibly support. Unfortunately, when it comes to our children, and particularly disorders as puzzling as autism, everyone dealing with it – parents, professionals, administrators – is desperately hungry for any advance in understanding. This being the case, information presented in this shorthand way quickly becomes accepted fact, despite it’s not being an accurate representation of the available information. Given this, I am very concerned that we will very soon be hearing it stated, as if it were established fact, that television causes autism.
In light of this, let me say up front that this entry probably should be more accurately titled “this article does not show that TV watching causes autism”.
The study, the paper for which can be found here, is a correlational and regressional analysis of the rate of television watching as compared to the rates of autism diagnosis. The word correlation is very important here, as it refers to a relationship between two or more factors. In a positive correlation, one factor increases in relation to another factor. An example of a positive correlation would be the relationship between height and weight – as a general rule, as height increases, weight increases.
A negative correlation is one in which, as one factor increases, the other factor decreases. An example of this would be the average rate of traffic citations as related to age – as your insurance company will tell you, as driver age increases the rate of traffic citations declines.
It is extremely important to understand that, while a correlational analysis shows a relationship, the one thing that it cannot show is cause. It is a mantra that beginning statistics students learn to chant: Correlation does not imply causation. Weight does not cause height, and being young does not cause traffic citations to occur (the parents of adolescents will undoubtedly shake their heads at this statement – what I mean to say is that the age, by itself, does not cause the citations to occur. There is a third factor – impulsivity and poor judgment – which is also associated with youth, that causes the traffic citations to occur). Unfortunately, the author’s of this article, particularly in Section III, further confuse this issue with their liberal and inaccurate use of the word “trigger” to refer to the correlational findings in other studies.
A regressional analysis is a means by which one factor can be used to predict another. If I know your height, for example, I can make predictions about what your weight, within a certain range, is likely to be. It also does not speak to causes.
Another thing that media reports often do not explain well is the quality of the study, or the data upon which it is based. This study was conducted by a pair of economists using existing available data to examine relationships between the rate of television watching in the homes of children under three years of age, and comparing this to the rates of autism diagnosed. As is often the case with existing data, the authors were limited to what was available. Given this, it is vital to understand that the authors had absolutely no information about the television viewing habits of the children. The data that they had available to them was about the television viewing habits of the adults in the home. They then used data about the weather to determine whether it was likely or not that the children would have been spending more, or less time inside during the time the adults were watching television.
As you can probably begin to see, there are an awful lot of “ifs” and “mights” that we must agree to before we begin to agree that there is even any useful information from this report. We must assume that the time adults spend watching television is equivalent to the time the children are watching television. The authors attempt to bolster this by using their weather analysis to make it seem unlikely that the children are playing outside during those times. This may seem a reasonable effort, but all it really tells us is that they are likely to have been inside. During that time they may have been watching television with their parents, but it seems just as likely that they may have been playing on their own, playing with one parent while another was watching television, napping, eating or a myriad of other potential activities. In addition, it must be noted that, for many households it seems likely that some of the adult television watching time occurs after the young children in the household have gone to bed. We must assume, based upon the author’s presentation, that the children are living in the same location at eight years of age that they were in from birth to three years of age. We must agree that the populations living on the west coast of California and Oregon are comparable to the populations in the eastern half of those states (which they are clearly not, the largest population centers in both states being on the west coast). We are supposed to accept the author’s results, even though they do not show a positive correlation in California, the most populous state in the union, and, as such, arguably the most valid population upon which to conduct the analysis. We are supposed to accept their initial results as valid, and then take the leap to believing that the rate of autism has grown because of the growth of cable television. As I noted, an awful lot of “ifs” and “mights”
Finally, even if we agree to the ifs and mights all we have at the end of this study is a correlation between adult television watching and autism. This, despite the unfortunate use of the word “trigger” by the authors. What we can say is that there appears to be a relationship between the amount of time adults in a home watch television, and the incidence rate of autism. Assuming this relationship holds true (and it is the only study of its kind that I am aware of, so there is no guarantee that it will hold true under repeated examination), there are a wide array of variables that could account for this relationship. One obvious example is the fact that, for many children with autism, the symptoms show up very early on. Most children with autism are considerably less socially engaging than typically developing children. In those homes, it may simply be the case that the adults spend less time on social interaction with their child (because the child is not as interested in social interaction) and, as a result, more time watching television. It’s also the case that the increased rates to which the authors refer occur in larger populations centers, which would be expected to have more diverse populations (including a greater variety of people with disabilities), and a far greater array of professionals trained in recognizing autism. Or it may simply be coincidental, with no causative variable at all.
What is certain is that there is no information about this study that implicates anything about the watching of television itself as a cause of autism. There is not one iota of research information in this study that suggests that the lighting, the type of story pattern, the rapid changes in subject matter, or any other factor consistent with television watching, is causative of autism. It is simply not there. Unfortunately, the authors of this study do not report their results in the typical, responsibly cautious manner usually seen in research studies. Rather, they fragrantly refer to television “triggering” autism several times throughout the study, despite the fact that they are not providing any information that actually supports this. In fairness, the authors try to argue, in a very convoluted fashion, that because precipitation results in higher likelihood of TV watching, and TV watching is correlated to autism, and because having autism cannot cause precipitation, it stands to reason that TV watching triggers autism. They are careful to qualify these statements later in the study. However, the study itself is 67 pages long, and written in very convoluted technical language. The paper’s title – “Does Television Cause Autism”, is the clearest, simplest portion of the paper to read, and that title is what has gotten media attention.
If there is irresponsibility here, it is in the release of the information to the general media and allowing it to be summarized in the way it has been. This is certainly not a limitation common only to this study – it’s happening all of the time nowadays.
There has been much discussion about the rising rate of autism in our country, and what this actually means. Some camps hold that there is an epidemic, while others feel that it’s due to a broadening of diagnostic criteria, or better understanding and recognition on the part of the people doing the diagnosing.
Update: We've moved to a new comment system - one that will work better with the website overall, I think. However, as a part of that transition we would have lost a thoughtful and heartfelt comment made by a parent in relation to this article. That comment is printed in it's entirety below - EJW:
Thanks for such an interesting article. However I would pray that parents and professionals who know the slightest thing about Autism, realize this is just a stab at money or fame at seeing their name published with such a crazy article. What kid doesn't sit around and watch the TV? Oh yeah my son with Autism does not watch TV. The sounds and vibrations are very annoying to his nervous system. And how about the article in my local newspaper earlier this year that headlined "Rainfall levels could cause increases in autism rates". It reads: An analysis of data from three states suggests that counties with higher precipitation levels also have higher autism rates......I wonder about those other states as well??? Three states isn't a very big percentage. And of course we can't forget Jenny McCarthy who states her son is "cured" and the biggest one of all....vaccinations! So my own opinion as a parent, is read what you will, but really in the end, its up to us the parents to love and protect and be sure we are reaching every possible avenue to provide for our loved ones the therapies, structure and routine to make their daily living the best possible, even though it isn't exactly perfect for us. Thanks for the great website Dr Wade. Great work! Kim Zera | 04.22.09 - 6:59 pm
