Development of a Novel Self-Report Tool to Aid in the Assessment of Autism in Adulthood: The Autism Spectrum Trait Scale

Abstract

Background:

Autism spectrum disorder is a neurodevelopmental condition characterized by a spectrum of neuropsychological and behavioral differences. Due to historical changes in diagnostic criteria and a rapid increase in autism rates, there are likely many undiagnosed autistic adults. The current state of formal assessment for autism poses a severe ethical crisis as these assessments are often difficult to obtain, especially in adulthood. Self-report tools are sometimes used to alleviate the burden of identifying autism. However, commonly used self-report tools to aid in the identification of autistic-like traits in adults have significant limitations that threaten their reliability and validity. The purpose of this study was to develop and assess the psychometric properties of the Autism Spectrum Trait Scale (ASTS), a novel self-report tool that may aid in the identification of autistic-like traits in adults, developed using principal axis factoring procedures.

Methods:

Exploratory factor analysis (n = 764) was conducted to develop the factor structure, and confirmatory factor analysis (n = 761) was performed in a nonoverlapping sample to assess the factor model. Out of the 1525 participants, 507 were autistic and 1018 were not. Of the 507 autistic adults, 177 reported having a formal diagnosis, whereas 330 were self-diagnosed.

Results:

The results showed a stable four-factor model with good model fit (Tucker-Lewis Index [TLI] = 0.92, Comparative Fit Index [CFI] = 0.92, Goodness-of-Fit [GFI] = 0.88, root-mean-square error of approximation [RMSEA] = 0.05), strong internal consistency (α = 0.94), and criterion validity (r = 0.90, p < 0.001). The model demonstrated metric measurement invariance (Δ CFI = −0.008, Δ RMSEA = 0.001), acceptable sensitivity and specificity (area under the curve = 0.92), and the presence of a superordinate factor (TLI = 0.91, CFI = 0.92, GFI = 0.90, RMSEA = 0.05).

Conclusions:

These findings provide preliminary evidence for the use of the ASTS as a component of assessment for autism in adults.

Community Brief

Why is this an important issue?

Many undiagnosed autistic adults face difficulty in getting assessed and diagnosed because the process is often complicated and hard to access. However, getting a diagnosis is important because it reduces stigma and can help autistic individuals better understand themselves, connect with others, and access support and resources that can improve their overall well-being.

What was the purpose of this study?

This study tested the psychometric properties of the Autism Spectrum Trait Scale (ASTS), a new self-report tool developed by the researchers, to identify autistic-like traits in adults and facilitate proper assessment and diagnosis.

What did the researchers do?

The researchers attempted to follow best practices for developing and analyzing new scales. They first conducted an exploratory factor analysis with 764 participants to figure out how the scale should be structured. Then, they conducted a confirmatory factor analysis with a different group of 761 participants to check if the structure they found was accurate. A total of 507 participants had autism, with 177 formally diagnosed and 330 self-reporting their diagnosis. They conducted analyses comparing formal and self-reported diagnoses to better understand the group differences.

What were the results of the study?

The results showed that the new scale had four clear factors, was valid and reliable, and accurately identified autistic-like traits across different groups. In addition, an analysis of the final items revealed an overall factor based on the total score, demonstrating that the new scale effectively measured the construct of autism as a whole.

What do these findings add to what was already known?

There have been ongoing discussions about the need for a reliable and easy-to-use tool. However, there was no existing tool that covered such a wide range of autistic traits and met the highest standards for validity, reliability, and clinical accuracy—until the ASTS.

What are potential weaknesses in the study?

This study included both “self-reported” and “clinically diagnosed” autistic adults. Additional analyses compared the differences between these groups on the new scale and an already established scale for detecting autism in adults. While not all autistic participants were clinically diagnosed, this approach contributes to the literature by highlighting the number of undiagnosed adults who believe they meet diagnostic criteria for autism and offering additional insights into this population.

How will these findings help autistic adults now or in the future?

The new scale is available online for free, and people who take the ASTS can score it themselves using the ASTS scoring interpretation sheet to get results immediately. It can also be downloaded for printing. Undiagnosed autistic adults can use it to identify whether they may have autistic traits, and they can also bring the results to a formal assessment to discuss with a clinician. However, more research is needed before the ASTS can be officially used as part of the diagnostic process without the need for additional components.

Background

Autism

The Diagnostic and Statistical Manual (DSM) formally conceptualizes the characteristics defining autism as two overarching constructs: (1) “persistent deficits in social communication and social interaction across multiple contexts” (Criterion A) and (2) “restricted, repetitive patterns of behavior, interests, or activities” (Criterion B).¹ It is important to note that autistic individuals often do not perceive autism as a disability, deficit, or impairment.^2,3 In fact, 87% of autistic adults prefer identifying as an “autistic adult” (known as identity-first language or IFL) rather than “an adult with autism” (known as person-first language) as IFL is believed to decrease stigma and embrace all aspects of identity.⁴ The Autism and Developmental Disabilities Monitoring Network reported a growing prevalence of autism in the United States, estimating that 1 in 36 children had autism in 2023.⁵ This represents a significant increase from previous estimates, such as the 1 in 500 children reported to have autism in 1999.⁶ Consequently, there are likely many undiagnosed autistic adults.⁷

Undiagnosed autism in adults

Autistic adults may remain undiagnosed or unaware of their condition until later in life due to various factors, including sex assigned at birth, gender, socioeconomic status, barriers to reaching a formal diagnosis, changes in diagnostic criteria, the ability to “mask” or “camouflage” symptoms, and so on.^7–11 For example, research shows that autism is often underdiagnosed in girls, with about 80% remaining undiagnosed by 18.^12,13 Autistic women may experience fewer socio-communication difficulties compared with men¹⁴ possibly due to autistic females developing strategies like mimicking “popular girls” in class and adjusting their behavior to fit social settings.^15–17 These strategies likely contribute to the underdiagnosis of girls and women, as their socialization and ability to mask could lead to their autistic traits being overlooked.¹³ Similarly, researchers have primarily focused on White males in autism studies, which may further explain the gap in recognizing autism in females.^18,19 Gender-diverse individuals are three to six times more likely to be autistic than cisgender individuals and five times more likely to remain undiagnosed.^20–22

Cultural differences, shaped by diagnostic criteria based on Western norms, may further hinder recognition of autism in individuals from non-Western cultures.²³ For example, Black parents of autistic children report fewer concerns about autism symptoms, possibly due to cultural stigma,^24–26 and White children are up to 65% more likely to receive an autism diagnosis than Black or Latinx children, highlighting racial and sociocultural disparities in diagnosis.^24,27 Moreover, research shows that a late autism diagnosis increases the risk of depression and self-harm.²⁸ Adults over 50 who were recently diagnosed reported feelings of isolation, alienation, and confusion about their identity due to a lack of childhood diagnosis.¹¹ While perspectives differ, an autism diagnosis can reduce stigma, improve support, foster acceptance, and enhance self-concept.^29,30 Despite these benefits, studies on best practices for supporting autistic adults, particularly those diagnosed later in life, remain limited.^10,31,32

Complications with the assessment and diagnosis of autism

Autism is a complex neurobiological condition that often requires a comprehensive evaluation for diagnosis, including self-report measures that assess behavioral and developmental factors.^33–35 Professional consensus recommends that qualified clinicians facilitate autism diagnoses.³⁶ However, there is a well-documented shortage of clinicians that conduct autism assessments, such as clinical psychologists and neuropsychologists.^35,37 Other types of clinicians also report being hesitant to diagnose autism due to historical inconsistencies in its definition and an overall limited understanding of what autism is.^37–39 Even after finding a clinician who conducts autism assessments, patients face historically long wait-list periods and high out-of-pocket costs to be assessed.^35,40,41 While self-diagnosis appears to be gaining credibility,^7,42 official assessments remain time-consuming, costly, and hard to acquire.⁴⁴ While clinicians commonly use tools like the Autism Diagnostic Observation Schedule, Second Edition or the Autism Diagnostic Interview-Revised with children,⁴⁵ tools for assessing autistic traits in adults are lacking.^46,47 Delayed diagnoses have also occurred due to doctors’ lack of preparedness and knowledge about autism.⁴⁸ These limitations make it extremely difficult for adults who suspect they may be autistic, as well as for clinicians to accurately assess autism-related traits.^{8,34,46,49–51}

Limitations in current tools used to assess for autism

Few diagnostic tools are available for assessing autism in adults, and those that exist have significant limitations.^6,50,52–54 For example, the Systemizing Quotient (SQ) and Autism Spectrum Quotient (AQ) do not reflect the latest DSM-5 (Text Revision) (DSM-5-TR) criteria, have validity issues, and were not developed using factor analytic procedures.^55,56 The AQ may not distinguish autism from other neurodevelopmental conditions.⁵⁷ Other self-report scales, such as the Ritvo Autism Asperger Diagnostic Scale revised (RAADS-R) and Ritvo Autism Asperger Diagnostic Scale shortened version (RAADS-14), omit behaviors that are common in autistic adults with lower support needs.⁵⁸ The Adult Repetitive Behaviors Questionnaire-2 (RBQ-2A-R)¹⁰² and Adult Social Behavior Questionnaire (ASBQ)^58,59 were not originally tested via confirmatory factor analysis (CFA)⁴⁶ and also focus too heavily on repetitive behaviors, neglecting social challenges. Finally, the Comprehensive Autistic Trait Inventory (CATI) lacks a large enough sample to test measurement invariance or confirm their model fit using CFA.⁶⁰ To improve quality-of-life outcomes and support accurate diagnoses, it is essential to develop a better assessment tool for undiagnosed autistic adults.^{6,7,32,46,50,52–54,61,62}

The need for a novel self-report measure for detecting autism in adults

Currently, no self-report assessment exists that evaluates autistic traits in adults while encompassing the most recent diagnostic criteria and meeting the highest standards for validity, reliability, and clinical accuracy.⁶³ Asperger’s syndrome, characterized by “social deficits and restricted interests of the type seen in autism, but, in contrast to autism, relative preservation of language and cognitive abilities—at least early in life,” is now included under the umbrella of autism.^64,65 This change has evoked controversy and identity confusion, especially for individuals previously diagnosed with Asperger’s syndrome.^66–68 Moreover, sensory processing differences in autistic individuals have been long overlooked and are typically neglected in self-report measures.¹ Since these perceptual differences significantly impact how individuals experience the world, they should be carefully considered in diagnostic assessments.⁶⁹ Therefore, a new self-report scale that incorporates traits associated with both Asperger’s and autism, including sensory and perceptual differences, is essential to effectively assess and detect autism in adults.^66–68

The current study

Psychometric studies have outlined the critical steps for developing high-quality measures,^70–72 yet flawed assessments continue to plague the field.^71–75 To address these shortcomings while developing the Autism Spectrum Trait Scale (ASTS), we (1) integrated modern diagnostic content and adaptive items, including sensory processing integration; (2) utilized the updated DSM criteria and subscales; (3) conducted an informal review with four autistic adults to gather feedback and critique the items; (4) used reversed questioning to reduce validity issues and personal bias in self-reporting; and (5) tested the scale’s factor structure to ensure it functions as a psychometrically sound self-report instrument. Both exploratory factor analysis (EFA) and CFA are considered “best practice” for scale development and psychometric research, as both are necessary to ensure that the hypothesized factor structure from EFA fits the observed data.⁷⁶ Therefore, we conducted EFA and CFA on nonoverlapping samples while adhering to preferred extraction methods for scale development.⁷⁶ This study is an important first step in testing a self-report instrument that aims to identify autistic-like traits and undiagnosed autism in adults.

Methods

Item development

The first (S.M.A.) and last (A.M.P.) authors developed the items on the ASTS. A.M.P. is an experienced neuropsychologist specializing in scale development and assessments of neurodivergent populations. S.M.A, a clinical psychology PhD student, has over 3 years of experience working with autistic individuals and aims to focus her research and career on autism, attention-deficit/hyperactivity disorder (ADHD), and related neurobiological conditions. To create the new items, the authors reviewed existing adult autism scales, including the RAADS-R, and used these insights to design new items that accurately reflected the underlying latent constructs. Next, the authors examined the DSM-5-TR factors and subscales for autism, focusing on Criteria A and B and their subcategories.¹ To ensure representation of the hierarchical criteria and their subcategories, as well as to establish content validity, approximately seven items were developed for each of the seven subscales (1–3 for Criterion A and 1–4 for Criterion B). This approach was used so that the items comprehensively reflected the diagnostic constructs and examples as outlined, capturing the full scope of the targeted domains. Novel items were designed to reflect each subcategory’s diagnostic criteria. Given that autism manifests in early childhood, we prioritized “indirect” and “early life” questioning to minimize response bias inherent in self-report, by including at least two indirect or early life questions per subscale. This approach improves on existing scales, which often neglect these critical features.⁷⁹

Inclusion of autistic adults

Although none of the authors are autistic, the first and last authors compiled the final set of items and sought feedback from four autistic adults (three male and one female-identifying). Research suggests that “autistic adults should be considered autism experts and involved as partners in autism research.”⁸⁰ Iteratively, the autistic adults reviewed the original 52 items in an informal evaluation to assess face validity, content validity, item clarity, and scale comprehensiveness. They suggested revisions to the wording to enhance item clarity. While a formal evaluation using Delphi methods would have been ideal,⁷² resource constraints limited the review to this informal process.^81,82

Participants

Recruitment

Participants were recruited through ResearchMatch and social media, with most participants from ResearchMatch.⁸³ While there were no direct benefits for participating, volunteers were entered into a random drawing for five $5.00 Amazon gift cards, distributed electronically.

Total sample

The dataset included 1525 participants, aged 18 and older. The sample was randomly split to allow for independent EFA (n = 764) and CFA (n = 761) analyses (see Table 1).^7,71,72 Additional analyses included comparisons between autistic (n = 507) and non-autistic groups (n = 1018) (see Table 2), demographic comparisons between formally (n = 177) and self-diagnosed (n = 330) autistic adults (see Supplementary Table S1), performance evaluations on the ASTS across under-represented groups and levels of education (see Supplementary Table S2), and an age distribution of the total sample (Supplementary Fig. S3).

Table 1.

Demographics of EFA and CFA Samples

	EFA (n = 764)		CFA (n = 761)
Variables	Mean	SD	Mean	SD
Age	M = 48.20	SD = 17.69	M = 48.65	SD = 17.67

	N	%	N	%
Gender
Male	231	30.20	230	30.20
Female	485	63.50	477	62.70
Non-binary/Third	48	6.30	50	6.60
Prefer not to say	0	0.00	4	0.50
Race/Ethnicity
American Indian/Alaskan Native	3	0.40	5	0.70
African American	21	2.70	17	2.20
Asian	14	1.80	13	1.70
Hispanic/Latino	19	2.50	22	2.90
Pacific Islander	1	0.10	1	0.10
White	686	89.80	668	87.80
Other	20	2.60	35	4.60
Education
Some high school	5	0.70	4	0.50
High school/GED	24	3.10	22	2.90
Some college credit	108	14.10	116	15.20
Associate’s degree	61	8.00	45	5.90
Bachelor’s degree	272	35.60	269	35.30
Master’s degree	207	27.10	230	30.20
Doctoral degree	87	11.40	75	9.90
Diagnosis of autism
Yes	240	31.40	267	35.10
No	524	68.60	494	64.90
Who made the autism diagnosis
Self	164	68.30	166	62.20
Psychiatrist	27	11.30	44	16.50
Primary care physician	6	2.50	4	1.50
Psychologist	33	13.80	45	16.90
Social worker	9	3.80	3	1.10
Nurse practitioner	1	0.40	5	1.80
Co-occurring diagnosis
Autism
Yes	224	93.30	244	91.40
No	16	6.70	23	8.60
Non-autism
Yes	351	67.00	322	65.20
No	173	33.00	172	34.80

CFA, confirmatory factor analysis; EFA, exploratory factor analysis; SD, standard deviation.

Table 2.

Demographics of Autism Status (N = 1525)

	Autistic participants (n = 507)		Non-Autistic participants (n = 1018)
Variables	Mean	SD	Mean	SD	p-Values
Age	41.45	16.06	51.90	17.42	0.000

	N	%	N	%
Gender
Male	129	25.44	332	32.61	<0.001
Female	306	60.36	656	64.44
Non-binary/Third	69	13.61	29	2.85
Prefer not to say	3	0.59	1	0.10
Race/Ethnicity
American Indian/Alaskan Native	5	0.99	3	0.29	0.009
African American	10	1.97	28	2.75
Asian American	9	1.78	18	1.77
Hispanic/Latino	18	3.55	23	2.26
Pacific Islander	1	0.20	1	0.10
White	435	85.80	919	90.28
Other	29	5.72	26	2.55
Education
Some high school	5	0.99	4	0.39	<0.001
High School/GED	20	3.94	26	2.55
Some college credit	100	19.72	124	12.18
Associate’s degree	44	8.68	62	6.09
Bachelor’s degree	198	39.05	343	33.69
Master’s degree	111	21.89	326	32.02
Doctoral degree	29	5.72	133	13.06

Measures

The study was delivered via the Qualtrics Experience Management (XM) Platform,⁸⁴ including an informed consent form, a demographic questionnaire, three measures, and a debriefing form. The three measures were ASTS, RAADS-14, and the Pictures of Facial Affect (POFA). Although POFA⁸⁵ data were collected, only ASTS and RAADS-14 data were used in the current study. The RAADS-14, a shortened version of RAADS-R,^81,82 consists of 14 items with one reverse-coded item. Participants rated items on a 4-point Likert scale ranging from 0 (never true) to 3 (true now and when I was young), with higher ratings indicating more autistic traits.

The Autism Spectrum Trait Scale

The ASTS initially had 52 items, including 14 reverse-coded and 7 early/indirect questioning items (see Supplementary Table S3). After item deletion during EFA, the final version of the ASTS contains 29 items with 3 reverse-coded items and 5 early/indirect questioning items (see Table 3). Participants rated items on a 4-point Likert scale ranging from 1 (false) to 4 (true), with higher ratings indicating higher levels of autistic traits.

Table 3.

The Autism Spectrum Trait Scale Items Grouped Per Factor

		Factor loadings
Item	Question	F1	F2	F3	F4
Factor 1: Social-emotional reciprocity challenges
Q1	I like to repeat words or phrases that I’ve previously heard during conversations.	0.43
Q4	I often take things people say literally, and I have been told I am misunderstanding the point of what they are saying.	0.76
Q6	People sometimes tell me that I talk too loudly or softly, but I don’t notice myself doing it.	0.51
Q10	People have commented that I have difficulty maintaining eye contact.	0.51
Q11	When I was a teenager, some people commented that I talk in a strange way.	0.67
Q12	I have been told that my facial expressions don’t match how I really feel.	0.68
Q18	It is difficult for me to KNOW when someone says one thing but actually means something else.	0.58
Q26	When I was a teenager, I had the tendency to flip conversations back on to myself, or to topics I was interested in.	0.46
Q28	I have been told I am very literal in my thinking.	0.62
Q29	I have trouble understanding why people use expressions such as “break a leg,” “bite the bullet,” “stabbed in the back.”	0.75
Q34	I find myself staring at objects such as fans and lights.	0.49
Q42	When I was young, I often engaged in repetitive behavior such as rocking OR hand-flapping.	0.41
Q17R	When talking to someone, I have no trouble telling when it is my turn to talk or to listen.	0.40
Factor 2: Hyper- or hyporeactivity to sensory input
Q35	I tend to notice details in patterns of ordinary objects.		0.33
Q36	From a very young age, I have been sensitive to certain sounds.		0.86
Q37	I feel uncomfortable wearing certain clothing.		0.65
Q40	I try to avoid certain foods because of their texture.		0.45
Q41	I often feel uneasy when I am in a place where there are many smells, noises, or bright lights.		0.79
Q44	It is difficult for me to tolerate things I dislike (like smells, textures, sounds, or colors).		0.79
Factor 3: Developing, maintaining, and understanding relationships
Q27	It takes a lot of effort for me to make friends.			0.69
Q30	When I was young, I had few close friends.			0.41
Q31	I have a hard time forming romantic relationships or getting past the first date.			0.59
Q45	It can be difficult for me to work in a group for projects. I often prefer to work alone.			0.46
Q15R	I can easily chat and make small talk with strangers.			0.65
Q49R	It is easy for me to ask potential partners on dates.			0.59
Factor 4: Highly fixated interests
Q47	I like collecting items on certain topics, but many people tend to show little interest in my collection(s).				0.82
Q50	I like to collect information about certain topics like cars, birds, trains, computers, TV shows, etc.				0.76
Q51	I have special hobbies that only a few people have.				0.67
Q52	I enjoy collecting special or “rare” items.				0.80

The factor loading that is bolded indicates the highest factor loading for each factor.

Overview of procedures: Lay summary of statistical analyses

We used principal axis factoring (PAF) for EFA to identify patterns in the data and conducted CFA to confirm our four-factor model. We also tested a single-factor model to ensure that the items contributed to an overall score and assessed whether the model worked equally well across groups. In addition, we measured the tool’s accuracy in identifying autistic adults and developed a version of the ASTS for public use (see Supplementary Description S1 for technical summary of statistical analyses).

Procedures

Data cleaning

The dataset originally included 1883 participants. We removed 24 individuals who did not provide consent, 9 people who were under the legal age of consent, and 176 participants for incomplete participation or completing the study in under 10 minutes. Following Schober and colleagues (2021), we excluded 72 participants whose response patterns (i.e., endorsing “4” on all items) suggested they did not respond meaningfully.⁸⁶ In addition, 18 multivariate outliers were identified and removed using Penny’s methodology.⁸⁷ The final dataset of 1525 participants was randomly split into two independent groups for EFA (n = 764) and CFA (n = 761).

Exploratory factor analysis

Extraction Method

PAF accounts for measurement error and identifies items with shared systematic variation⁷²; it is often considered the “preferred” method of factor extraction and was therefore applied in this study.^59,71,77,78

Testing Assumptions

We used the Statistical Package for Social Sciences (SPSS) Version 28 software to conduct EFA.⁸⁸ We tested all assumptions, including the use of metric variables, sufficient sample size, independence of errors, linearity, absence of outliers, underlying factor structure, no extreme multicollinearity, and homogeneous intercorrelations by subgroups. No assumptions were violated, so we proceeded with the factor retention procedures.

Factor Retention Procedures

Four statistical tests determined the number of factors to retain for the EFA, including: (1) the latent root criterion (eigenvalues ≥1.00), (2) percent of total variance (factors ≥60%), (3) individual factor contribution (factors ≥5.00%), and (4) scree plot (visual inflection point). We used Promax rotation to allow the factors to correlate. Next, we iteratively deleted “bad” items that did not load on any factor or that cross-loaded on multiple factors above 0.30.⁷¹ Items with high cross-loadings (>0.30) may not effectively discriminate between factors, as they load onto more than one factor. Therefore, we removed these items to ensure each factor accurately represented its intended construct.⁸⁹

Confirmatory factor analysis

Preparation for Conducting CFA

We used Analysis of Moment Structures [AMOS] to conduct the CFA.⁹⁰ Before transferring the dataset from SPSS, we deleted cases listwise because AMOS does not allow analyses with missing data. This resulted in removing 59 participants, leaving 761 of the original 820 participants in the final CFA. We tested all assumptions, including calculating the number of pieces of information inputted into AMOS, estimating the number of parameters, and determining the degrees of freedom. Furthermore, we tested for adequate sample size and local and overall model identification assumptions. No assumptions were violated, so we proceeded with the analysis, and set the first item’s path loading on each factor to 1.00.

Testing Model Fit

We conducted a CFA using maximum likelihood estimation and standardized regression coefficients by delineating the factor structure found during EFA procedures. We assessed construct validity through convergent and discriminant validity analyses. Convergent validity relied on average variance extracted (AVE) values. We obtained AVE values by calculating the average of the squared standardized regression weights for each factor.⁹¹ We deemed discriminant validity adequate if the AVE value was higher than the squared correlation value.

Testing Higher-Order Models

The authors theorized that the factors might relate to a higher order construct associated with either Criteria A or Criteria B for autism in the DSM-5-TR or alternatively, that all factors might load onto one superordinate factor, indicating a general autism factor.^1,92 To test this, we ran a model where each factor loaded onto one of the two second-order latent constructs as follows: either Criterion A (Factors 1 and 3) or Criterion B (Factors 2 and 4) allowing them to correlate. We tested discriminant validity by comparing the higher order model’s AVE values to the squared correlation. Regardless of the model fit for the DSM-5-TR model (Criteria A and B), we also tested the superordinate autism factor.

Testing Measurement Invariance

To test for measurement invariance, we ran two baseline models in AMOS using multiple groups confirmatory factor analysis (see Table 4).⁹³ Model 1 included individuals with and without autism (unconstrained baseline model). Model 2 constrained all of the paths to set values and tested configural (weak) invariance, which indicates that items load onto their intended factors. Model 3 added additional constraints to covariances and tested metric (moderate) invariance, confirming that items contributed to the latent construct (aka factor) similarly across groups. Model 4 tested scalar (strong) invariance by constraining the variance of the error terms, covariances, and paths. We assessed model fit using the chi-square (χ²), CFI, and root-mean-square error of approximation (RMSEA), with CFI ≥0.90 and RMSEA ≤0.05 indicating good fit.^93,94 We compared changes (Δ) in RMSEA and CFI to the cutoff values (Δ CFI ≥−0.01 and Δ RMSEA ≥0.015) indicating that the model fit got significantly worse with additional constraints.^93,94

Table 4.

Measurement Invariance Fit Statistics

Model	Fit indices					Comparison	Δ RMSEA	Δ CFI
Model	χ ²	dfs	p	RMSEA	CFI	Comparison	Δ RMSEA	Δ CFI
AUTISM	761.786	371	0.000	0.063	0.805
Non-AUTISM	907.010	371	0.000	0.054	0.871
Model 1	1669.101	742	0.000	0.041	0.850
Model 2	1735.083	767	0.000	0.041	0.843	2 vs. 1	0.000	−0.007
Model 3	1798.411	777	0.000	0.042	0.835	3 vs. 2	0.001	−0.008
Model 4	2285.594	806	0.000	0.049	0.760	4 vs. 3	0.007	−0.075

The criteria cutoffs are as follows: Δ CFI ≥−0.01 and Δ RMSEA ≥0.015 indicate a lack of invariance. Model 1 (unconstrained), Model 2 (measurement weights), Model 3 (structural covariances), Model 4 (measurement residuals).

χ², chi-square; dfs, degrees of freedom; RMSEA, root-mean-square error of approximation.

Testing Internal Consistency and Correlation Between Measures

We assessed internal consistency (reliability) using Cronbach’s α. To evaluate criterion validity, we conducted a bivariate correlation between the total score of the RAADS-14 and the ASTS, obtaining the Pearson correlation coefficient.⁹⁵

Testing Whether the ASTS Can Detect Autism

We calculated the overall mean and standard deviation (SD) for the total ASTS scores and each factor to assess its ability to detect autism in adults. t-Tests compared scores between autistic and non-autistic participants to determine if autistic participants scored significantly higher on each factor and the overall ASTS score. While t-tests help to establish group differences among autistic and non-autistic adults, further analyses were conducted to establish the diagnostic potential of ASTS. To do this, we established the sensitivity and specificity of the ASTS by examining receiver operating characteristic (ROC) curves and area under the curve (AUC) scores for the total 29 items and each of the four factors. To identify a visual threshold for elevated autistic-like traits, we analyzed the ROC curve by finding the point closest to the top left corner and its corresponding threshold value. For statistical thresholds, we calculated the mean and SD of the total and factor scores of the participants in the CFA sample.

Results

Exploratory factor analysis

EFA descriptive statistics

The participants in the EFA sample (n = 764) ranged in age from 18 to 93, with a mean age of 48.20 years. The majority of the participants in the EFA sample were female (63.50%) and White (89.80%). Most of the participants had a degree in higher education, with either a Bachelor’s degree (35.60%) or a Master’s degree (27.10%; Table 1). Approximately one-third of the participants indicated being diagnosed with autism (31.40%), and the majority of those diagnosed indicated having a self-reported diagnosis (68.30%; Table 1). We conducted preliminary exploratory analyses of demographic differences based on autism status. A significant difference in gender distribution was found between autistic and non-autistic populations (p < 0.001; see Table 2); racial composition (p = 0.009), education level (p < 0.001), and age (p < 0.001) also significantly differed, with non-autistic participants being about 10 years older.

Factor loadings

The latent root criterion (1.02) and the percent of total variance (60.53%) supported a nine-factor solution; however, the individual factor contribution statistic (6.88%) and the scree plot both suggested a two-factor solution. When discrepancies arise in determining factors retention, researchers rely on theory.⁹⁹ Given autism’s complexity and item development targeting Criteria A and B (including seven subscales), we evaluated the nine-factor solution. During item retention procedures, 23 of the initial 52 items were deleted, resulting in a 29-item scale. As we removed items, the eigenvalues indicated retaining fewer factors, reducing the nine-factor solution into a four-factor solution.

Description of the factors

The final EFA produced a four-factor solution (Table 3). We named the factors based on a comprehensive understanding of their corresponding items as follows: Social-Emotional Reciprocity Challenges (13 items); Hyper- or Hyporeactivity to Sensory Input (6 items); Developing, Maintaining, or Understanding Relationships (6 items); and Highly Fixated Interests (4 items).

Confirmatory factor analysis

CFA descriptive statistics

The CFA sample (n = 761) ranged in age from 18 to 89 years (mean [M] = 48.65). Most participants were female (62.70%), White (87.80%), and had a degree in higher education (35.30% Bachelor’s, 30.20% Master’s; Table 1). Approximately one-third (35.10%) reported having an autism diagnosis, and the majority of those diagnoses were self-reported (62.20%; see Supplementary Table S1).

Model fit indices

The model converged with a significant chi-square (χ² [371] = 1200.10, p < 0.001; see Supplementary Fig. S1). The TLI and CFI were each 0.92, indicating good model fit, and the GFI was 0.88, just slightly below the threshold. RMSEA was 0.05, further confirming a good model fit.⁹⁴ Given the preponderance of good fit statistics, the four-factor model was deemed acceptable.¹⁰⁰ All item loadings were significant on their intended factors, supporting the hypothesized structure.

Construct Validity

The AVE values of Factors 1–4 were 0.39, 0.52, 0.45, and 0.60, respectively. A value of greater than or equal to 0.50 indicates good convergent validity⁹¹; by this metric, Factors 2 and 4 had good convergent validity (Supplementary Table S4). For discriminant validity, the squared correlation value for Factors 1 and 2 was 0.64, Factors 1 and 3 was 0.57, Factors 1 and 4 was 0.61, Factors 2 and 3 was 0.48, Factors 2 and 4 was 0.49, and for Factors 3 and 4 was 0.37 (Supplementary Table S4). These results indicate that the discriminant validity of the factors was not ideal. However, this was not necessarily surprising, as we used a Promax rotation—a type of oblique method—because we expected the components to correlate under “autism” as a construct.

Hierarchical testing

The DSM Criteria A and B model fit

The fit statistics showed that the higher-order construct for Criteria A (Factors 1 and 3) or Criteria B (Factors 2 and 4) in the DSM-5-TR had a good fit. The squared correlation value was 0.97, and the AVE values for Criteria A and Criteria B were 0.77 and 0.70, respectively. The comparison of AVE values to the squared correlation indicated poor discriminant validity. The strong intercorrelation suggests that a single higher order factor, representing an overarching “autism” factor, may be more theoretically appropriate than two separate factors for Criteria A and B.

Overarching autism factor model fit

The superordinate factor model converged with a significant chi-square (χ² [373] = 1207.35, p < 0.001). The TLI and CFI had values of 0.91 and 0.92, respectively, and the GFI and RMSEA had values of 0.90 and 0.05, respectively. With all fit statistics indicating a good model fit, we concluded that the model was adequate. This also supports using a total score for the ASTS rather than relying on the subscales independently, as reflected in the scoring instructions in Supplementary Data.

Measurement invariance

We established configural invariance by comparing Model 2 and Model 1, with a Δ CFI of −0.007 and Δ RMSEA of 0.000 (Table 4). We then established metric invariance by comparing Model 3 with Model 2, with a Δ CFI of −0.008 and Δ RMSEA of 0.001. When comparing Model 4 with Model 3, we found a Δ CFI of −0.075 and Δ RMSEA of 0.007. Since the Δ CFI exceeded the cutoff value of Δ CFI ≥−0.01, scalar invariance was not supported. However, we did find evidence for metric invariance, suggesting that the ASTS functions comparably across autism and non-autism groups (Table 4).^93,94,101

Internal consistency and correlation between measures

Both scales demonstrated excellent internal consistency (Cronbach’s α), with values of 0.91 for the RAADS-14 composite score and 0.94 for the ASTS composite score in both the EFA and CFA samples. A strong positive correlation was found among the scales, (r = 0.90, p < 0.001; see Supplementary Fig. S2), which confirmed that the ASTS has criterion validity with an existing self-report measure of autism.

Determining the detection of autism using the ASTS

When comparing the total scores of all participants in the CFA dataset on the ASTS, autistic participants scored significantly higher than those without autism on all four factors, as well as on the total score (see Table 5).

Table 5.

t-Tests to Determine the Detection of Autism Using the ASTS

		All		Non-AUTISM		AUTISM
	Range	Mean	SD	Mean	SD	Mean	SD	t-Value	dfs	p-Value
Total	26–116	67.85	19.04	58.11	14.03	85.89	12.95	26.77	759	<0.001
F1	13–52	26.80	8.60	22.53	6.14	34.69	6.72	25.20	759	<0.001
F2	6–24	15.56	5.17	13.40	4.42	19.56	3.94	19.01	759	<0.001
F3	6–24	16.07	4.59	14.76	4.22	20.03	3.00	18.06	759	<0.001
F4	4–16	8.89	3.67	7.41	3.00	11.63	3.20	18.12	759	<0.001

The “range” refers to the entire possible range of scores that participants could have obtained. The t value represents the difference between the AUTISM and non-AUTISM group.

ASTS, Autism Spectrum Trait Scale.

Sensitivity, specificity, and hypothetical diagnostic “thresholds”

An ROC curve comparing the total 29-item scores of autistic participants (n = 267) and non-autistic participants (n = 494) yielded an AUC score of 0.92. ROC curves for the individual factors yielded AUC scores ranging from 0.82 to 0.92 (Table 6). The RAADS-14 total scores for both groups yielded an AUC score of 0.91. We calculated the hypothetical threshold for elevated autistic-like traits at a total score of 69.50 and applied the same procedure to each factor (Table 4). We defined score ranges based on how many SDs participants’ scores deviated from the mean: average (±1 SD), low (below −1 SD), and high (above +1 SD) from the mean. An average score does not necessarily indicate that an individual has autism, but rather that the score is in the average range. The ASTS, scoring instructions, and interpretation are available in Supplementary Data.

Table 6.

Sensitivity, Specificity, AUC, and Diagnostic Thresholds

	Range	# of items	AUC	Y axis	X axis	Threshold	Mean	SD	Low	Medium	High
Total	29–116	29	0.92	0.89	0.21	69.50	67.85	19.04	29–48	49–87	88–116
F1	13–52	13	0.90	0.81	0.18	28.50	26.80	8.60	13–17	18–35	36–52
F2	6–24	6	0.84	0.75	0.18	17.50	15.56	5.17	6–9	10–21	22–24
F3	6–24	6	0.84	0.75	0.25	17.50	16.61	4.59	6–11	12–21	22–24
F4	4–16	4	0.82	0.75	0.25	9.50	8.89	3.67	4	5–13	14–16

The Y axis represents sensitivity, and the X axis represents 1.00 minus specificity. The low–high levels were determined based on how many SDs the scores deviated from the mean. The SD is a measure of variation in scores.

AUC, area under the curve.

Comparison of formally diagnosed and self-reported autistic adults

An independent samples t-test showed that formally diagnosed autistic adults (M = 91.12, SD = 11.12) scored significantly higher on the ASTS than self-diagnosed autistic adults (M = 83.94, SD = 12.74), p < 0.001 (Supplementary Table S1). Similarly, formally diagnosed autistic adults (M = 32.47, SD = 6.70) scored significantly higher on the RAADS than self-diagnosed autistic adults (M = 27.66, SD = 8.08), p < 0.001. Demographic analyses for gender, race/ethnicity, and education showed no significant differences between the groups, but age was higher in self-diagnosed individuals (M = 42.33, SD = 16.25) compared with those with formal diagnoses (M = 39.81, SD = 15.62), p = 0.09 (Table 6).

Discussion

Overview of findings

This study aimed to evaluate preliminary evidence for ASTS, a novel self-report tool. The ASTS may serve as an initial self-evaluation for undiagnosed autistic adults who suspect they are autistic or eventually as one of several assessment tools used by clinicians to help identify autistic-like traits in adults who may qualify for a diagnosis. If the ASTS reveals elevated autism-related traits, it may suggest the need for further assessment. While the ASTS shows promise, further studies are needed before it can be used in clinical settings as part of an official diagnostic assessment for identifying autism in adulthood without additional assessment methods. The study aimed to reduce barriers in adult autism assessment by developing a psychometrically sound self-report tool to help identify autistic-like traits while addressing the limitations of existing scales. Unlike the ASBQ and CATI, which lack CFA confirmation and use the less-preferred principal component analysis (PCA) method, we used PAF to accurately identify shared variance among items.^58,60 The ASTS also reflects modern diagnostic criteria, unlike scales such as the RBQ-2A-R, AQ, and SQ.^55,56,59,102

We tested a nine-factor solution, which was refined into a four-factor, 29-item scale with good model fit. To assess construct validity, we conducted convergent and discriminant validity analyses. Convergent validity was good for half of the factors, although some factors showed less-than-ideal discriminant validity, which is expected given the scale’s focus on a single construct (autism) and the presence of a single superordinate factor with a good model fit. In the CFA sample (n = 761), the ASTS showed internal consistency comparable to RAADS-14 and a strong positive correlation with it. While these findings are specific to our sample, they remain notable. Although the RAADS-14 is widely used, it does not align with the DSM-5-TR criteria, making the ASTS potentially more useful to clinicians. Sensitivity and specificity analyses showed that the ASTS effectively identified autistic participants, yielding a slightly higher AUC score than the RAADS-14. These results, along with the superordinate factor analysis, support the ASTS as a potentially valuable self-report tool for aiding clinicians in identifying of autistic-like traits.

Diagnostic “Thresholds”

We determined that using a single diagnostic “threshold” to assess autistic traits was less appropriate than establishing a range of scores indicating varying levels of characteristics—low, medium, or high (Table 6). We calculated the factor ranges to provide deeper insights into the components of the ASTS. Given autism’s multifaceted nature, encompassing a wide range of characteristics, individual factors alone should not identify autistic traits. Instead, we recommend using the total ASTS score for diagnostic purposes by qualified clinicians, once further validation in other samples occurs. Nonetheless, the factor scores offer valuable insight into specific areas that may require support.

Scoring differences based on self-diagnosis versus formal diagnosis

Mean testing showed that autistic adults scored significantly higher than non-autistic adults on both total and individual factor scores of the ASTS. Analyses comparing self-diagnosed and formally diagnosed autism revealed that formally diagnosed autistic adults scored higher on both the ASTS and the RAADS-14 (Supplementary Table S1). This difference likely reflects more pronounced autistic traits in formally diagnosed individuals. For self-diagnosed adults, these findings highlight the large number of underdiagnosed individuals, particularly those who display fewer observable traits or those who mask. However, this does not invalidate their autistic identity. Another possible explanation for the higher scores in the formally diagnosed group is inaccurate self-reporting among some self-diagnosed participants, which could have skewed the results and created the appearance of lower scores in that group as a whole, even though many may share similar levels of traits with formally diagnosed adults.

Self-diagnosed individuals tended to be older than those with formal diagnoses. This may be due to older generations missing diagnoses because of outdated criteria or the stigma surrounding autism in the past. In addition, the likelihood of receiving a diagnosis decreases with age, as noted in the literature, such as by McDonald (2020),⁷ which highlights the “lost generation” of autistic individuals overlooked in their earlier years.^7,19,103 While the age difference of about 3 years is small, it aligns with these broader trends.^7–11 Upon conducting preliminary statistics to examine demographic differences based on autism status, there were more non-binary participants in the autistic group, aligning with the literature indicating that autism is associated with gender diversity.^20–22 In addition, non-autistic participants were more likely to hold master’s or doctoral degrees, which may stem from autistic adults, particularly those undiagnosed, lacking access to the support and resources that facilitate higher education.^96–98

Item development procedures

A key strength of the ASTS item development process was its alignment with diagnostic criteria, an update not commonly seen in other scales. However, some of the criteria and item wording could have been better represented, which is a significant limitation. Clarity and precision in item development are crucial to ensure understanding of the underlying concept represented by an item. While indirect and reversed questions were included to reduce response bias, many were removed during the EFA due to wording issues. Likewise, although the ASTS included items addressing sensory differences, it lacked items on hyporeactivity to sensory input. Given the diverse nature of autism, where some individuals experience hyporeactivity rather than hyperreactivity, this omission is notable.¹⁰⁴ This shortcoming may highlight the need for more extensive iterative feedback.⁷² In future revisions or adaptations of the ASTS—for example, for autistic adults requiring higher support levels—formal evaluations involving autistic experts are essential. In addition, feedback from autistic laypeople would help distinguish between content validity and face validity.¹⁰⁵

Limitations

The target population of autistic adults

The target population for the ASTS aimed to include a broad range of autistic adults, but the final sample mainly consisted of individuals with lower to moderate support needs. This outcome may have occurred due to recruitment via ResearchMatch and the need for participants to have sufficient comprehension to complete the items, resulting in a sample likely skewed toward the higher functioning end of the autism spectrum.

We conducted a reading level analysis of the original 52 items using Microsoft Word, which revealed a Flesch–Kincaid Grade Level of 6.3. This finding suggests that autistic adults with higher support needs, such as nonspeaking individuals or those with co-occurring intellectual disabilities, likely did not participate in the study. The high education level among participants further indicates that the sample skewed toward individuals on the higher masking end of the autism spectrum. Future research should adapt the ASTS for use with caregivers of individuals with higher support needs to increase its inclusivity. Consulting a plain language expert would also improve accessibility, especially considering the higher prevalence of intellectual disabilities among autistic individuals compared with the general population.¹⁰⁶

This study has several other limitations, including that most participants self-reported having autism. Before starting the ASTS research, individuals were asked, “Are you currently diagnosed with autism or suspect you might have it?” Those who answered “yes” specified whether the diagnosis was self-made or from a clinician. In the EFA sample, 68.3% self-reported compared with 62.2% in the CFA sample.

Excluding self-reported autism would have made the sample size too small to conduct analyses with the highest psychometric standards. The prevalence of self-reported diagnoses underscores the need for reliable tools for undiagnosed autistic adults, a growing group likely seeking autism diagnoses.^7,32 While self-reported cases cannot be verified, this approach enabled the collection of a larger sample than clinical data alone would allow. This study represents a key step in developing the ASTS as an assessment tool for future research. Future studies should include sub-analyses that use only formally diagnosed cases to further validate diagnostic utility.

Constraints on generality

One limitation is using ResearchMatch as the primary recruitment method, which primarily includes individuals from the United States, resulting in limited cultural and regional diversity. The sample was also predominantly White women, despite autism being reported as 3.8 times more prevalent in males. However, this disparity may reflect the underdiagnosis of females in childhood rather than a male-to-female ratio of 4:1.¹³ In fact, recent research suggests that this ratio is more likely closer to 4:3.¹³ Regardless, it is important to recognize that there are serious mental health implications for women who remain undiagnosed, underscoring the need for improved diagnostic practices and recognition of autistic women.⁹

Sample size constraints limited our analysis to two-group comparisons (Supplementary Table S2). Despite this, the findings reveal valuable insights. Minority groups—including non-White participants, individuals with non-female gender identities, and those without a college degree—consistently scored higher than the majority sample. These results are significant given that non-White groups and individuals with lower education levels often face more barriers to diagnosis, treatment, and care, which may influence how autism manifests within these populations.

Implicit bias in question wording may influence responses across demographic groups. For example, non-White participants may struggle to form friendships for reasons unrelated to autism, such as racial discrimination (e.g., Question 27: “It takes a lot of effort for me to make friends”). This highlights the limited generalizability of diagnostic criteria developed using Western norms, which may overlook cultural differences in behavior and social interaction, leading to underdiagnosis in non-Western populations.^24,27

Future directions

Due to the homogenous nature of the sample in this study, the psychometric properties and norms of the ASTS may not be generalizable to other groups within the United States, including males and minority populations. Although preliminary statistical analyses were conducted, future research using the ASTS should focus on its utility for individuals from diverse and underrepresented backgrounds, including gender-diverse individuals, and ensure representation across various racial, educational, and sociodemographic groups.

Gathering more data on the ASTS and evaluating its performance among underrepresented groups are essential for broader application. The limited diversity in the current sample restricted comprehensive exploratory analyses across demographics. Future studies should examine score differences among diverse groups, considering both gender and sex assigned at birth, as autistic individuals are more likely to report gender identities differing from their sex assigned at birth.²² In addition, item-level analyses should identify potential biases in specific questions. Detailed demographic data would enable research into scoring patterns across racial groups and education levels.¹⁰⁷ Such data could offer valuable insights into the unique symptoms and challenges faced by gender-diverse autistic individuals.

Future research should assess test–retest reliability, retest criterion validity using the final 29 ASTS items, and examine correlations between the four factors with neurocognitive markers, such as POFA by Ekman and Friesen.⁸⁵ The deletion of 59 participants with missing data during CFAs, due to AMOS software limitations, highlights the need to use software capable of handling missing data, allowing for the inclusion of incomplete cases.¹⁰⁸ Another option is data replacement, which was not utilized in this study due to the early stage of the ASTS, but could be explored in future research to strengthen findings.

Future studies should further validate the ASTS by incorporating additional measures, like RAADS-14, and comparing it to other constructs, enhancing its robustness against other instruments assessing autism in adulthood. Embedding the ASTS items into a broader psychological questionnaire could also mitigate acquiescent response styles and provide deeper insights into its performance in a more heterogeneous context.

Conclusion

The study provides preliminary evidence that the 29-item ASTS is a reliable and valid self-report measure for identifying autistic traits in adults. The ASTS demonstrated a stable four-factor structure with good model fit, verified through EFA and CFA (TLI = 0.92, CFI = 0.92, GFI = 0.88, RMSEA = 0.05). The factors showed promising psychometric properties, including high internal consistency (α = 0.94), criterion validity (r = 0.90, p < 0.001), metric measurement invariance across diagnostic groups (Δ CFI = −0.008, Δ RMSEA = 0.001), adequate sensitivity and specificity (AUC = 0.92), and a superordinate factor (TLI = 0.91, CFI = 0.92, GFI = 0.90, RMSEA = 0.05).

These results support the use of ASTS in future research and as an initial component of assessment for autism in adults. However, further studies with diverse samples are needed to confirm the findings before integrating the ASTS into clinical settings as part of the official diagnostic process without requiring additional components.

Footnotes

Acknowledgment

The authors would like to thank Jacob Sherman, David Cole, and Frank G. Culkar IV—friends of the first author—for their contributions to the development of the ASTS and their invaluable insights into autism in adulthood.

Authorship Confirmation Statement

S.M.A.: Conceptualization, data curation, formal analysis, funding acquisition, investigation, methodology, validation, visualization, writing—original draft, and writing—review and editing. N.M.F.: Data curation, formal analysis, software, validation, visualization, and writing—review and editing. K.W.R.: Methodology, project administration, supervision, and reviewing and editing. A.M.P.: Conceptualization, methodology, project administration, resources, supervision, and reviewing and editing. The article has been submitted solely to Autism in Adulthood.

Ethical Considerations

We have complied with the American Psychological Association’s ethical standards and guidelines. This study was not preregistered. We reported how we determined our sample size, all data exclusions, all manipulations, and all measures in the study. We are committed to promoting full transparency of our research study. Materials and analysis code for this study are available by emailing the corresponding author.

Author Disclosure Statement

The authors have no conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding Information

This study was made possible, in part, by the Cleveland State University’s Graduate Student Research Award, which granted the first author funds for the following: a summer stipend, conference travel, and the random drawing for five Amazon gift cards, each worth $5.00.

Supplementary Material

References

APA. Diagnostic and Statistical Manual of Mental Disorders | Psychiatry Online. DSM Library. 2022. Available from: https://psychiatryonline.org/doi/book/10.1176/appi.books.9780890425596 [Last accessed: September 3, 2024].

Cage

, Di Monaco

, Newell

. Experiences of autism acceptance and mental health in autistic adults. J Autism Dev Disord. 2018; 48(2):473–484; doi: 10.1007/s10803-017-3342-7

Kintzinger

. Equity: What model should we use when we talk about autism? CJAE. 2021; 1(1):32–39; doi: 10.15173/cjae.v1i1.4982

Taboas

, Doepke

, Zimmerman

. Preferences for identity-first versus person-first language in a US sample of autism stakeholders. Autism. 2023; 27(2):565–570; doi: 10.1177/13623613221130845

Maenner

, Warren

, Williams

, et al. Prevalence and characteristics of autism spectrum disorder among children aged 8 years—Autism and Developmental Disabilities Monitoring Network, 11 sites, United States, 2020. MMWR Surveill Summ. 2023; 72(2):1–14; doi: 10.15585/mmwr.ss7202a1

Filipek

, Accardo

, Ashwal

, et al. Practice parameter: Screening and diagnosis of autism: Report of the Quality Standards Subcommittee of the American Academy of Neurology and the Child Neurology Society. Neurology. 2000; 55(4):468–479; doi: 10.1212/wnl.55.4.468

McDonald

TAM

. Autism identity and the “Lost generation”: Structural validation of the autism spectrum identity scale and comparison of diagnosed and self-diagnosed adults on the autism spectrum. Autism in Adulthood. 2020; 2(1):13–23; doi: 10.1089/aut.2019.0069

Brosnan

. An exploratory study of a dimensional assessment of the diagnostic criteria for autism. J Autism Dev Disord. 2020; 50(11):4158–4164; doi: 10.1007/s10803-020-04474-8

Leedham

, Thompson

, Smith

, Freeth

. “I was exhausted trying to figure it out”: The experiences of females receiving an autism diagnosis in middle to late adulthood. Autism. 2020; 24(1):135–146; doi: 10.1177/1362361319853442

10.

Lewis

. Realizing a diagnosis of autism spectrum disorder as an adult. Int J Ment Health Nurs. 2016; 25(4):346–354; doi: 10.1111/inm.12200

11.

Stagg

, Belcher

. Living with autism without knowing: Receiving a diagnosis in later life. Health Psychol Behav Med. 2019; 7(1):348–361; doi: 10.1080/21642850.2019.1684920

12.

Goss

, Huynh

, Taing

, Brumback

. Approaching autism diagnosis and care through the lens of gender diversity. Child Neurol Open. 2023; 10:2329048X231219201; doi: 10.1177/2329048X231219201

13.

McCrossin

. Finding the true number of females with autistic spectrum disorder by estimating the biases in initial recognition and clinical diagnosis. Children (Basel). 2022; 9(2):272; doi: 10.3390/children9020272

14.

Lai

, Lombardo

, Pasco

, et al.; MRC AIMS Consortium. A behavioral comparison of male and female adults with high functioning autism spectrum conditions. PLoS One. 2011; 6(6):e20835; doi: 10.1371/journal.pone.0020835

15.

Cage

, Troxell-Whitman

. Understanding the reasons, contexts and costs of camouflaging for autistic adults. J Autism Dev Disord. 2019; 49(5):1899–1911; doi: 10.1007/s10803-018-03878-x

16.

Bargiela

, Steward

, Mandy

. The experiences of late-diagnosed women with autism spectrum conditions: An investigation of the female autism phenotype. J Autism Dev Disord. 2016; 46(10):3281–3294; doi: 10.1007/s10803-016-2872-8

17.

Hull

, Petrides

, Mandy

. The female autism phenotype and camouflaging: A narrative review. Rev J Autism Dev Disord. 2020; 7(4):306–317; doi: 10.1007/s40489-020-00197-9

18.

Brickhill

, Atherton

, Piovesan

, Cross

. Autism, thy name is man: Exploring implicit and explicit gender bias in autism perceptions. PLoS One. 2023; 18(8):e0284013; doi: 10.1371/journal.pone.0284013

19.

Mandell

, Novak

, Zubritsky

. Factors associated with age of diagnosis among children with autism spectrum disorders. Pediatrics. 2005; 116(6):1480–1486; doi: 10.1542/peds.2005-0185

20.

Tollit

, Maloof

, Hoq

, et al. A comparison of gender diversity in transgender young people with and without autistic traits from the Trans 20 cohort study. Lancet Reg Health West Pac. 2024; 47:101084; doi: 10.1016/j.lanwpc.2024.101084

21.

Van Der Miesen

AIR

, Hurley

, De Vries

ALC

. Gender dysphoria and autism spectrum disorder: A narrative review. Int Rev Psychiatry. 2016; 28(1):70–80; doi: 10.3109/09540261.2015.1111199

22.

Schuck

, Flores

, Fung

. Brief report: Sex/gender differences in symptomology and camouflaging in adults with autism spectrum disorder. J Autism Dev Disord. 2019; 49(6):2597–2604; doi: 10.1007/s10803-019-03998-y

23.

de Leeuw

, Happé

, Hoekstra

. A conceptual framework for understanding the cultural and contextual factors on autism across the globe. Autism Res. 2020; 13(7):1029–1050; doi: 10.1002/aur.2276

24.

Aylward

, Gal-Szabo

, Taraman

. Racial, ethnic, and sociodemographic disparities in diagnosis of children with autism spectrum disorder. J Dev Behav Pediatr. 2021; 42(8):682–689; doi: 10.1097/DBP.0000000000000996

25.

Turnock

, Langley

, Jones

CRG

. Understanding stigma in autism: A narrative review and theoretical model. Autism Adulthood. 2022; 4(1):76–91; doi: 10.1089/aut.2021.0005

26.

Malone

, Pearson

, Palazzo

, Manns

, Rivera

, Mason Martin

. The scholarly neglect of black autistic adults in autism research. Autism Adulthood. 2022; 4(4):271–280; doi: 10.1089/aut.2021.0086

27.

Durkin

, Maenner

, Baio

, et al. Autism spectrum disorder among US children (2002–2010): Socioeconomic, racial, and ethnic disparities. Am J Public Health. 2017; 107(11):1818–1826; doi: 10.2105/AJPH.2017.304032

28.

Hosozawa

, Sacker

, Cable

. Timing of diagnosis, depression and self-harm in adolescents with autism spectrum disorder. Autism. 2021; 25(1):70–78; doi: 10.1177/1362361320945540

29.

Thompson-Hodgetts

, Labonte

, Mazumder

, Phelan

. Helpful or harmful? A scoping review of perceptions and outcomes of autism diagnostic disclosure to others. Res Autism Spectr Disord. 2020; 77:101598; doi: 10.1016/j.rasd.2020.101598

30.

Crawshaw

. Should We Continue to Tell Autistic People that Their Brains are Different? - Daniel Crawshaw, 2023. May 2023. Accessed September 3, 2024. Available from: https://journals.sagepub.com/doi/full/10.1177/00332941231174391

31.

Lupușoru

, Oroian

, Cămănaru

, Cojocaru

, Szalontay

. Navigating challenges and supportive strategies in adult autism. BPI. 2024; 100(1):161–170; doi: 10.36219/BPI.2024.1.16

32.

Jadav

, Bal

. Associations between co‐occurring conditions and age of autism diagnosis: Implications for mental health training and adult autism research. Autism Res. 2022; 15(11):2112–2125; doi: 10.1002/aur.2808

33.

Bölte

, Girdler

, Marschik

. The contribution of environmental exposure to the etiology of autism spectrum disorder. Cell Mol Life Sci. 2019; 76(7):1275–1297; doi: 10.1007/s00018-018-2988-4

34.

Carpenter

. Diagnosis and assessment in autism spectrum disorders. Adv Ment Health Intellect Disabil. 2012; 6(3):121–129; doi: 10.1108/20441281211227184

35.

Thabtah

. An accessible and efficient autism screening method for behavioural data and predictive analyses. Health Informatics J. 2019; 25(4):1739–1755; doi: 10.1177/1460458218796636

36.

Bishop

, Lord

. Commentary: Best practices and processes for assessment of autism spectrum disorder—the intended role of standardized diagnostic instruments. J Child Psychol Psychiatry. 2023; 64(5):834–838; doi: 10.1111/jcpp.13802

37.

Zhai

, Li

, Zhou

, et al. Correlation and predictive ability of sensory characteristics and social interaction in children with autism spectrum disorder. Front Psychiatry. 2023; 14:1056051; doi: 10.3389/fpsyt.2023.1056051

38.

Maddox

, Crabbe

, Beidas

, et al. “I wouldn’t know where to start”: Perspectives from clinicians, agency leaders, and autistic adults on improving community mental health services for autistic adults. Autism. 2020; 24(4):919–930; doi: 10.1177/1362361319882227

39.

Okoye

, Obialo-Ibeawuchi

, Obajeun

, et al. Early diagnosis of autism spectrum disorder: A review and analysis of the risks and benefits. Cureus. 2023; 15(8):e43226; doi: 10.7759/cureus.43226

40.

Huang

, Arnold

SRC

, Foley

, Trollor

. Choose your own adventure: Pathways to adulthood autism diagnosis in Australia. J Autism Dev Disord. 2022; 52(7):2984–2996; doi: 10.1007/s10803-021-05169-4

41.

Shattuck

, Grosse

. Issues related to the diagnosis and treatment of autism spectrum disorders. Ment Retard Dev Disabil Res Rev. 2007; 13(2):129–135; doi: 10.1002/mrdd.20143

42.

Rutter

, Howard

, Lakhan

, Valdez

, Bollen

, Lorenzo-Luaces

. “I haven’t been diagnosed, but I should be”—Insight into self-diagnoses of common mental health disorders: Cross-sectional study. JMIR Form Res. 2023; 7:e39206; doi: 10.2196/39206

43.

Zwick

. Neuropsychological assessment in autism spectrum disorder and related conditions. Dialogues Clin Neurosci. 2017; 19(4):373–379; doi: 10.31887/DCNS.2017.19.4/gzwick

44.

Lord

, Rutter

, Le Couteur

. Autism diagnostic interview-revised: A revised version of a diagnostic interview for caregivers of individuals with possible pervasive developmental disorders. J Autism Dev Disord. 1994; 24(5):659–685; doi: 10.1007/BF02172145

45.

Jia

, Steelman

, Jia

. Psychometric assessments of three self-report autism scales (AQ, RBQ-2A, and SQ) for general adult populations. J Autism Dev Disord. 2019; 49(5):1949–1965. | Accessed September 3, 2024. https://link.springer.com/article/10.1007/s10803-019-03880-x

46.

Matson

, Wilkins

, González

. Reliability and factor structure of the autism spectrum disorders—Diagnosis scale for intellectually disabled adults (ASD—DA). J Dev Phys Disabil. 2007; 19(6):565–577. | Accessed September 3, 2024. https://link.springer.com/article/10.1007/s10882-007-9070-8

47.

Aguiar

M D

, Pondé

. Autism: Impact of the diagnosis in the parents. J bras psiquiatr. 2020; 69(3):149–155; doi: 10.1590/0047-2085000000276

48.

Bishop

, Seltzer

. Self-reported autism symptoms in adults with autism spectrum disorders. J Autism Dev Disord. 2012; 42(11):2354–2363; doi: 10.1007/s10803-012-1483-2

49.

Hayes

, Ford

, Rafeeque

, Russell

. Clinical practice guidelines for diagnosis of autism spectrum disorder in adults and children in the UK: A narrative review. BMC Psychiatry. 2018; 18(1):222; doi: 10.1186/s12888-018-1800-1

50.

Karmiloff-Smith

. Development itself is the key to understanding developmental disorders. Trends Cogn Sci. 1998; 2(10):389–398; doi: 10.1016/s1364-6613(98)01230-3

51.

Fleiss

. Design and Analysis of Clinical Experiments. John Wiley & Sons; 2011.

52.

Guthrie

, Swineford

, Nottke

, Wetherby

. Early diagnosis of autism spectrum disorder: Stability and change in clinical diagnosis and symptom presentation. J Child Psychol Psychiatry. 2013; 54(5):582–590; doi: 10.1111/jcpp.12008

53.

Kaufman

. Rethinking “gold standards” and “best practices” in the assessment of autism. Appl Neuropsychol Child. 2022; 11(3):529–540; doi: 10.1080/21622965.2020.1809414

54.

Baron-Cohen

, Richler

, Bisarya

, Gurunathan

, Wheelwright

. The systemizing quotient: An investigation of adults with Asperger syndrome or high-functioning autism, and normal sex differences. Philos Trans R Soc Lond B Biol Sci. 2003; 358(1430):361–374; doi: 10.1098/rstb.2002.1206

55.

Baron-Cohen

, Wheelwright

, Skinner

, Martin

, Clubley

. The autism-spectrum quotient (AQ): Evidence from Asperger syndrome/high-functioning autism, males and females, scientists and mathematicians. J Autism Dev Disord. 2001; 31(1):5–17; doi: 10.1023/a:1005653411471

56.

Ketelaars

, Horwitz

, Sytema

, et al. Brief report: Adults with mild autism spectrum disorders (ASD): Scores on the autism spectrum quotient (AQ) and comorbid psychopathology. J Autism Dev Disord. 2008; 38(1):176–180; doi: 10.1007/s10803-007-0358-4

57.

Horwitz

, Schoevers

, Greaves-Lord

, de Bildt

, Hartman

. Adult manifestation of milder forms of autism spectrum disorder; autistic and non-autistic psychopathology. J Autism Dev Disord. 2020; 50(8):2973–2986; doi: 10.1007/s10803-020-04403-9

58.

Gruijters

. Using principal component analysis to validate psychological scales: Bad statistical habits we should have broken yesterday II. Published online July 8, 2020. doi:10.31234/osf.io/qbe6k

59.

English

MCW

, Gignac

, Visser

TAW

, Whitehouse

AJO

, Enns

, Maybery

. The comprehensive autistic trait inventory (CATI): Development and validation of a new measure of autistic traits in the general population. Mol Autism. 2021; 12(1):37; doi: 10.1186/s13229-021-00445-7

60.

Sonido

, Hwang

YI(J)

, Trollor

, Arnold

SRC

. The mental well-being of informal carers of adults on the autism spectrum: A systematic review. Rev J Autism Dev Disord. 2020; 7(1):63–77; doi: 10.1007/s40489-019-00177-8

61.

Balasco

, Provenzano

, Bozzi

. Sensory abnormalities in autism spectrum disorders: A focus on the tactile domain, from genetic mouse models to the clinic. Front Psychiatry. 2019; 10:1016; doi: 10.3389/fpsyt.2019.01016

62.

Sandercock

, Lamarche

, Klinger

. Assessing the convergence of self-report and informant measures for adults with autism spectrum disorder. Autism. 2020; 24(8):2256–2268; doi: 10.1177/1362361320942981

63.

Barahona-Corrêa

, Filipe

. A concise history of Asperger syndrome: The short reign of a troublesome diagnosis. Front Psychol. 2015; 6:2024; doi: 10.3389/fpsyg.2015.02024

64.

Skuse

. Few people mourn Asperger syndrome’s loss from diagnostic manuals. The Transmitter: Neuroscience News and Perspectives. May 9, 2018. Accessed February 26, 2025. https://www.thetransmitter.org/spectrum/people-mourn-asperger-syndromes-loss-diagnostic-manuals/

65.

Topal

, Adıgüzel

, Tufan

. Symptoms extending from identity confusion to personality pattern, autistic spectrum to psychotic spectrum via an adolescent case. Turk J Child Adolesc Ment Health. 2021; 28(1):65–68; doi: 10.4274/tjcamh.galenos.2020.80774

66.

Volkmar

, Klin

, Schultz

, et al. Asperger’s syndrome. J Am Acad Child Adolesc Psychiatry. 1996; 35(1):118–123; doi: 10.1097/00004583-199601000-00020

67.

Freeman

, Cronin

, Candela

. Asperger syndrome or autistic disorder?: The diagnostic dilemma. Focus Autism Other Dev Disabl. 2002; 17(3):145–151; doi: 10.1177/10883576020170030401

68.

Busch

. A Deep Learning Approach to Understanding Real-World Scene Perception in Autism. Dartm Coll Undergrad Theses. Published online January 1, 2020. https://digitalcommons.dartmouth.edu/senior_theses/273

69.

Kaiser

. An index of factorial simplicity. Psychometrika. 1974; 39(1):31–36; doi: 10.1007/BF02291575

70.

Clark

, Watson

. Constructing validity: New developments in creating objective measuring instruments. Psychol Assess. 2019; 31(12):1412–1427; doi: 10.1037/pas0000626

71.

Boateng

, Neilands

, Frongillo

, Melgar-Quiñonez

, Young

. Best practices for developing and validating scales for health, social, and behavioral research: A primer. Front Public Health. 2018; 6:149; doi: 10.3389/fpubh.2018.00149

72.

Opitz

, Newman

, Alvarado Vázquez Mellado

, Robertson

MDA

, Sharpe

. The psychometric properties of Orthorexia Nervosa assessment scales: A systematic review and reliability generalization. Appetite. 2020; 155:104797; doi: 10.1016/j.appet.2020.104797

73.

Strauss

, Pelletier-Baldelli

, Visser

, Walker

, Mittal

. A review of negative symptom assessment strategies in youth at clinical high-risk for psychosis. Schizophr Res. 2020; 222:104–112; doi: 10.1016/j.schres.2020.04.019

74.

Dunning

, Heath

, Suls

. Flawed self-assessment: Implications for health, education, and the workplace. Psychol Sci Public Interest. 2004; 5(3):69–106; doi: 10.1111/j.1529-1006.2004.00018.x

75.

Wymer

, Maria

. A review of scale development practices in nonprofit management and marketing. E&S. 2012; 5(2):143–151.

76.

Ballester

MAG

, Linguraru

, Aguirre

, Ayache

. On the adequacy of principal factor analysis for the study of shape variability. In: Medical Imaging 2005: Image Processing. Vol 5747. SPIE; 2005:1392–1399. doi:10.1117/12.593333

77.

Costello

, Osborne

. Best practices in exploratory factor analysis: Four recommendations for getting the most from your analysis. Pract Assess Res Eval. 2005; 10(1); doi: 10.7275/jyj1-4868

78.

Morris-Rosendahl

, Crocq

. Neurodevelopmental disorders—the history and future of a diagnostic concept. Dialogues Clin Neurosci. 2020; 22(1):65–72; doi: 10.31887/DCNS.2020.22.1/macrocq

79.

Gillespie-Lynch

, Kapp

, Brooks

, Pickens

, Schwartzman

. Whose expertise is it? Evidence for autistic adults as critical autism experts. Front Psychol. 2017; 8:438; doi: 10.3389/fpsyg.2017.00438

80.

Ritvo

, Ritvo

, Guthrie

, et al. The Ritvo Autism Asperger Diagnostic Scale-Revised (RAADS-R): A scale to assist the diagnosis of autism spectrum disorder in adults: An international validation study. J Autism Dev Disord. 2011; 41(8):1076–1089; doi: 10.1007/s10803-010-1133-5

81.

Eriksson

, Andersen

, Bejerot

. RAADS-14 screen: Validity of a screening tool for autism spectrum disorder in an adult psychiatric population. Mol Autism. 2013; 4(1):49; doi: 10.1186/2040-2392-4-49

82.

Harris

, Scott

, Lebo

, Hassan

, Lightner

, Pulley

. ResearchMatch: A national registry to recruit volunteers for clinical research. Acad Med. 2012; 87(1):66–73; doi: 10.1097/ACM.0b013e31823ab7d2

83.

Temkin

. The Global State of XM, 2020. Published online 2020.

84.

Ekman

, Friesen

. Measuring facial movement. J Nonverbal Behav. 1976; 1(1):56–75; doi: 10.1007/BF01115465

85.

Schober

, Mascha

, Vetter

. Statistics from A (Agreement) to Z (z Score): A guide to interpreting common measures of association, agreement, diagnostic accuracy, effect size, heterogeneity, and reliability in medical research. Anesth Analg. 2021; 133(6):1633–1641; doi: 10.1213/ANE.0000000000005773

86.

Penny

. Appropriate critical values when testing for a single multivariate outlier by using the Mahalanobis distance. J R Stat Soc Ser C Appl Stat. 1996; 45(1):73–81; doi: 10.2307/2986224

87.

IBM Corp. SPSS for Windows. Published online 2021.

88.

Wirth

, Edwards

. Item factor analysis: Current approaches and future directions. Psychol Methods. 2007; 12(1):58–79; doi: 10.1037/1082-989X.12.1.58

89.

Arbuckle

. IBM® SPSS® Amos^TM 25 User’s Guide.

90.

Cheung

, Cooper-Thomas

, Lau

, Wang

. Reporting reliability, convergent and discriminant validity with structural equation modeling: A review and best-practice recommendations. Asia Pac J Manag. 2024; 41(2):745–783; doi: 10.1007/s10490-023-09871-y

91.

Shek

DTL

, Yu

. Confirmatory factor analysis using AMOS: A demonstration. Int J Disabil Hum Dev. 2014; 13(2):191–204; doi: 10.1515/ijdhd-2014-0305

92.

Putnick

, Bornstein

. Measurement invariance conventions and reporting: The state of the art and future directions for psychological research. Dev Rev. 2016; 41:71–90; doi: 10.1016/j.dr.2016.06.004

93.

L T

, Bentler

. Fit indices in covariance structure modeling: Sensitivity to underparameterized model misspecification. Psychol Methods. 1998; 3(4):424–453; doi: 10.1037/1082-989X.3.4.424

94.

Henson

. Understanding internal consistency reliability estimates: A conceptual primer on coefficient alpha. Meas Eval Couns Dev. 2001; 34(3):177–189; doi: 10.1080/07481756.2002.12069034

95.

Nelson

, Switalsky

, Ciesielski

, et al. A scoping review of supports on college and university campuses for autistic post-secondary students. Front Educ. 2023; 8; doi: 10.3389/feduc.2023.1179865

96.

Bakker

, Krabbendam

, Bhulai

, Meeter

, Begeer

. Predicting academic success of autistic students in higher education. Autism. 2023; 27(6):1803–1816; doi: 10.1177/13623613221146439

97.

Syriopoulou-Delli

, Sarri

, Papaefstathiou

, Filiou

, Gkiolnta

. Educational programs supporting higher education individuals with autism spectrum disorder: A systematic literature review. Trends High Educ. 2024; 3(3):710–724; doi: 10.3390/higheredu3030040

98.

Knekta

, Runyon

, Eddy

. One size doesn’t fit all: Using factor analysis to gather validity evidence when using surveys in your research. CBE Life Sci Educ. 2019; 18(1):rm1; doi: 10.1187/cbe.18-04-0064

99.

Hahs-Vaughn

. Applied Multivariate Statistical Concepts. Routledge; 2016. doi:10.4324/9781315816685

100.

Jiang

, Mai

, Yuan

. Advances in measurement invariance and mean comparison of latent variables: Equivalence testing and a projection-based approach. Front Psychol. 2017; 8:1823; doi: 10.3389/fpsyg.2017.01823

101.

Barrett

, Uljarević

, Baker

, Richdale

, Jones

CRG

, Leekam

. The adult Repetitive Behaviours Questionnaire-2 (RBQ-2A): A self-report measure of restricted and repetitive behaviours. J Autism Dev Disord. 2015; 45(11):3680–3692; doi: 10.1007/s10803-015-2514-6

102.

Boulton

, Hodge

, Jewell

, Ong

, Silove

, Guastella

. Diagnostic delay in children with neurodevelopmental conditions attending a publicly funded developmental assessment service: Findings from the Sydney Child Neurodevelopment Research Registry. BMJ Open. 2023; 13(2):e069500; doi: 10.1136/bmjopen-2022-069500

103.

MacLennan

, O’Brien

, Tavassoli

. In our own words: The complex sensory experiences of autistic adults. J Autism Dev Disord. 2022; 52(7):3061–3075; doi: 10.1007/s10803-021-05186-3

104.

Parrott

. Performance tests in human psychopharmacology (2): Content validity, criterion validity, and face validity. Human Psychopharmacology. 1991; 6(2):91–98; doi: 10.1002/hup.470060203

105.

Tonnsen

, Boan

, Bradley

, Charles

, Cohen

, Carpenter

. Prevalence of autism spectrum disorders among children with intellectual disability. Am J Intellect Dev Disabil. 2016; 121(6):487–500; doi: 10.1352/1944-7558-121.6.487

106.

Bungert

, Li

, Cardinaux

, et al. Proportional overrepresentation of gender-diverse identities in two US-based autistic adult samples from the SPARK database. Autism Adulthood. 2024; doi: 10.1089/aut.2023.0121

107.

Hutcheson

. Missing data: Data replacement and imputation. J Model Manag. 2012; 7(2); doi: 10.1108/jm2.2012.29707baa.002

Supplementary Material

Please find the following supplemental material available below.

For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.

For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.

0.00 MB

0.08 MB

0.52 MB