Date Revised: October 22, 2021
You have questions: What causes autism? What therapies and services work best? And one more thing: After decades of research, why don’t we have the definitive answers to those questions?
First, it is important to follow the road researchers travel from the beginning, as they plan an autism study, to the end, when they publish their results for others to see. Along the way, scientists take steps to make sure that studies are accurate and fair.
They start with a question to be researched, called a hypothesis. They read what other researchers have already discovered about that question. They ask if the hypothesis makes sense, given what is already known about biology, genetics, or psychology, for instance.
They must select the kind of study they want to do, such as:
- A case study is a detailed report about one person or case.
- A correlational study uses data from participants to see how different factors relate to each other. For example, is having insomnia related to having gastrointestinal problems in autism?
- A longitudinal study follows a group of people over time to see how they develop and change. For example, a researcher could follow 500 people with autism from childhood through adulthood.
- An experimental study and clinical trial involve changing one thing to test its effect on other factors. Does a new medicine work better than a sugar pill for insomnia?
Removing Bias from Research
We all have opinions. But the role of science is to look for evidence in a way that is unbiased. So how do researchers do that?
Let’s say researchers are planning an experimental study to test a new treatment for speech delays in children with autism.
One common way to create a fair experiment is to divide participants into two groups. One group, the treatment group, will get the new therapy that is being tested. The second group will get either no treatment or an established treatment. This group is called the control group. Scientists will know that the new treatment works if the treatment group makes more progress than the control group.
But selecting people for the treatment and control groups must be done fairly, too.
It would be unfair to put only children with mild speech delays in the treatment group and only children with severe speech delays in the control group. At the end of the study, the researchers could say that the treatment group scored better on speech tests than the control group. But that may be because the treatment group had fewer speech problems at the start. This could lead the researchers to think that the new speech program was better than it really was.
One way to be fair is to assign people to the groups randomly, or to make sure that both groups have similar make-ups. For example, each group might have the same percentage of boys and girls, the same percentage of children with mild, moderate, and severe speech delays, and so on.
Researchers hoping to make a scientific breakthrough may be affected by their own expectations. Let’s say that they are studying a new drug for a medical problem. They might be more likely to notice an improvement if they know someone was taking the medicine and ignore an improvement if they know someone was taking a fake pill, called a placebo. That is why many medical studies do not allow the participants or the researchers to know who is getting the medicine and who is getting the placebo.
When the study is done, researchers usually submit a detailed description of their procedures, methods, and results to a professional journal. Ideally, experts would then see if the research makes scientific sense. That process is called peer review, and it is used by many journals to weed out research that may be unfair or invalid. The research paper will be published if the peer reviewers believe the study is acceptable.
Research and Repeat
Other researchers may want to repeat the experiment to see if they get the same results, often using larger groups of people. Say a study shows that an autism therapy improved social skills for 7 of the 10 people who participated. Will the therapy give the same results for 70 percent of all people with autism? It’s possible that there was something about the design of that study, or about the 10 people who were picked for it, that made the therapy look better than it really is.
So researchers may want to test the treatment in 50, 100, or 500 people to see how it works, while also using control groups. These are called replication studies.
It can take time to design a fair study, find participants, conduct the research, analyze data, and publish the results in a peer-reviewed journal.
Maybe you want answers now. Can scientists skip a few steps?
In some cases, when research has been sloppy, biased, or inadequately reviewed, new drugs and medical devices have come onto the market that did not work or were dangerous. Some of these medicines or medical devices have been recalled, as a result. Sloppy research could lead to effective treatments being abandoned too soon or ineffective ones looking better than they really are.
Making Research More Efficient
But that does not mean that research has to move slowly. The SPARK study hopes to make autism research more efficient by providing a huge number of people to study. SPARK has enrolled more than 275,000 people, including 100,000 people with autism and their relatives. With so many study participants, researchers hope to learn about the biology and genetics of autism.
SPARK also encourages research on other autism topics. Through its Research Match program, SPARK makes it easier for outside researchers to find people who want to participate in their studies.
Maybe your questions about autism have not been answered yet. But you can see what researchers have learned since SPARK began by browsing a list of research publications that rely on SPARK data and participants.
Photo credit: SPARK photo of Dr. Pamela Feliciano and Dr. Wendy Chung
For information about research on autism in adults, see the SPARK article, “Answering the Questions about Autism in Adulthood.”