The people most visible in autism discussion are not the whole autistic population. The opinions most often repeated are not necessarily the most widely held. And the people who disappear may have something important to teach us.
I call the wider problem selective visibility: our picture depends on who can enter the discussion, who finds it worthwhile, who attracts attention and who remains. This is not a claim that an unseen majority secretly agrees with me. Silence does not give anybody a mandate.
An engaged audience is not the population
Singapore election discussion provides a useful warning. A stream of enthusiastic opposition posts can make a landslide seem plausible to somebody reading it, only for election day to deliver a very different picture. The mistake is treating an information environment as though it were the electorate. The eventual result does not, by itself, tell us which voters were previously silent or what the parties privately expected.
Political participation also has different levels. The US Census Bureau reported that 65.3% of the citizen voting-age population voted in the 2024 presidential election. Voting, campaigning and posting frequently about politics are not the same activity. A politically active online audience is not a census of either voters or non-voters.
For autism, we must distinguish the general public, autistic people and caregivers, and the much smaller circles that happen to supply a particular discussion. An organisation can faithfully report what its participants said while still overstating what those participants establish about everybody else.
Why the visible picture becomes distorted
Network structure and message volume. The majority illusion describes how something uncommon across a network can appear common among many people’s immediate contacts. Repetition adds another distortion: in Pew’s study of US adult Twitter users, the most active quarter produced 97% of tweets during the three-month 2021 observation period, including reposts and replies. Counting messages is not counting people.
Selection, coverage and nonresponse. A service list misses people who do not use the service. A public survey misses people who never encounter it or choose not to respond. A meeting excludes those who cannot use its format. Inviting more people through the same narrow channels does not necessarily correct the original gap.
Spiral of silence. People may withhold views when they expect disapproval or isolation. Pew’s research on willingness to discuss a controversial issue found greater willingness where people believed their audience agreed. Lack of interest is a different explanation from inhibited expression.
Pluralistic ignorance. People can also misjudge what others privately believe, helping sustain a norm that many do not personally endorse. Research on college drinking norms provides an example; it does not establish the views of an autism community.
Selective attrition. As people leave, the remaining participants can become less representative of everyone who entered. In this context, survivorship bias concerns whoever remains observable, not necessarily whoever has the most successful life.
These mechanisms can overlap, but they are not interchangeable explanations for a single scientifically established “silent majority effect”. The practical question is which of them is operating in the situation before us.
Public support is not the same as dependable inclusion
I am concerned about the gap between public enthusiasm for inclusion and what happens when inclusion becomes personal, inconvenient or consequential. Imagine someone who celebrates autistic achievement online but refuses to assess a qualified autistic applicant fairly. Another person never posts about inclusion but quietly makes their workplace usable. Public visibility would rank their contributions incorrectly.
Singapore’s Disability Trends Report 2024 recorded positive attitudes towards autistic people falling from 69.9% in 2019 to 56.2% in 2023. In 2023, 34.8% were classified as neutral and 9.0% as negative. These questionnaire categories do not identify who speaks publicly or establish how people behave.
There is also evidence against assuming that support is confined to a tiny vocal minority. An IPS survey of 1,801 respondents found substantial stated support for disability-inclusive measures even when possible drawbacks were specified. Neither blanket reassurance nor blanket cynicism describes these findings adequately.
Silence can conceal hostility, agreement, uncertainty, indifference or attention directed elsewhere. We need to distinguish what people believe, what they feel able to say, and what they actually do. Even positive changes in one measure of acceptance need not imply change in every other.
Nor is an ableist majority necessary for serious exclusion. In a hypothetical workplace, one person controlling a crucial appointment or promotion can block an opportunity despite goodwill from everyone else. Conversely, a few competent allies may make a large difference when they can change the relevant decision. Count authority and consequences as well as opinions.
Who has enough reason to keep speaking?
To become a visible participant, someone must recognise autism as relevant, find a channel, have a usable way to contribute, consider participation worthwhile, tolerate its costs and continue. Public attention adds another filter: not every contribution receives the same amplification.
I wonder whether sustained advocacy disproportionately attracts people with unusually strong reasons to share or act. These might include intellectual interest, a wish to help, concern about injustice, professional work, a need for companionship or the importance of explaining personal experience. These are questions about motivation, not diagnoses inferred from posting frequency.
Initial motivation and continued participation are separate stages. A highly committed person may join because they want improvement, challenge established practices for the same reason, and eventually leave when the response makes further effort unattractive. I recognise aspects of this in my own autism work. Losing willingness to work under particular conditions is not necessarily losing concern for the purpose.
Those who stay may be unusually patient, have few alternatives, value genuine friendships, enjoy sharing, use only selected channels or simply encounter a different side of the group. Remaining does not prove approval; leaving does not establish superior judgement.
Here, “vocal” includes writing and communication aids. It is not a synonym for fluent speech or low support needs. Someone can be strongly motivated to discuss autism while struggling with unrelated daily tasks. Activity, expertise, leadership readiness and general independence are different qualities.
Testimony is not automatically a mandate
“This happened to me”, “I have relevant expertise”, “this study estimates a population’s views”, and “these members authorised me to represent them” are four different claims. None automatically supplies the others.
An advocate need not be typical to identify something true, important or useful. Minority status does not invalidate an argument, and protection from mistreatment should not depend on winning a popularity contest. Equally, sharing a diagnosis does not authorise someone to announce everybody else’s preferences.
A member-elected organisation has a defined constituency. A public speaker may have expertise without that mandate. A consultation needs to name the people it reached and the people it did not. Neither visible advocates nor their opponents can claim the silent population by default.
Presence Is Not Influence asks whether participation produces real authority. The prior question here is whose experiences reached the room at all. Selection determines what decision-makers can hear before they decide whose contribution to use.
Private lives and lives outside autism organisations
Hidden Autistics include people with private diagnoses, selective disclosure, private self-identification or uncertainty about whether autism describes them. Non-membership is another distinction: someone can be openly autistic and have no interest in an autism organisation. Others want to participate but cannot access a suitable format. These circumstances require different responses.
Not wanting an autism-centred life is different from not caring about autistic people’s welfare. A person may value an adjustment, support a practical reform and still prefer to spend their time on work, family or another cause. Agreement with one proposal should not require buying an entire package of terminology, treatment positions and political loyalties.
A private adult may already have a sustainable life, be struggling without suitable help, or be doing well in some areas and poorly in others. Privacy is not evidence of either success or concealed suffering. My hidden-success majority hypothesis remains a question for research, not a reason to populate the unseen world with uniformly accomplished people.
Hidden Caregivers may also remain outside public networks because of privacy, time, disappointment or uncertainty about what participation offers. But the caregiver’s account and the autistic person’s wishes are separate. A parent may want publicity that their child does not; an adult may want understanding that a family would prefer to avoid. Caregiver participation does not automatically establish autistic representation.
Keeping information private is different from denying someone useful self-understanding or support. Where communication is difficult, assistance should help the person express their position rather than silently replace it.
Disclosure also distributes benefits and risks unevenly. A movement may gain another public example while the individual bears the consequences. A positive story from a selected disclosure situation does not guarantee safety elsewhere. Making an experience understandable need not make its owner publicly identifiable.
Consensus can increase without persuasion
Consider a hypothetical group with 60 members supporting a position and 40 opposing it. If 35 opponents leave, the remaining group has 60 supporters out of 65 members: about 92.3% support instead of 60%. Nobody has changed their mind.
A community can become more unanimous while becoming less representative. Reduced criticism might mean persuasion, but it might also mean that disagreement became too costly. A survey of those who stayed cannot establish why the others left, and even an exit survey misses people who considered joining but decided against it.
The direction of distortion is not always pessimistic. Successful people may stop needing a service, leaving persistent difficulty overrepresented among current clients. People exhausted or harmed by participation may disappear too, making remaining satisfaction look better. Meanwhile, selected success stories can dominate publicity. Quiet successes and quiet failures can vanish from the same account for opposite reasons.
Departure itself is not proof of harm. People may graduate, change priorities or oppose a justified conduct restriction. In my own groups, some critics left and some later returned because they valued information I shared. Their return did not establish agreement with my moderation. A useful offering and an approving constituency are different things.
When Inclusion Quietly Excludes examines the conditions behind such decisions, including my experience of both challenging organisations and making difficult decisions myself.
What disappears with the people?
A departing contributor takes more than a membership number. They may take practical methods, professional connections, corrective feedback and knowledge of alternatives. Recruiting an enthusiastic newcomer does not necessarily replace everything a previous contributor had learned.
Repeated hostility to feedback can also discourage people from advising someone else who would welcome it. The costs extend beyond the original exchange. If useful outside standards disappear, a group’s familiar way of working can start to look inevitable.
The same visibility problem affects future planning. People facing an immediate problem can insist on attention; people who will face an avoidable problem years from now are not yet there to ask. My discussion of reactive drift explains how an initiative can stay busy while losing its strategic purpose.
Departure can mean rejecting one association, becoming wary of organised advocacy, leaving the wider movement or redirecting contribution elsewhere. These outcomes should not be merged. Neither an organisation nor its current leaders own the cause.
Learn without making belonging compulsory
Research should distinguish finding useful experiences, estimating how common they are and testing which methods transfer. A set of biographies cannot establish a majority; a population percentage cannot explain how somebody else could develop.
Recruit through broad, voluntary invitations as well as service and advocacy networks. Include quiet participants, former participants, never-joiners and private adults with contrasting outcomes. Ask separately about beliefs, public expression, behaviour, usefulness, safety, costs and reasons for leaving. Allow “I do not know” and refusal without interpreting either as agreement.
A larger convenience sample is not automatically representative. Report the actual denominator, recruitment channels and missing perspectives. Where missing outcomes could change the conclusion, examine that uncertainty rather than filling it with a preferred story. A small organisation need not conduct a population study, but its public claims should fit its evidence and authority.
Community can be beneficial. A study of 196 autistic participants associated greater community connectedness with better wellbeing, while noting its online recruitment and the need to include people who reject community involvement. The finding supports voluntary opportunity, not compulsory belonging.
For accomplished private adults, offer a worthwhile, bounded contribution: a review, explanation, research interview or mentorship with clear scope, appropriate recognition and control of attribution. The AFAM’s confidential pathway does not require public identification, client status or lifelong advocacy. Professional success does not automatically confer mentoring skill; useful methods still need testing.
The purpose is to recognise both neglected harm and neglected value. Sometimes something is unknown. Sometimes it is unsafe to report. Sometimes it is already known but discounted. Those are different problems, and more publicity will not solve all of them.
The task is not to make everybody visible. It is to stop making visibility the price of being counted.