Understanding Online Job and Housing Search Practices of Neurodiverse Young Adults to Support Their Independence

Securing employment and housing are key aspects of pursuing independent living. As these activities are increasingly practiced online, web accessibility of related services becomes critical for a successful major life transition. Support for this transition is especially important for people with autism or intellectual disability, who often face issues of underemployment and social isolation. In this study, we conducted semi-structured interviews and contextual inquiries with neurotypical adults and adults with autism or intellectual disability to understand common and unique goals, strategies, and challenges of neurodiverse adults when searching for employment and housing resources online. Our findings revealed that current interfaces adequately support practical (e.g., finance) goals but lack information on social (e.g., inclusivity) goals. Furthermore, unexpected search results and inaccessible social and contextual information diminished search experiences for neurodivergent users, which suggests the need for predictability and structured guidance in searching online. We conclude with design suggestions to make neurodivergent users’ online search experience an opportunity to demonstrate their independence.


INTRODUCTION
The use of browsers and mobile application technology is critical to many aspects of modern life.This is particularly true in the wake of the COVID-19 pandemic, as services related to personal and professional activities are increasingly available primarily or exclusively online [48,55].With the increased activities online, many employment and housing providers now require applicants to apply through online portals.As securing employment and housing are key in transitioning to adulthood [83], the shift of related services to online spaces makes web navigation of these job and housing websites critical for young adults in pursuing independence.While ofering these services online makes them accessible to people regardless of geography, information and services should also remain accessible to all people regardless of their physical or cognitive abilities.Yet for many individuals with disabilities who are in a transitional phase, inaccessibility of websites and inadequate information regarding accessibility and accommodations remain a signifcant barrier that aggravates existing employment and housing disparities [18,25,31].
Transition planning for independent living is crucial for everyone, but it holds particular signifcance 1 for individuals with autism [39,83] and those with intellectual disabilities [94].While both neurodivergent and neurotypical young adults may share the desire for independent living, it is worth noting that the unemployment and underemployment rates for these two specifc neurodivergent subgroups, people with intellectual disability and/or autism, are higher than average (even in comparison to other disability groups), emphasizing the need for tailored support [73].Furthermore, supporting neurodivergent young adults to reside in non-family settings can help them acquire new skills [54] and become more integrated into society beyond their immediate families [43].This marks the importance of designing online interfaces for job and housing searches that are efective for both neurotypical and neurodivergent individuals since inaccessible websites may further exacerbate un(der)employment and housing inequalities.Following the principles of universal design, which aims for designs that are "usable by all people, to the greatest extent possible" [21], our objective is to inspire designs that are not merely usable but truly useful for individuals across the neurodiversity spectrum.Despite the fact that identifying user goals and motivations is essential to user experience and interaction design [30], there is limited research that comprehensively investigates the goals and needs of neurodiverse young adults during online job and housing searches and how useful the existing sites are in light of their goals.
Prior works on how users with disabilities use the internet for information search and web navigation have largely focused on the usability of the current technology such as the graphical layout, structure, and the language used [32,52,63].However, the usefulness of a system is determined by both the usability and utility [75].In other words, even if the system is easy and pleasant to use, it may lack the features to meet the users' needs.Thus the focus on accessibility only from a usability standpoint leaves out an important aspect of a system, its utility.This was demonstrated in Liu et al.'s study of blind people doing online shopping, which has shown that in addition to the usability issues such as navigation and product recognition, social and cultural aspects largely afected the shopping behavior and experience of blind people [61].Building on this idea of holistic usefulness as a new framework for web accessibility, we aim to examine both the usability and utility of current online services for transition planning.
For this purpose, we conducted a contextual inquiry and a semistructured interview with 16 participants as they searched for housing and employment.We chose to study these two search domains (i.e., housing and job) since fnding satisfactory employment and housing are key aspects of pursuing independent living.By studying these two types of searches together, we aimed to identify common search strategies and challenges that infuence participants' web searches throughout their transitional journey.While the concept of neurodiversity can be interpreted diferently by various individuals [26], we use Gillespie-Lynch et al.'s expanded framework that defnes neurodivergence to encompass a spectrum of conditions, including but not limited to attention defcit hyperactivity disorder (ADHD), anxiety, depression, autism, and intellectual disability [40].Our study focuses on a small subset of this population -people with autism and/or intellectual disability -as these two subgroups face a similar set of challenges in their transition toward independent living including underemployment [64,73] and social exclusion in communities [49,74].Due to the shared challenges in transition, these two subgroups have often been studied together in prior research on transition-aged youth [13,24].Similarly, our participant grouping centered around the potential challenges they face.
Our results showed that although most job and housing searchrelated websites were usable by all participants, there was a misalignment between the transition goals mentioned by the participants and their web search behavior during the study.While most of the practical (e.g., fnancial, vocational) goals were actively considered in the job and housing search, contextual (e.g., social, safety, inclusivity) goals were left out as they were largely unsupported by the interface.In other words, the participants' search was driven by features on the interface rather than by their goals.We also uncovered a diference between the groups in their reactions when the system behavior diverged from their mental model, such as unexpectedly seeing an empty search result page.Neurotypical participants often adapted to the situation by adjusting the search terms, undoing the latest action, or ignoring irrelevant information.On the other hand, participants with intellectual disability or autism restarted the search from the beginning or sometimes abandoned the site altogether.Yet, once these participants formed an adequate mental model for the system by extensively exploring the site, they were able to successfully and even creatively use the sites to achieve their goals.Furthermore, the experience of a successful search made the participants feel a sense of accomplishment and independence.Based on our fndings, we make design suggestions for building systems that better conform to users' mental models, help users efciently integrate social and contextual information, and guide the users to a successful transition to independence.
Author Positionality.We acknowledge that this research is shaped by the unique perspectives, experiences, and biases of the authors.Three of the authors were heavily involved in data analysis and writing the manuscript.Two of these authors have over eight years of research experience with people with autism and their families.One of them and the other author have personal connections to the autism community as they have a sibling or a family member on the spectrum.Our collective research experience with the neurodivergent population is limited, as we have primarily worked with participants who are profcient in verbal communication and adept at using technology.Consequently, our interpretations of the data and presentation of results and discussion may have included nuanced insights that may be infuenced by personal biases and experiences.

RELATED WORK 2.1 Supporting Independence of Neurodivergent Young Adults
The post-school transition to adulthood is a period of change and anxiety for all youth, but especially so for neurodivergent (ND) young adults [68].The Individuals with Disabilities Education Act (IDEA '97) is a law by the U.S. Department of Education that requires transitional services to be provided for students with disabilities in the United States [28].Two key areas of support are employment and independent living.However, despite going through transitioning planning as a part of their Individual Education Program (IEP), many students with disabilities face poor post-school outcomes [15] and families often describe the end of special education services after graduation as "falling of a clif" [83].The U.S. National Longitudinal Transition Study-2 (NLTS-2) dataset that tracked around 11,000 young adults with disabilities revealed that only 63% of individuals with autism out of high school up to 8 years had been employed at some time compared to 91% of individuals with other disabilities [73].In the area of independent living, only 38.7% (SE = 24.3) of students with mild intellectual disability met their transitional goal within 2 to 4 years of graduation [15].Prior research has examined ways to empower low-resource and disadvantaged job seekers by designing a tool that identifes required skills and provides resources related to their dream job [35] and by investigating current employment tools (e.g., providing resume feedback) and the desired features [34].
To understand factors that compose current transitional barriers, it is also important to examine how young adults navigate the web for employment and housing.In recent years, the internet played an essential role in transition as 79% of job seekers in the United States used online resources and information as of 2015 [92] and job seekers who use the internet had a shorter duration of unemployment [93].Nearly all (99%) of home buyers in 2021 used the internet to fnd homes [33].As digital house hunting and job seeking become the norm, housing and employment-related websites should be accessible to all those who need the platform, regardless of their cognitive styles.Thus, our study examines how users with intellectual disability and/or autism fnd and use housing and employment-related websites to assess the usability of the existing sites (e.g., How long does it take them to identify and use appropriate flters?Can they easily change the location for the search?).We also examine the utility of the sites by assessing how well they assist the users in meeting their transitional goals.Studies show that people with intellectual disability are motivated by a desire to be productive and enjoy work environments where they are admired and have positive social connections [85,95].On a similar note, adults with autism reported self-actualization, job roles and work content, and social and collegial factors as the top three positive experiences in employment while pay and benefts were mentioned less frequently [11].Individuals with intellectual disability are also likely to consider transportation in their employment searches and expressed a desire to work in familiar environments [95].In comparison, NT groups place a higher value on salary and advancement opportunities over social factors and location [50].In housing search, everyone places importance on compatible roommates and housing cost, but people with disabilities also consider family and support networks nearby, accessible commutes, and landlords who support housing for people with disabilities [98,100].Considering the broad range of transitional goals, a website cannot be considered fully inclusive if particular goals are not appropriately supported, even if the website is usable by all.To discover ways to fll the gap between neurodiverse young adults' goals and the actual transitional outcomes, we studied how they search for jobs and housing online and examined whether the accessibility of the relevant sites could be one of the causes of the gap.

Web Accessibility for Digital Inclusivity
People with disabilities may have diverse physical and/or cognitive challenges.Guidelines for ensuring accessibility of online spaces and mobile applications for people with disabilities were published by the World Wide Web Consortium (W3C), an international group that guides the long-term growth of the internet [29,99].W3C's accessibility guides present recommendations with three levels of compliance (A-low, AA-mid, and AAA-high) for websites and applications.Meeting AA compliance has been adopted as a requirement by many bodies, including the US Government, for web content [1].Most A and AA-level standards focus on accessibility for users with physical disabilities, while accommodations that beneft those with cognitive disabilities are often rated AAA [37,87].Since many sites and mobile applications only follow A or AA-level standards, people with cognitive challenges may be marginalized in online spaces [37,69,87].Furthermore, the design guidelines to accommodate disabilities center around the usability of the system and thus overlook how the unique goals of ND users would be addressed in the system to make it fully inclusive.Responding to continued accessibility challenges for people with cognitive disabilities, researchers have developed specialized platforms [12,87], recommendations and guidelines [37,81], and design tools (e.g., persona, empathy map) to promote a designer's awareness of users with cognitive diferences [66,67].However, most guides and tools are based on expert knowledge from professionals or parents [37,66,67,81] and few have directly involved people with cognitive disabilities in design [12,78].The direct involvement of participants with intellectual disability or autism is key to our study since it allows us to observe what they value and search for while preparing for a transition.
Prior works on web and technology use of users have examined various disability groups (e.g., dyslexia [70], ADHD [65], down syndrome [56]) and domains (e.g., tourism [32], shopping [61], socialization [22,46,80]).Benefts of using computer-mediated communication for socialization included increased control over the communication, the ability to fnd similar others, and reduced stress from non-verbal signals [22,41].However, difculties remained in the areas of trust, disclosure, and online social norms [22].Morris et al.'s study provides a glimpse into the internet search behavior of ND people by focusing on a specifc sector of the ND population, people with dyslexia [71].Their study found a diference between dyslexic and non-dyslexic adults' search behaviors in the areas of query formulation, results triage, and information extraction as dyslexic users found the tasks more challenging.We build on this research to investigate the accessibility and use of internet searches by people with intellectual disability or autism in the domain of employment and housing search.Hu and Feng investigated information search behaviors of people with cognitive disabilities by comparing how they navigated deep and broad web content structures [47].They found that participants with cognitive disabilities navigated and searched more efectively on user interfaces with a deeper structure and fewer choices at each level compared to ones with a broader structure and more choices on fewer levels.Participants were able to detect and fx most types of search errors (e.g., typos, broad keywords, redundant keywords), but faced difculty in recognizing the target result, which was one of the main causes of failed searches.Results also showed that they frequently used a single type of interaction (i.e., clicking on the back button) when correcting mistakes during the search (e.g., navigating a wrong path, or entering a misspelled word).In contrast, Aula et al. 's study showed that NT users without disabilities formulate diverse search terms and strategies for error recovery [10].While Hu and Feng's study provides a preliminary understanding of web search-related difculties experienced by people with cognitive disabilities, they noted that future studies are needed that include NT participants as well, and where the tasks are performed on more realistic websites for ecological validity.As such, our study examines the search behaviors of both ND and NT participants on existing job and housing search websites to build on their fndings.
A few studies have studied ND adults in the context of employment [70,79], but they focused on the overall experience of the ND people at their workplace rather than their online employment search process.Lazar et al. looked at the workplace-related computer skills of expert users with Down Syndrome and found that they were able to engage in complex interaction tasks like flters and sorting for targeted search [57].As their study showed that people with various cognitive disabilities can learn computer skills which opens new possibilities for a successful career, we further study how they navigate webpages when they are given complex criteria to fnd these opportunities independently via online search.Bill and Ng worked on automating a part of the process by developing a job-matching algorithm for connecting adults with autism with potential employers based on the required job skills [14].As they tested the matching algorithm in a simulated environment, we yet have to study the experience of neurodivergent users using job search websites in real-world settings.In his literature review on technology for neurodiverse users, Motti emphasized that introducing strengths-based design principles can increase independence in web use [72].Thus, our research examined various aspects of the online job and housing search experience of neurodivergent2 and neurotypical young adults, not only focusing on the users' needs and challenges during the current search process but also on their strengths and coping strategies in light of their goals and motivations.Cooper stated that identifying user goals and motivations is central to interaction design [30].When it comes to independent living, users with diferent cognitive styles were shown to have diferent priorities [11,50,98,100] and motivation [85,95], and this may infuence the way they interact with the interface.Hence, our study was structured to frst understand the user goals, and then to identify if the existing interaction design caters to these goals.More specifcally, we aim to answer the following research questions through our study: (1) What are the similarities and diferences in transition goals between neurodivergent and neurotypical users, and how are the goals refected in their web searches?(2) What are the strengths of neurodivergent users when using housing and employment search websites?(3) What are the challenges and coping strategies of neurodivergent and neurotypical users when using housing and employment search websites?

Participants
For both ND and NT participants, our inclusion criteria were (a) young adults between the ages of 18 to 35, (b) those who have completed or are planning a major life transition such as searching for new employment or housing within the last or upcoming 6 months, and (c) searching online for resources related to the transition.While the categorizations and severity of cognitive disabilities can vary greatly, our study involved people with autism and/or intellectual disability who are profcient in using technology, competent in verbal communication, and actively pursuing independent lifestyles.We included these two subsets of the neurodivergent population as underemployment is a serious issue for both groups [64,73], and they have been studied together in previous studies on postsecondary education [45] and on inclusive workplaces [62].
We sent out recruitment fyers to general university mailing lists, regional developmental-disability-related mailing lists (e.g., the Autism Self-Advocacy Network), and the subreddit r/AutisticAdults.Participants were also recruited verbally at events for people with autism and intellectual disability, through members of personal and professional networks, and from Transition and Postsecondary Programs for Students with Intellectual Disability (TPSID).NT participants were recruited through fiers, university mailing lists, and word-of-mouth in the college groups.Both undergraduate and graduate students who have recently experienced a transition or are in the planning stages were targeted.Participants were asked screening questions on their transition planning and the self-identifcation of their disability, when applicable.We decided not to use one participant's data after examining his screening answers since he did not meet our inclusion criteria.
We conducted our study with 16 participants which included 6 participants with intellectual disability (ID), 2 participants with autism (AS), and 8 neurotypical (NT) participants.Table 1 shows demographic information and the work and housing experience of each participant.All of our participants were living in a large city or nearby suburban area in the Southeast U.S. when the study took place.The median asking rent in the area is close to the national monthly median (around $1,500), and the unemployment rate is 3.4% [6,51].The city is rated as slightly unsafe with a crime rate that is 1.75 times the national average but a lower crime rate compared to similarly sized metro areas [7].The city received moderate walk, transit, and bike scores [8] indicating that a person is likely cardependent for accessing amenities due to limited reliable public bus and rail options and limited infrastructure for bikers.
All six participants with intellectual disabilities had recently graduated from the TPSID program, which is a post-secondary program that serves students with intellectual disabilities focusing on socialization, independent living skills, and integrated work experiences.They all know basic mathematics, can use a tablet or computer, have no signifcant behavioral/emotional challenges, and are able to live and work independently for long periods of time.In addition, all ID and AS participants were able to use language at a conversational level, independently operate computers to search for online resources, and seek resources to support independent living.All participants were highly motivated individuals with a major upcoming transition and looking for an independent lifestyle in the long run.All participants used online searches on an everyday basis and were adept at using a search engine to perform information retrieval-based tasks.

Study Procedure
Participants were given the option to interview in person or remotely using a video conferencing platform with the fexibility to turn of their video at any time.This was to ensure that participants felt comfortable during the study.All participants chose to interview remotely over Microsoft Teams.On average, a session took about 90 minutes for the ND participants and 60 minutes for the NT participants.The researchers recorded the conference calls for a detailed analysis.ND and NT participants received a $20 and $15 Amazon gift card respectively.We compensated ND participants understand whether they were able to meet the goals they enumore than NT participants since the tasks and the interview could merated in the frst phase.The contextual inquiry consists of a require more time and efort from ND individuals.For participants participants-driven search and a researcher-promoted search.Durwith ID, we had a comprehension check after going over the consent ing the participants-driven search, participants performed web form together, where we asked them to explain the basic phases of searches for housing and employment starting from a search enthe study in their own words to verify understanding.At the begin-gine of their choice, while screen sharing.In order to discuss the ning of the study, the research team emphasized that participants strengths and weaknesses of existing websites, at least two diferent could stop the study anytime and take a break or leave without websites for each domain needed to be compared.Therefore, particany penalty.The participants were also informed that they would ipants who only used a single website for their search were asked receive the full compensation amount no matter when they ended to repeat the search using a diferent interviewer-provided webthe study.All the participants completed the full study, except for site.Depending on the site that the participant had already used, ID1 who ended the study after viewing one housing site.He was we chose Apartments.com [2] or Zillow [5] for housing-related frustrated that his computer was freezing during the search and the search, and Indeed.com[3] or Linkedin [4] for employment search search process did not go as he expected.When he expressed his since they were the most popular sites in their respective domains desire to exit the study, we thanked him for his participation, high- [42,77].We designed the participant-driven search to allow for lighted his contribution, and compensated him for the full study greater ecological validity by seeking participants in transition curamount.
rently or recently and allowing them to search on the websites of Our study was conducted in three phases: a pre-study interview, their choice based on their own criteria.Despite these eforts, the contextual inquiry, and a post-study interview.The three-phase experience may have difered from real-world experiences due to structure of the study was inspired by the fndings of Rose et al.
the limited time and the requirement to complete the entire search which highlighted the importance of understanding user's search in one session.We monitored the diferent features and tasks the goals and behavior, and not only how people search [82].Therefore, participants engaged with, and the level of ease or difculty in comthe pre-study interview was designed to understand participants' pleting these tasks.We also observed how the participants reacted goals and motivations in their search before assessing the utility of to the success or failure in fnding the desired search result.Note the existing websites (i.e., how well they address the users' needs).
that we designed the participant-driven search to allow for greater All interview questions were phrased at the 6th grade level for ecological validity.This involved recruiting participants who were comprehension.They covered the participants' major transition currently or recently in transition, as described in the Participants goals, diferent housing criteria (e.g., amenities, rent) and employ-section, and allowing them to search on the websites of their choice ment criteria (e.g., pay, location), and information on their search using their own computer or laptop.Despite these eforts, the exresources (e.g., websites, friends).We collected participants' job and perience may have difered from real-world experiences due to the housing search goals before the contextual inquiry to ensure that limited time and the requirement to complete the entire search in the reported goals were not infuenced by their search experience one session.during the contextual inquiry.
After completing the participant-driven search, the researchers In the second phase, through contextual inquiry, we aimed to prompted the participants to use more specifc search criteria to uncover the online search strategies employed by participants and ground the searches on their previously stated goals and criteria (e.g., under $2k per month, public transportation friendly).For example, if the participant had price and safety as their primary criteria while searching for independent living, the interviewer prompted for an apartment search "within $2000 budget in the desired location."This ensured that all participants used various features (e.g., flters) available on a site to examine the usability of the site and allowed us to assess how adequately the existing sites were designed to address their initial goals.During this phase, we ofered gentle nudges for all participants when there were multiple failed searches (e.g., choosing the wrong flter of "house" instead of "apartment"), and empathized with the participant's difculties.If the participants voiced or showed signs of discomfort, we asked if they wanted to take a break or move to a diferent task.Note that we observed increased use of advanced features (e.g., interactive map, fltering, sorting, dropdown menus) in the researcher-prompted phase when compared with the participant-driven phase.However, we did not observe a signifcant diference in participants' abilities to complete the participant-driven searches and the researcherprompted searches.
After the contextual inquiry, the fnal phase, a post-study interview was conducted where participants refected on their performed tasks, shared whether they were able to accomplish their search criteria successfully, and evaluated their overall experience.Our study was approved by the university's Institutional Review Board.

Data Analysis
All interviews were video-recorded and transcribed using Microsoft Teams.The interview was the frst auto-transcribed from the recording and then for better context, the interviewers manually added descriptions of sections where the participants shared their screen but did not verbalize the task performed.Two of the authors employed open coding to identify diferent themes using word-level and sentence-level qualitative data analysis techniques [84].The generated codes were validated by the inter-rater reliability test.Common themes included transition goals (e.g., job-related goals, housing goals), challenges and fears, search resources, search strategies (e.g., contextual search, targeted search), and positive and negative user experiences.These open codes were then discussed among the research team to identify commonalities and diferences between ID, AS, and NT participants.We identifed how the ID and AS participants had diferent goals and priorities for their future independence as compared to NT participants.Furthermore, we also recognized how the success or failure of the intended online search afected these two groups diferently.With the newly surfaced themes, we re-analyzed the contextual inquiry part of the interview with a focused coding technique to identify diferent search strategies attributing to diferent goals and priorities, as well as patterns in response to the success or failure of search by these two diferent groups.The identifed overarching themes are presented in the results section.

RQ1. Infuences of goals and interfaces on search behavior
To identify the participants' independent living and transition goals, we asked them what the most important factors were as they searched for a job or housing and what they were most excited about for their future.Then we studied whether the existing websites' interfaces support these user goals and requirements, and if they do, whether the participants were able to actively incorporate their goals in their search strategies (RQ1).When reporting themes brought up by participants, we report the count of participants with autism as (AS=), those with intellectual disability as (ID=), and neurotypical participants as (NT=).
4.1.1Job-related goals and search paterns.The fndings showed similarities in the job-related goals between ID, AS, and NT participants and diferences in their priorities among those goals.Both ID and AS participants saw employment as an opportunity for self-actualization while NT participants prioritized fnancial goals.
Aligning with prior research [11], prominent job-related goals for ID and AS participants included self-actualization (n=6; ID=4, AS=2), job roles (n=5; ID=3, AS=2), and social factors (n=3; ID=2, AS=1).The self-actualization through work included "using knowledge and skills on cars" (ID4), fnding a job that "I would be good at and would match my strengths" (AS2), and trying "to be the best positive infuence and be the best I can be out there" (ID1).In the case of NT participants, the major themes in job-related goals included fnance (n=6), job roles (n=4), self-actualization (n=3), and job safety/security (n=2).While some NT participants mentioned self-actualization as their long-term goal, the participants that we interviewed prioritized total fnancial independence as the frst step to independent living and aimed for the job roles they would be best at.For example, participants mentioned, "I want to make as much money as I can [...] to do with entrepreneurship in life" (NT2).
Although participants with ID (ID=3) and autism (AS=2) also mentioned the fnancial aspect, their focus was not on making the most amount of money possible, but rather on making ends meet by getting "paid a decent amount" (AS2) and gaining the fnancial fexibility for casual things such as "comfortably buy takeout food" (AS1) and "help mom and dad with bills" (AS2).While AS1 mentioned that "not making enough" is a concern, she was also afraid of "getting complacent enough that I'm not necessarily continuing to strive to better myself, and being in a position where it's just sort of like I may not necessarily feel challenged, but I'm doing it because it pays the bills." Another priority of ID and AS participants was social factors (n=6; ID=5, AS=1), which included making new friends (ID=2), fnding/getting along with roommates (ID=3), and social acceptance (n=2; ID=1, AS=1).These results confrmed prior fndings that ND young adults see employment as an opportunity to grow and be valued members of society [11,85,95].While all participants listed expected goals for searching for jobs, many of these goals were not refected in their actual search during the contextual inquiry.For instance, several participants (ID2, ID3, AS1, NT4) had named social aspects as important criteria for a job-"I'm just looking for like a welcoming climate." (AS1), but could not consider those factors when looking through job postings online because the information was often invisible.Instead, they focused on job characteristics that were directly displayed on the page either as contextual tags or bullet points such as the job type (full-time, part-time), schedule, requirements, and qualifcations (See Figure 1b).Although one of the websites provided the social aspects such as scores on the company on factors like inclusion, learning, and belonging, none of the participants had referred to these scores when evaluating the job positions because the information was available on the company's profle page rather than directly on the search results.Diferences were found in the detailed job-search criteria mentioned by the two groups.Three ID participants (ID3, ID5, ID6) named the location as a critical criterion for an acceptable job while all of the NT participants were very fexible on the location.In their search strategy, NT participants revolved their housing search based on their fexibility to relocate for work.ID5 shared that she wanted to work at a daycare that is close to home since her mom drives her to work and "she didn't wanna drive over 30 to 40 minutes to get to a daycare." ID3 mentioned that she used public transportation so it was important for her work to be near a station or close enough to use a ride-sharing app (e.g., Lyft, Uber).ID6 named location as the most difcult aspect of the search as she wanted more details about where she would be working at.These fndings align with prior fndings that ID groups are more likely to consider location and transportation in their employment searches compared to NT groups (n=9; ID=4, AS=2, NT=3) [50,95,97].This diference in priorities was also refected in their search where ID5 used Microsoft Bing map to search for "Daycare jobs near stone mountain." NT participants used job postings from various job listing websites (e.g., Google Jobs, Linkedin) without ever using a map in this context.4.1.2Housing-related goals and search paterns.Housing-related goals were similar for ID, AS, and NT participants, who both named proximity to various places (e.g., work, grocery stores), safety, and having good roommates as top priorities.While proximity to family was never mentioned by NT participants, several ID participants mentioned it as a critical factor.NT participants mentioned having fexible housing criteria (e.g., willing to change budget and location based on the result) whereas more ID and AS participants had a fxed set of criteria on budget and location which refected in their search.Furthermore, ID participants specifcally mentioned social aspects as an important goal for their housing search more often than AS or NT participants.For example, two ID participants mentioned a need for a common area or lounge to "meet new people" (ID2) and do leisure activities like "drawing and reading a book" (ID1).This attribute conforms with Hall et al. 's fnding that people with ID develop feelings of belonging to artistic social spaces that have the potential to re-inscribe their social inclusion [43].
In housing search, we also found that the existing interfaces often did not efectively support users in fnding information related to their goals.Safety was named as an important factor when searching for housing by around half of the participants (n=9; ID=3, AS=1, NT=5).Even though websites like Apartments.com had a separate tab listing user reviews and ratings, none of our ID and AS participants consumed this information or correlate it with possible reviews on safety.Only three NT participants checked for safety of the region when they were assessing an apartment's quality, which required the use of external resources.NT4 and NT5 read through Google reviews of the properties and searched for safety-related keywords, and NT8 used a combination of Google Maps and a crime map site to check whether certain apartments were in a safe region.Instead of searching for their initial criteria, participants focused on apartment characteristics that were displayed on the summary panel (e.g., rent price, number of bedrooms) or those that were listed as drop-downs or flters (e.g., move-in date, amenities).In other words, they engaged in an interface-driven search rather than a goal-driven search.The infuence of the interface on the search behavior was also shown by the heavy use of the auto-completed and pre-flled search terms.All but one participant (ID1 who terminated the study early) used these guided search terms, with an average of 55% of the search terms of the ID and AS participants and 29% of the search terms of the NT participants being auto-completed or pre-flled by the interface.
While all the ID and AS participants were looking for transition in a region they were already familiar with, four NT participants had transition goals of moving to a new city they had never lived in before.We noticed that the locality of transition could infuence the search priorities as transitioning to a new city requires more contextual knowledge.When looking for an apartment in a new city, NT5 desired to get additional social and contextual information such as the safety of the neighborhood and information on public transit.While NT participants prioritized social information only when moving to a completely new location, ID and AS participants valued this even in local searches (n=5; ID=3, AS=2).For example, when asked about moving to an independent living situation, a participant with ID (ID5) pointed out that she would like to fnd an apartment that is "not too far away from my mom [...] right in the middle and the city." As mentioned in Section 3.1, participants lived in a metropolitan area where transportation could be a challenge, so proximity to public transit and to family was an important requirement for these participants.When looking for social information, many NT participants relied on external resources (e.g., social network channels like WhatsApp and Facebook groups of people with similar transition goals) whereas ID and AS participants solely focused on the information presented on the site that they were using.

RQ2. The strengths of users with intellectual disability and/or autism in web search for transition
The contextual inquiry helped us identify the strengths of ID and AS participants when using housing and employment search websites (RQ2).We identifed three unique strengths of our ID and AS participants as they searched for housing and jobs.First, they displayed attention to detail, especially when establishing criteria for housing.Several ID and AS participants (n=4; ID=3, AS=1) shared a comprehensive list of amenities when asked what they were looking for in an apartment.For example, ID3's list started from how the electricity bill is paid, to having a stable wi-f connection and a TV, a good washer and dryer, to having a toilet, a sink, and a bathtub that works.Another participant (AS1) took note of an apartment's amenities and furniture while browsing through the pictures of an apartment and commented that although "it's furnished [but] I'd always have to [bring] my own couch.This is not something that they provide." ID and AS participants were also more likely to notice when a criterion was missing -"They don't have the common space thing.That's defnitely.I don't see that." (ID2)showing their adherence to their initial plan.On the other hand, NT participants never commented on the misalignment between their initial searching criteria and what was available on the sites.Instead, they often based their search criteria on the features that were available on the system, and two of them shared how missing the details could lead to regrets when fnding housing.For example, NT4 had a strict requirement of "don't want to go to the laundromat" and wanted laundry as an in-house amenity but did not actively seek this information while searching-"This is a bad thing on my part now.Now that you've reminded me, I will make sure I do that." Similarly, NT2 shared a past experience when he "actually did one mistake [...] I did not see where they [train] stops on this route."As a consequence of missing this detail, he ended up walking for 15 minutes to reach the nearest train station every day.
The concrete checklists for housing and employment also unveiled an opportunity for ID and AS participants to get answers during their search without the fear that it may be too "intuitive" and not worth explaining or exploring, which they often struggle to deal with in the real world.AS1 shared a detailed description of how fnding out what is normal or acceptable was difcult in her everyday life both for housing search and at work, "I remember I had questions galore when I was [seeing] this other place, and they were so surprised that I had those questions that they didn't actually answer them and just sort of like left me in the dark and made me feel really stupid." On online platforms, one can check all the details at their own pace without getting judged.
Furthermore, ID and AS participants were able to learn and use a website creatively and efectively once they got familiar with it.ID2 said while looking at the housing website shown in Figure 1a, "It's pretty exciting.It's just overwhelming cause there's a lot on the page and you have to learn how to use a website, [but] I get a pretty good hang of websites easily cause I tried to not rush it.Even if there's a lot on one page, I look at what you can do with it, you know." Two ID participants (ID2, ID4) even repurposed the sites that they were familiar with to meet their specifc needs that are not addressed by the existing job and house-search sites.For example, one of the primary goals of ID participants that was difcult to identify on existing interfaces was fnding a housing location near public transit.One participant (ID2) overcame the issue by using a public transit website for housing search instead of a site that was specifcally designed for housing search.Another participant (ID4) used an automobile company website that he was familiar with to look for a job opening as it matched his goal of using his skills and knowledge of cars in his work.Although reaching the job opening page required him to go down several levels of menus and sub-menus, he was able to navigate the site successfully because he had a good mental model of how the site was structured.During the contextual inquiry, all ID and AS participants were able to easily use features such as flters and sorting options showing high usability of the current websites.For example, when asked to search for a two-bedroom apartment, all participants immediately used the relevant dropdown menu to change the number of bedrooms.A few usability issues were identifed during the study including the visual complexity of the map feature and users entering the available budget as the minimum price instead of the maximum price.However, most usability issues were minor and quickly fxed without any prompting.Our results align with the prior fndings that with sufcient experience and training, neurodivergent people are capable of using technology and search engines efectively [44,47,57].
Lastly, four ID participants felt a sense of achievement through the search process.Before conducting the study, we hypothesized that the search process could be a hurdle for them to overcome on their way toward independence.ID5 who shared that she had found a job all by herself, was asked a follow-up question on whether she would have liked to have someone to help her.She frmly answered, "I like searching by myself because I know that I am very, very, very independent." ID6 explained her feelings about the house search process as "proud and nervous at one time." In our study, we saw that for our ID participants, the search process in itself was an opportunity to show independence and boost self-confdence.This shows the importance of designing interfaces that support independent search for the ND population to empower them in their frst step towards independent living.

RQ3. The challenges and coping strategies of neurodiverse users
Despite the benefts of online search mentioned in the previous section, the current online search process presented challenges and barriers for AS, ID, and NT participants.We also observed how diferent users employed diferent coping strategies to overcome these challenges, and what kind of impact it had on their overall search experience (RQ3).

Overwhelmed by visual complexity.
A key barrier in online search was the visual complexity of the search results.While several participants noted that the initial number of pin icons on the map on the housing search websites (as shown in Figure 1a) was overwhelming (n=7; ID=3, AS=2, NT=2), NT participants were able to quickly adjust their search by drawing a boundary around the region of interest or fltering for characteristics to trim down the number of pin icons.In contrast, ID and AS participants preferred not to engage with the map feature in their search due to the visual complexity: "I don't understand the little tools on the map either.So it's like, it's like a lot for one page" (ID2).ID6 shared that she found the interface overwhelming and "It's giving [her] a migraine." Thus, even though visual complexity was a nuisance for both groups, it placed a higher barrier for ID and AS participants.
Our study also showed how the layout of the search results could afect people's preferences between search websites.Three ID participants indicated that they preferred the housing-search website with a list layout over the one with a grid layout.ID4 even refused to use the job-search website that we suggested, which had a grid layout, saying that it was overwhelming.ID5 also had a similar preference.When she referred to a public transit website for housing search, she said "I like how they give me [information] in bullet points.It's easy to fnd.With Zillow [which has a grid layout by default], I don't know where to go.It's all over the place." (ID5) We hypothesize that the preference for the list layout is because of its clear linear order in which the user can consume the information on a page whereas a grid-based layout is more compact but visually more complex with no apparent order.

Diferent responses of ND and NT participants to unexpected or inaccurate results
. The search behaviors also demonstrated a noticeable diference in how the NT and ND users interpret and respond to search results, especially the unexpected ones.Most websites for housing and job searching display results that are exact matches as well as broader matches to the search query.For example, a preschool teacher position may appear when searching for a position at a daycare or jobs in San Francisco may appear when searching for a job in San Jose.NT people were largely unafected by the broader set of results as they glossed over the unrelated results and focused on the ones that matched their interests.One NT participant (NT4) even found this as a good opportunity to discover jobs that he did not explicitly search for, but could still be a good ft.On the other hand, ID and AS participants (n=3; ID=1, AS=2) expressed their frustrations or discouragement on how irrelevant search results were appearing when they had entered precise search terms: "One thing I don't like about Indeed [...] is how even if you like to choose a certain destination or a certain distance, it still gives you things that aren't results[, those] that might not be exactly in that area" (ID2).
A diference in coping strategy was found when the two groups encountered an unexpected situation such as getting an empty results page.NT users were most likely to undo their last action (e.g., resetting a flter, going back to the previous search term).For example, when NT4 opened a job listing that did not directly match his search criteria, he casually perused the details, identifed its irrelevance, and moved on to the next most relevant option.On the other hand, a common coping strategy of ID participants was restarting the search process from the very beginning.At other times, ID participants who struggled to move forward in their search when they encountered an unexpected situation, switched to a diferent website altogether.For example, ID5 started the search process by entering the name of a popular job search site on a search engine and clicking on the frst result.When the home page displayed the log-in screen instead of the job search page that she had expected, she switched her search strategy to searching for jobs directly on a general search engine instead of the job search site.Another ID participant (ID3) ran into an empty results page during her search when she entered $1,500 as both the minimum and maximum value for the price range on a housing site.After exclaiming "Wow, no results found!" she tried to fx the issue by changing the values.Since she kept on getting the "No rentals found" page after adjusting the values, she quit the search on the site with the phrase, "this is ridiculous."The most extreme case was seen when ID1, who tried to search for an apartment near his school, had to stop in the middle of the user study because he was emotionally overwhelmed and frustrated by the search process that did not work as he expected.The situation was exacerbated due to technology issues as his computer was freezing during the search, further delaying the search process.These responses of participants to unexpected or inaccurate results highlight the importance of having a clear and consistent interface for the search process.

Missing implicit social and contextual cues.
In line with previous research that showed that subtle social and contextual cues can be difcult for people with autism and/or intellectual disability to interpret [90,91], two ID (ID2, ID5) and two AS participants (AS1, AS2) in our study sometimes interpreted search results at face value.One example is the use of posting dates for job postings.A posting date of "30+ days ago" sends a cue that the posting is not recent and that the employer might have already found a candidate or might not be actively looking anymore.NT users leverage the cue by either skipping those postings or lowering their expectations of hearing back based on prior experience or knowledge.However, ID5 commented that she did not know why the posting date was present on the listings, and later shared her previous dilemma of not hearing back from most of the places she had applied to.Another negatively interpreted signal surfaced during the employment search when AS1 came across a page where most of the results on a page had the tag "Not qualifed."She shared, "when you see like this, you don't necessarily want to apply for anything.[...]You know, like as soon as you see someone say, like, not qualifed, you're just like, OK.There's no need to bother them.But then there are a lot of places where people are not necessarily qualifed, but they still apply and get it." As such AS1 was discouraged by the search results and demotivated from further searches.Instead of changing the search term or going to the next page of results to fnd more jobs that she might be qualifed for, she dwelled on the result: "it [the search result] makes me feel less confdent as an engineer [...].I don't have the same experience other people do.So like, there's nothing that can beat experience.So [I'm] basically looking for a place that [...] aren't necessarily belittling me because of it" (AS1).Overall, we saw stronger emotional responses from the ID and AS participants than from the NT participants during the search process.As we mentioned in Section 5.2, the search process for participants with ID and/or autism in our study was not just a process of looking for information but often their frst step towards independence.Therefore, the overwhelming and discouraging search experiences can have a great impact on neurodivergent people's confdence, self-efcacy, and perspective about their capabilities in successful transition.

DISCUSSION
The search patterns and the stories shared by our participants provided a vivid illustration of how products can meet most usability standards but still lack utility.For example, the social motivations and needs (e.g., safety, inclusivity) of our participants were inadequately addressed in current job and housing search websites whereas most practical goals (e.g., fnancial, job roles) were well addressed.The analysis of the search patterns further showed that ID, AS, and NT participants mostly engaged in an interface-driven search, rather than a goal-driven search.In other words, they shifted their search goals based on what was available on the interface instead of following their original intended goal.These fndings endorse Shinohara and Wobbrock's call for action: inclusive design should go beyond usability to support utility [88].As we saw from our results, a successful search experience was especially crucial for our ID and AS young adults during the transitional period, as it goes beyond fnding adequate jobs and housing to gaining confdence in their frst step towards independence.In this section, we present two design suggestions to better align the search interface with user goals and increase the utility of job and housing websites based on our fndings.

Increasing the predictability and relevance of the system
One of the main issues our ID and AS participants faced during the search process was broad matching, where the search result includes a wider pool of related search terms rather than only including exact matches.While related matches can provide additional opportunities for users who are fexible on their search criteria, the "inaccuracy" of the results caused distress in our ID and AS participants who had set expectations on what types of results they would see.ID and AS participants' strength in their attention to detail may have led to a bigger impact of the unexpected results on ID and AS users, as they better noticed the discrepancy between their search terms and the results.Nielsen et al. 's design recommendation for such systems was "to give customers only what they ask for [76]."Shneiderman's usability guidelines also endorse the importance of ofering predictable results stating that "one experience with misleading data or unexpected results could undermine for a long time a person's willingness to use the system [89]."Based on our fndings and existing design principles, we urge designers to take a strength-based approach which focuses on providing an opportunity for ID and AS users to leverage their strengths that were discovered in our study -searching with attention to detail and using websites creatively and efectively -rather than solely focusing on the challenges they face.
Websites that provide more predictable and relevant results can ofer search experiences that better align with users' expectations.One way to increase predictability is by showing exact matches by default and providing a toggle option for users who want to see a broader set of results.The relevance of the results could be supported by allowing users to flter by a wider range of options, going beyond practical aspects of the position (e.g., job title, location) to the social and contextual aspects (e.g., a quiet room, gender-ratio).An alternative way to provide more relevant results is by using an inferential algorithm to understand users' preferences through their interaction patterns and browsing history [23].While this algorithmic selection of relevant results shifts the burden of selecting relevant results from the users to the system, new ethical risks are introduced.For example, after identifying a user's cognitive condition and challenges, tools that identify new potential employees based on a company's current employee data [86] could start hiding these job opportunities from the ND user due to homogeneity and lack of representation of ND people in companies' existing workforce [20,38].Our results have shown that the display of biased and/or discriminatory results is particularly harmful for ID and AS participants, as several of these participants internalized their search results and evaluated their self-efcacy based on the search experience.Therefore, future research is needed to strike a balance between search results based on inferences from user interaction patterns and preventing discriminatory search results to optimize the experience for a wide range of ND users.
We envision a middle-ground approach between fully manual and fully automated tailoring of results by using visual signifers (e.g., labels, tags).The predictability of the results could be increased by using a tag to indicate which recommendation is within the strict query results and which is an extended option.The system can further expand the query flters and recommend based on the accepted results.By displaying all the results while visually marking the ones predicted to be more relevant, the users are given the option to make the selection without the lost opportunities or cognitive burden.This idea of visual signifers is not only limited to indicating the relevance of a result but it can also be used for providing social and contextual information, which will be further discussed in the next subsection.
Next, adaptive systems could increase the relevance of the results by adjusting the number of results shown on a screen, which tailors the level of visual complexity to the user.While all of our participants shared that they valued location-based information such as proximity to public transport or family, walk-able distance from college, or access to the grocery store, most did not use the map feature to identify this information due to its visual complexity.In such cases, an adaptive system could display a simplifed version of the map with fewer pins by default for decreased visual complexity and ask users about increasing the number of results based on their use.This would allow the users to easily assess the relevance of each search result (by checking its proximity to key places) for a manageable number of results.Such design patterns can be applied more broadly to web searches in other domains where interacting with a map is essential such as restaurant or hotel reservations where the location is important.While current platforms focus on adapting the information of the website using various methods such as content recommendation and targeted marketing [27,58], prior works have suggested adapting the user interface (UI) to accommodate diferent users [9,87].Adaptive user interfaces dynamically add, remove, and adapt UI elements and features based on user interactions.As Akiki et al. have recommended, this approach can involve end-users in the adaptation process to increase their control and acceptance of the system [9].Since our study revealed that establishing autonomy in search could increase ID and AS users' self-efcacy, it further highlights the usefulness of adaptive interfaces.Therefore, we envision that adaptive UIs, where each user sees a page with a personalized level of visual complexity and UI elements, can complement the prominent accessibility framework of universal design ("one design fts all").In fact, customization (e.g., providing options to change the number of elements and their arrangements in the interface) was listed as the second most important web accessibility guideline for people with autism [19].Providing fexibility in map visual complexity is also compatible with the "competency-based" approach that Bayor et al. took in their co-design study with people with intellectual disability [12].Their results showed that ID participants had gained competencies (e.g., common web navigation skills, functional association with icons) through their use of mainstream technologies.Furthermore, leveraging these competencies in an app's design enhanced their confdence in using it, as well as their engagement.Given that our participants valued location information in their search and many people frequently use map-based applications, having the option to customize the visual complexity of the map could enable ND users to fully beneft from their competencies when using the map, rather than ignoring the feature entirely.

Enhancing the visibility of social and contextual information
Existing HCI and neurodiversity research has shown the importance of making social and contextual information visible such as others' facial expressions of emotions to efectively support social interactions of ID and AS people [16,17,36].Our research fndings contribute to this line of research by uncovering three types of under-utilized social and contextual information in current search interfaces that can facilitate a successful search for ID and AS users.

5.2.1
Implied or subtle information on the interface.Social and contextual cues in search websites such as job posting dates or reviews on safety were often overlooked by our ID and AS participants during their job and housing search process.To better support ID and AS young adults' information search and interpretation during their preparation for the transition, we recommend making social and contextual information on the websites more visible through visual signifers.For example, an existing job search site (Indeed.com)has a clear signal of how likely it is to hear back from an employer by showing a "responsive employer" tag and the supporting data in the description (e.g., "Responded to 80% or more applications in the past 30 days, typically within 7 days.")as shown in Figure 1b.As the tag and description provide more explicit contextual information than the job posting date, this design could help all users set better expectations when applying for jobs.This is in line with the web accessibility content guideline for people with intellectual disabilities, which recommends avoiding abstract concepts and, instead, ofering concrete indicators [52].Enhancing the visibility of social and contextual cues could also be benefcial in other domains of web search where interpreting user reviews is a fundamental aspect of decision-making, such as online shopping sites or online databases for movies.For example, the system could visualize the likelihood that a review is promotional or genuine and highlight the key phrases.Prior works have shown how making the invisible social cues visible can help evaluate trustworthiness in medical crowdfunding campaign web-pages [53] and mitigate bias in information seeking for NT users [59].Our study result calls for future work to explore how surfacing subtle social cues could beneft ND users, especially those with autism and/or ID, in their various online activities.

5.2.2
Hidden or missing contextual information.Currently, tags on job and housing websites are limited to summarizing the practical aspects of the job position (e.g., pay, full/part-time) or housing (e.g., rent).Some social aspects were not included on the sites at all such as information on common areas in the apartments although having a social space to meet new people (ID2) or engage in creative work (ID3) was an important criteria for participants with intellectual disability during their housing search.Job search sites were missing location-related data such as the distance to the nearest public transportation station, which was essential information for ID and AS users.Other types of information were available on the existing sites (e.g., workplace inclusivity) but not easily accessible on the search result list or the job position summary page, which were the two main spaces on which our participants focused.Thus, search sites should more clearly display the social aspects of a job position or housing that are important to the ID and AS applicants such as diversity, inclusivity, the learning environment, and safety.Quantitative measures on diversity could be included such as the racial and gender makeup of the workplace, and qualitative evidence such as the safety of the area or the learning environment by highlighting relevant reviews.Surfacing this information would beneft both ND and NT users as they can make a comprehensive assessment of jobs and housing.

Non-incorporable social information.
Another way to support the users' goals is to allow them to incorporate their social or contextual data (e.g., school or work location), which are currently non-incorporable, into the system.Our study results showed that most of our ID and AS participants considered social aspects such as living near parents, and NT participants relied on their social network for information about safety and neighborhood.While safety was noted as an important factor in housing search, all ID and AS participants disregarded this aspect in their search due to lack of clearly visible information on safety within the same website.Some NT participants found this information externally through Google reviews.This lack of necessary information to address user goals within the context of the search violates the Tognazzini's principle of data integration which states "bring to the user all the information and tools needed for each step of the process [96]."He argued that one should not expect users to leave their current screen to collect necessary information, and the tools required to successfully interact with the system should be visible, accessible, and usable within the context of the interface [96].Housing-related websites, therefore, should show information on the social aspect of a neighborhood in a more integrated and personalized manner.For example, the interface could support a deeper understanding of the neighborhood by highlighting safety-related reviews, allowing the users to integrate their social resources (e.g., getting comments from peers on the area or property), and showing the commute information to their workplace.In addition, maps should allow the user to pin locations of their social circle (e.g., houses of friends and community) to gauge the neighborhood and tie in the social aspect of a place.This could be benefcial for both ND and NT users based on the context of their search as it will alleviate the burden of mentally or manually integrating information from various sources, which is especially helpful for the ID and AS users [90].

LIMITATIONS
Our study is limited by the small sample size.To minimize the recall bias, we had a specifc inclusion criterion -having completed or actively looking for transition within 6 months.These specifc criteria unfortunately restricted the pool of potential participants.Therefore, despite our broad recruitment calls on various online and ofine autism and intellectual disability communities, we could only recruit 6 participants with ID and 2 with autism.Despite the small sample size, our participants' transitional goals aligned with those mentioned in existing literature [15,73].Thus, the discovery of unfulflled needs of ID and AS participants during the web search could refect on the current struggles and barriers faced by other young adults with similar goals.As our study participants involved only people with intellectual disability and/or autism, it is not representative of the true diversity of cognitive abilities present in the neurodiversity movement [40].However, collecting rich data on individuals' experiences and current practices of web search through contextual inquiry helped us uncover real and existing design failures not only at the usability level but also at the utility level.Thus, we call for future exploration of the goals and experiences of broader neurodivergent groups (e.g., ADHD, dyslexia) in these domains to promote universal design.Next, all of our ID participants were young adults who graduated from TPSID, a program specifcally created to help them live independently.It is possible that our participants with intellectual disabilities were more motivated and skilled in pursuing independent living than other people with intellectual disabilities who are not in this program.In addition, our neurodivergent participants were competent in verbal communication and in using technology.Thus, our fndings may not be generalized to the entire population with autism or intellectual disability, as our results are specifcally relevant to those who have a comparatively high level of adaptive living and planning skills.Additionally, all the NT participants in our study were postsecondary students.Future work is needed to generalize this study for broader neurodivergent groups who have no training or prior experience with the search for transition and include neurotypical participants without postsecondary education since education level could impact job search priorities and strategies.

CONCLUSION
We studied the goals and web search behaviors of neurodiverse young adults when searching for employment and housing resources online to examine the usability and the utility of related web services.Our fndings revealed that current interfaces support participants in reaching their practical (e.g., fnance, job role) goals but lack information on the contextual (e.g., safety, inclusivity) goals.Furthermore, we identifed a need for more predictable and adaptable search results since unexpected search results presented various challenges to users intellectual disability and/or autism, despite their ability to successfully use the sites when they matched their mental model.Designing a positive search experience is crucial for these young adults during the transitional period, as they not only fnd adequate jobs and housing through the search but also gain confdence in their frst step towards independence.Based on the results of our study, we present inclusive design suggestions that support a range of goals and facilitate all users in their transition toward independence.

Figure 1
Figure 1: a) The housing search website consists of a map panel on the left side and a property list panel on the right side.A dropdown flter is selected for fnding "2+ Bed" apartments, and pins on the map indicate available properties.b) The job search website displays the job list panel on the left and the job position detail on the right.The job list view includes various contextual tags such as "Full-time" and "Easy apply"

Table 1 :
Summary of study participants.Based on the TPSID program enrolment criteria, the age range is reported for two participants who did not share their age.ID1's work experience was not reported as the participant ended the study early.ID = intellectual disability, AS = autism spectrum, NT = neurotypical, Apt.= apartment, Undergrad.= undergraduate.