Using off-the-shelf solutions as assistive technology to support the self-management of academic tasks for autistic university students

ABSTRACT

Having the skills to self-manage the demands of academic life in third-level education is critical to the successful completion of courses taken there. Independent study behavior is an aspect of university life that pervades all topics but one that requires the self-management of time in relation to study goals. Individuals with additional educational needs, such as autism, often have difficulty self-managing independent study. This may result in students failing to meet the standards required for successful course completion. The current study (n = 2) used a range bound changing criterion design to evaluate the effects of a behavioral intervention that included assistive technology in the form of a smartphone and wearable smartwatch. The intervention aimed to increase the duration of independent study behavior among university students with autism. The intervention combined self-management (goal setting, self-monitoring, self-recording) together with assistive tech. to prompt engagement in, and recording of, independent study behavior. Findings showed the intervention to be effective at increasing independent study duration for autistic adults attending third-level education.

KEYWORDS:

• ASD
• higher education
• intervention
• self-management
• smart device
• smartphone
• smartwatch

Introduction

Autism Spectrum Disorder (ASD), sometimes termed Autism Spectrum Condition (ASC), is a complex neurodevelopmental condition characterized by deficits in social communication and interaction and is accompanied by repetitive, restrictive behavioral patterns and interests, and a preference for routine (American Psychiatric Association, 2013). Autistic¹ individuals sit along a spectrum of condition severity, with behavioral difficulties hindering greater life independence. If left unaddressed, these behavioral absences are associated with poorer long-term outcomes in independent living (Enner et al., 2020) and employment (Scott et al., 2019). Psychologists seek to teach adaptive skills designed to increase independence across many facets of life. Independence in higher education is important and is arguably more so for the increased number of autistic individuals taking up places there (Gurbuz et al., 2019). It is difficult to provide an accurate estimate of numbers of autistic students at the third level given that many decide not to disclose their diagnosis (Hillier et al., 2018; Huang et al., 2022), additionally some may only receive a diagnosis during or after that time (Wenzel & Brown, 2014). In the UK, students with social or communication impairment (including but not exclusively autism) made up 0.7% of the students in 2018–19 compared to 0.2% in 2010–11 (Office for Students, 2019). In Ireland, 4,324 (7.5%) of new entrant students to third-level institutions disclosed disabilities while registering (Ahead, 2022). Of these, 8.3% were autistic. This was an increase of 6.1% on the previous academic year, where 4,076 new entrants registered a disability of whom 7.5% disclosed an autism diagnosis (Ahead, 2021). This pattern of increase has been seen over the years (Ahead, 2018, 2019); however, it is not clear if it represents an increase in autistic students or an increase in disclosure rates. Autistic young people are at significant risk of complete disengagement from post-secondary education and employment in the immediate 2 years after leaving school. Nearly 62% of autistic college students report significant challenges with planning and organizing, the lack of structure and predictability, and increased academic demands (Bell et al., 2017; Cai & Richdale, 2016; Jansen et al., 2017; Kuder & Accardo, 2018; Van Hees et al., 2015). Academically, these students are often exceptional; however, there is an expectation that reasonable accommodations should be made to the learning environment that supports inclusion through better time management, organization, and planning, and which incorporates the type of predictability and routine to which autistic individuals respond well (United Nations, 2006 [ratified in Ireland, 2018]).

In a review of the literature, Aljadeff-Abergel et al. (2015) found self-management interventions to be successful at promoting independence for autistic individuals across a range of domains. These interventions aim to decrease the reliance placed on others, which can be an unintended consequence of a carer-led intervention (Aljadeff-Abergel et al., 2015). A follow-up review of the literature shows the higher the degree of independence, the better post-secondary outcomes in housing and employment are for this group (Hume et al., 2014). Aljadeff-Abergel et al. (2015) described self-management as a set of strategies that provide individuals with the skills to manage their own behaviors. Self-management interventions are often delivered as part of a package and may contain one or more packaged components: (a) goal setting, (b) self-monitoring, (c) self-recording, (d) self-evaluation, (e) self-reinforcement, and (f) self-initiation (Cooper et al., 2019; Miltenberger, 2016). Self-monitoring has been the most commonly used component for children aged 9–14 years, with self-management interventions reportedly under-used in adults. This is surprising as it is this age group that is likely to benefit most from this type of intervention (Aljadeff-Abergel et al., 2015).

Self-management interventions exist along a continuum upon which the individual has a predefined level of control (Cooper et al., 2019). Typically, these interventions consist of mastered component skills that participants self-manage in some manner (e.g., self-initiation, self-monitoring, self-recording, self-reinforcement, etc.). This promotes independence for the person concerned and reduces the number of interventionists/support persons required to assist with these tasks. In a two-phase report that reviewed 31 independent studies, self-management interventions were considered “established” (National Autism Center, 2009, 2015), meaning there is compelling evidence in support of its effectiveness. These reports, and others, provide synthesized information on the level of scientific evidence that exists in support of behavioral or educational interventions used for autistic people (Busacca et al., 2015).

Applied researchers have attempted to increase the effectiveness of self-management interventions by combining them with various technologies. These studies have included the use of audio and vibrating tactile notification systems that would prompt participants to record their own behavior at specific moments in time (Boswell et al., 2013; Moore et al., 2013; Otero & Haut, 2016). Interestingly, Otero and Haut (2016) found that positive reinforcement paired with self-management resulted in higher levels of accuracy in self-monitoring. Hare et al. (2016) used a personal digital assistant (PDA) and a real-time experience sampling method to measure levels of anxiety and provide management techniques to autistic individuals. This was successful at reducing anxiety for participants and novel as they combined self-management with a mobile device. The type of technology employed by Hare et al. (2016) has moved on but may still be described as Assistive Technology (AT) with AT defined broadly as: “assistive products and related systems and services developed for people to maintain or improve functioning and thereby promote well-being” (WHO, 2016, p. 1). Parents and caregivers highlight self-management as an important focus of the newly developed AT (O’Neill et al., 2020) and it is with that in mind that the current intervention was designed.

Widespread smartphone device ownership offers research avenues that were not available up to 10 or 15 years ago. Behavioral change procedures in combination with smartphone and wearable devices have been used to teach various self-management skills with a particular focus on supporting on-task behaviors (Bouck et al., 2014; Xin et al., 2017). However, as noted by Chia et al. (2018) in a review of technology enhanced interventions for self-management, the majority of these studies were school-based interventions. In one study at the third level, Wright (2016) taught college students with autism and/or intellectual disability to self-manage college-related tasks by prompting their initiation and completion using a helpful step-by-step walkthrough displayed on the smartwatches of participants. A more recent study by the same researchers used smartphones to teach a similar cohort of students to enter and attend appointments and complete tasks related to these (Wright et al., 2020). AT, when used in this way, has the potential to act as a leveler for autistic individuals by increasing access to environments and opportunities otherwise inaccessible, one such environment is higher education (Ayres et al., 2013). These studies leverage portability, social acceptability, and discreteness offered by mobile devices as intervention tools (Ayres et al., 2013; Cohen & Rozenblat, 2015).

The aim of the current study was to increase the duration of independent study, outside of the academic environment, across academic weeks, bringing much-needed organization and structure, predictability and time management to participating students. We used a self-management intervention that included goal setting and self-monitoring/recording together with a smartphone and wearable smartwatch device and paired this with positive reinforcement to prompt the initiation of study behaviors and record adherence.

Method

Participants and setting

Two students attending a university in Ireland were enrolled in the present study. Participants were two males, aged 30 and 23-years old, with a diagnosis of autism and autism and attention deficit hyperactivity disorder (ADHD), respectively.

Participants were informed about the existence of the study either via a group e-mail sent from the Disability and Learning Support Service located at the university or from seeing information sheets, in poster form, located around campus. Both recruitment drives described the upcoming project, its goals, and eligibility criteria. All interested individuals were asked to contact the project researcher if they met eligibility criteria and were interested in receiving more information on the project. Eligibility criteria included (a) enrollment at a university; (b) having a diagnosis of autism; (c) problems managing time and tasks whilst at university; (d) in attendance at the university’s Disability and Learning Support Service; (e) currently own or have experience with a smartphone: (f) be 18 years or above. Both participants reported difficulty engaging in independent study outside of programmed lectures in addition to a history of “cramming” during exam time. P1 was in year 1 of a three-year undergraduate degree programme in Physics, whilst P2 was repeating the final (3^rd) year having failed to graduate from the Physics programme the previous year. P2 was referred to the project as a priority from the Disability and Learning Support Centre on campus.

Materials

A Readiness for Behaviour Change Assessment (Vallis, 2013) was administered to each participant as part of the initial participant screening process. The assessment asked four questions to which participants could respond “yes” or “no.” Questions were: 1. Is this behavior a problem for you? 2. Does the behavior (or lack of) cause you distress? 3. Are you interested in changing this behavior? 4. Are you ready to take action toward changing this behavior? Participant responses were then graded on their readiness for change: green equated to four yes answers; yellow (2–3 yes answers); red (1 yes answer). Both participants gave four yes answers and were enrolled in the study. Following this assessment, and contingent upon inclusion, a semi-structured interview was conducted. This interview was used to narrow the intervention and module focus, to more clearly identify and operationally define the target behaviors within a SMART goal framework (e.g., Specific, Measurable, Achievable, Relevant, and Timely). Participants were encouraged to identify a module they found most difficult or least preferred. A bespoke social validity questionnaire was administered pre-and-post intervention (Appendix A).

Google Calendar was used in the case of both participants, study session days were established for the upcoming week, within which duration of in-session independent study behavior was targeted for incremental increase (see operational definition Appendix B). Each goal included a time and day of the week to alert the participant to engage in the target behavior, and for how long (duration). We asked participants to measure study time using an if-this-then-that (IFTTT) widget accessed on their smartphone. All goals were designed to be easily achievable to begin with and to avoid ratio strain. Goals were relevant to the course of study and were set at a timely segment of the academic semester.

Each participant received a smartwatch to be worn throughout the study. It was used to provide daily vibrating and textual prompts to the user. A trigger action programming widget called IFTTT was customized and used to allow participants to record their performance via their smartphone. Widgets are buttons that are placed on a separate screen of the smartphone and functioned to record independent study duration, topic, and type of study undertaken (Appendix B). This was done with either one or two button presses. Simplicity and level of effort required to self-record behavior was an important consideration of this study. In addition, participants were provided with a paper-based diary, which allowed them to record their performance in parallel. Given that off-the-shelf technology was employed, the diary would act as a backup should there be a technological failure, allowing us to assess the reliability of the technology. No backup was needed however. Android smartphones were used by participants and were compatible with the smartwatch (Pebble). Smartphones and smartwatches were paired at the outset and throughout.

Dependent measures

Independent study behavior was operationally defined between the experimenter and participant at the outset. Independent study duration acted as the main dependent measure. Participants recorded an independent study using clearly defined beginning and end points (Appendix B). The independent study time began when participants had all required study materials present and began to engage in one of the components labeled C1, C2, C3, or C4. The study time ended when the participant ceased to engage in work associated with one of the study components. Study components included C1 reading lecture notes, defined as follows: looking at each of the words contained in a sentence until all sentences on a slide had been read. This component included making notes on the same page or a separate page. Participants could choose to read these words silently or aloud. This continued until each slide had been read/completed; C2 additional reading, defined as follows: identifying a relevant topic in the index section of a textbook or journal, opening the textbook or journal to the relevant page, and reading a minimum of two relevant paragraphs. Each of the words contained in a sentence could be read silently or aloud until the paragraph ended; C3 solving a physics or mathematics-related problem, defined as follows: identifying or being assigned a problem related to a specific study topic and attempting to document a solution; C4 working on continuous assessment, defined as follows: any work assigned by a lecturer, to be conducted outside of a lecture environment, which contributes to a proportion of the marks for the module under study. It was made clear to participants that they did not have to complete each component in a sequence, but rather components could be completed in a “stand-alone” fashion. For a study session that included two or more study components, participants were instructed to record these separately using the corresponding widget on their smartphone, i.e., participants would end their current study session type (e.g., C1), press the widget for a second time, and begin a new type of study session (e.g., C3) using an alternative widget installed. Combined study time spent on multiple study components collectively contributed to the overall study session duration. The study behavior was concurrently measured using a paper-based diary.

Design

A single case experimental design called a range bound changing criterion design (RBCC; McDougall et al., 2006) was used to measure participant performance throughout this intervention. This experimental design uses baseline logic. Bias is largely controlled for as participants are exposed to the control condition (Baseline) and intervention condition in subsequent phases. This approach enhances internal validity as participants serve as their own control mitigating the extent to which other variables may influence their performance. Stability of performance (trend, level, or variability) criteria were applied before a new phase was introduced. This was of paramount importance as each phase served as a baseline for the subsequent phase (see Ledford & Gast, 2018 for a fuller description).

In keeping with Klein et al.’s (2017) suggestion, a minimum of three changes in criterion were planned for both participants. Specifically, as well as the baseline phase, there were four intervention phases, resulting in five study phases in total. This design allowed for graduated increases in the study duration criterion across the four adjacent intervention phases up to an optimal study duration. RBCC uses a criterion range consisting of both a lower and an upper limit which was systematically increased across the four study phases. Participants were asked to remain within the set criterion range. Importantly, participants were instructed to have matched or exceeded the lower limit set in each phase in order for that goal to be counted as completed. If a participant did not match or exceed the lower bound criterion, then that study session would not contribute toward the completed total for that week. Participants were not discouraged from exceeding their upper range criterion; rather, if participants found that more time was required to meet the needs of the course, then they should contact the experimenter to agree to a new upper criterion limit. Experimenters were keen to prioritize the attainment of lower study goals over upper ones. This was to ensure that should demands on a course increase in the short term, participants would not feel restricted in any way to meet these.

Procedure

Pre-baseline training

Prior to baseline, a number of assessments were carried out with each participant: Readiness for Behaviour Change; semi-structured interview, and social validity assessment. A task analysis was used to analyze the components that were required to be completed in order for participants to measure study duration using the IFTTT widget. A modified behavioral skills training (BST) procedure was used by the instructor (1. model/instruction, 2. practice, 3. feedback) during training. The instructor (first author) modeled how to acknowledge the prompt using the smartwatch and then to begin and end recording of independent study using the IFTTT widget on the smartphone (Lancioni et al., 2014). Participants then practiced by role-playing and recording their own behavior with the experimenter. Initially, the experimenter immediately modeled (0-s prompt) each component step in a total task chaining presentation. A 5-s constant prompt delay (CPD) strategy was combined with the BST procedure to fade the modeled prompt (if needed). Participants received feedback on their performance and opportunities for more practice. Participants were deemed to have mastered the task sequence if they engaged in each step of the behavior chain at 100% correct for two consecutive training sessions. Sessions took place on a single day at a meeting between the experimenter and an individual participant and lasted approximately 5 min with 5 min separating each session. It should be noted that both participants mastered the response chain following one model of the total task.

Baseline

During baseline, participants were required to record the frequency and duration of their study behavior using the IFTTT widget installed on their smartphone and in their daily self-management study diary. Both participants were instructed to treat the baseline as “business as usual” and to continue to engage in independent study as they would typically. Goal setting and reinforcement were not part of the baseline phase; however, for the purposes of capturing an accurate baseline of the target behavior, participants were asked to record their study duration (self-monitoring & recording).

Intervention phases 1–4

An outline of intervention phases can be seen in Table 1. Before commencing the intervention phases of the experiment, the experimenter and participant jointly agreed on a time of day to engage in an independent study and criterion range consisting of a lower and upper goal limit of study duration. This process was repeated across designated ”study days” of that week. A wearable smartwatch, linked to Google calendar via the participant’s smartphone, was used to prompt each participant to engage in independent study. This prompt consisted of a text message “e.g., time to study physics lecture notes,” in conjunction with a salient vibration pattern on the smartwatch. Immediately prior to the commencement of Phase 1, participants again practiced engaging in independent study following delivery of a prompt to the smartwatch. This was done in the same manner outlined for the baseline. Independent study duration was the main dependent measure. During each intervention phase, participants self-selected goals, self-monitored and recorded their independent study duration using the IFTTT widget on their smartphone.

Table 1. Outline of study phases. Actual times in minutes for phases 1–4 were tailored to the individual based on their baseline behaviors as well as the study time goal set.

Intervention phase	Time spent studying	Participant 1	Participant 2	Criteria for phase completion
Baseline	Time spent studying pre-intervention was measured.	Mean time studying: 0 mins	Mean time studying: 21.3 mins	Continued until - a stable pattern emerged - at least one week had passed - the participant expressed readiness for change
Phase 1 Occurring after SMART goal setting and reinforcement contingency put in place.	Target study range were tailored for each participant and set slightly above average baseline study times.	Target time per session: 25–60 mins Target number of sessions per week: 3	Target time per session: 30-60mins Target number of sessions per week: 3	Continued until - a stable pattern emerged over at least three consecutive sessions - the participant expressed readiness for change
Phase 2 Commenced following Phase 1	New target study time range was set 12–16% above Phase 1 levels.	Target time per session: 33–60 mins Target number of sessions per week: 4	Target time per session: 30–60 mins Target number of sessions per week: 4	Measured until - a stable pattern emerged over at least three consecutive sessions - the participant expressed readiness for change
Phase 3 Commenced following Phase 2	New target study time range was set 12–16% above Phase 2 levels.	Target time per session: 38–60 mins Target number of sessions per week: 6	Target time per session: 35–60 mins Target number of sessions per week: 4	Measured until - a stable pattern emerged over at least three consecutive sessions - the participant expressed readiness for change
Phase 4 Commenced following Phase 3	New target study time range was set 12–16% above Phase 3 levels.	Target time per session: 45–70 Target number of sessions per week: 5	Target time per session: 45–60 mins Target number of sessions per week: 3	Measured until - a stable pattern emerged over at least three consecutive sessions
Reversal Commenced following Phase 4 to demonstrate experimental control.	New target study time rage approximated baseline.	N/A	Target time per session: 25-40 mins Target number of sessions per week: 3	Measured until - a stable pattern emerged over at least 3 consecutive sessions

At commencement of the intervention phases, it was important to set achievable goals. Therefore, Phase 1 study time criterion was a target range for study time set slightly above that of the average time the participant demonstrated during baseline. Once participants showed stability in responding consistently with the Phase 1 criterion range across at least three consecutive sessions, they progressed to Phase 2. In Phase 2, the criterion range was increased so that the participant was required to spend longer studying. We used a criterion size change rule of between 12% and 16% (the extent to which the criteria were increased in the subsequent phase) where possible. Progression to Phases 3 and 4, and the target study time during each, was similarly determined. The incremental increases in time spent studying across intervention phases worked toward gradually achieving the participant’s self-selected goals and academic workload. If it became evident that the range during any phase was either too low or too high, the experimenter and participant would meet to discuss changing the criterion range in order to ensure that the targets were attainable. Although participants were instructed to remain within the study time range criterion in each intervention phase, they were not penalized for exceeding their upper limit; rather, they were asked if they would like to meet with the experimenter to modify their selected goals to better represent the demands of their course.

Feedback and reinforcement

A feedback graph, depicting their weekly performance, was sent to participants at the end of each week. This graph depicts their performance relative to each of their individualized goals. Participants also met with the first author once a week for approximately 1 h. During this time intervention goals were compared against actual behavior. While the feedback provided weekly during the intervention might act as both a reinforcer for previous behavior or prompt for future behavior, monetary rewards were also used during the intervention phase. This was employed both as a positive consequence for reaching the targets set but also as a means of retaining participants across the 3 months of the study. Monetary rewards were equally divided across weekly study goals. Contingent upon meeting or exceeding the minimum study duration limit, the goal was recorded as achieved with the monetary value in question allocated toward the total weekly earnings for that participant. Accumulated earnings were communicated to participants together with visual weekly feedback graphs at the end of each week. Participants could earn a maximum of six euros in vouchers at the end of each week and could choose whether to “cash” them in or to delay this until the end of the study. Both participants chose to delay cashing in until the conclusion of the study.

Reversal

The final phase of this experiment or P2 was a reversal phase. A bi-directional change was applied in order to test experimental control. Specifically, the criterion was set to approximate baseline performance levels during this phase. The inclusion of a reversal phase, where the behavior reverts to an earlier performance level, provides stronger evidence that the change in behavior is due to the intervention rather than to extraneous variables.

Generalisation

Following completion of the intervention phase of this study, which ran from mid-February to mid-April, both participants opted to apply intervention strategies to three additional academic modules (P1: Thermal dynamics, Electricity, Universe to atom; P2: Nonlinear dynamics, Digital signal processing, Continuous assessment) in preparation for end-of-year assessments. Generalization coincided and complemented revision time, mid-April to mid-May. The intervention targeted the completion of an independent study within a pre-agreed criterion time range throughout. For P1, this range was between 40 and 60 min, and for P2 this was between 30 and 60 minutes. The participant and the experimenter jointly set these duration goals. All other procedures were followed as outlined previously.

Social validity

Each participant was asked to rate the extent to which they agreed with 13 statements that evaluated three intervention categories: the social significance of intervention goals, the acceptability of the procedures used, and the social importance of intervention effects. They did this by rating each statement 1–5 (e.g., 1 – strongly disagree, 2 – disagree, 3 – neutral, 4 – agree, 5 – strongly agree). The social validity questionnaire was completed before and after the intervention (Appendix A). Both questionnaires presented the same items using differences to tense phraseology. Participant responses, pre-and-post intervention, were then compared.

Semi-structured interview

Following completion of the intervention, both participants were asked to carry out a post-intervention semi-structured interview. The interview consisted of nine items. Items 1–7 asked participants to rate statements in terms of importance, agreeableness, and preference, using a Likert-type scale (e.g., 1 – strongly disagree, 2 – disagree, 3 – neutral, 4 – agree, 5 – strongly agree). Items 8–9 were open-ended, although participants were free to make comments throughout (see Table 2).

Table 2. Interview responses for each participant. Questions 1–7 involved a rating of agreement from 1 to 5 with an option to make a comment. Questions 8 and 9 were open-ended questions.

Question	Participant 1	Participant 2
	Rating = 3	Rating = 4
Q.1 The monetary vouchers I earned for meeting my goals were an important feature of this project. Please rate your agreement with this statement (1 = strongly disagree, 5 =  strongly agree)	The money wasn’t a motivating factor for getting involved in the project; although it was nice. Two factors for getting involved: one was the ability to further research in this areas with a view that maybe my involvement could help others; second, to gain some help with my studies.	There’s probably no better incentive than money in any situation.
	Rating = 5	Rating = 4
Q. 2 How important would you say it was to set SMART goals when attempting to achieve your academic goals? (1 = not at all important, 5 = extremely important)	No comment	No comment
	Rating = 4	Rating = 4
Q. 3 How important to your success was it for you to record your performance? (1  = not at all important, 5 =  extremely important)	Recording my performance and “knowing” that I did my work was important - it created accountability. I could have cheated but I didn’t, it was important to do what I said I would do.	Recording was useful as it kept you accountable
	Rating = 4	Rating = 1
Q.4 How important to your success was it to receive feedback graphs? (1 = not at all important, 5 = extremely important)	I liked seeing my performance in graphical form	I didn’t look at the graphs. I know when I was meeting my daily minimum regardless of the graphs.
	Rating = 5	Rating = 5
Q.5 How would you rate the importance of having someone independent to monitor your performance and help you to set goals? (1 = not at all important, 5 =  extremely important)	Is there an option of 6? This was really important.	I’m confident I could set the goals myself. Seeing them through would require an independent monitor however.
	Rating = 5	Rating = 4
Q.6 I found the smartwatch to be very discreet when prompting me to do the things I had planned. (1 =  not at all discrete, 5 =  extremely discrete)	No comment	No comment
	Rating = 5	Rating = 4
Q.7 How would you rate the smartwatch? (1 = poor, 5 =  very good)	No comment	It’s a useful watch. Setting up prompts/reminders is easy and convenient.
Q. 8 What would you say was the most important component of this intervention?	The prompts were definitely the most important component!	Setting goals and having someone to report to with progress. Goal setting is easy. Following up on those goals is not. It’s important to have someone to work with and make sure things are being done.
Q. 9 If you were applying this intervention on your own do you believe it would have been as successful?	A sense of accountability was important, so I don’t believe so.	I’d feel less confident about it. I set goals for myself all the time, and put aside time to do achieve those. It’s never been successful prior to this project.

Items 1–5 were rated in terms of importance. These items were as follows: The monetary vouchers I earned for meeting my goals were an important feature of this project; how important would you say it was to set SMART goals when attempting to achieve your academic goals?; how important to your success was it for you to record your performance?; how important to your success was it to receive feedback graphs?; how would you rate the importance of having someone independent to monitor your performance and help you to set goals? Item 6 read: I found the smartwatch to be very discreet when prompting me to do the things I had planned. Item 7 asked: How would you rate the smartwatch? Items 8 and 9 were open-ended questions that asked: What would you say was the most important component of this intervention? and if you were applying this intervention on your own, do you believe it would have been as successful?

Results

At baseline, P1 (see Figure 1) shows a zero level of independent study duration on his chosen target module. For all phases of the intervention, the participant and the experimenter jointly set a study criterion range consisting of a lower and upper time limit or target. Participants were asked to remain within this range throughout. For participant 1, the study criterion range set in Phase 1 had a lower limit of 25 min and an upper limit of 60 min (e.g., L25 - U60-min). The study range criteria for Phases 2, 3, and 4 were set at L33 - U60-min, L38 - U60-min, and L45 - U70-min, respectively. With the exception of the third session in Phase 2 (72-min), and the final session in Phase 4 (72-min), P1’s performance remained within the pre-set study range for the duration of the experiment. Overall, data for P1 show an increase in independent study duration in line with pre-set criterion goals from zero average minutes at baseline to an average of 53 min of study at Phase 4. Phase 3 is notable due to the stability of performance observed.

Figure 1. Session by session data for P1 indicating number of minutes of independent study duration. Baseline (BSL) and study phases are indicated by the vertical lines and upper (U) and lower (L) bounds of the range criterion are indicated by the broken horizontal lines. Baseline began in mid-February with the intervention running until its conclusion in mid-April.

P2’s target module was selected because it was reported to be the most challenging/least preferred. This participant engaged in some independent study at baseline (see Figure 2). Prior to taking baseline data, P2 had reported no independent study for this module. This suggests that making P2 aware of his lack of study behavior had resulted in the occurrence of some independent study (BSL: 22, 34, 30, 30, 33, 0, & 0 min). This phenomenon, characterized as reactivity, may be observed at baseline in studies involving self-management or that involve wearable technology as part of an intervention whereby participants behave differently due to an awareness that they are being observed (Paradis & Sutkin, 2017; Seward & Redner, 2023). Despite this, P2 remained committed to participating in this project with the aim of increasing the duration of the study. A study range criterion for Phase 1 was set at L30 - U60-min. P2’s performance matched or exceeded the lower limit criterion in all three sessions at Phase 1. Phase 2 consisted of the same criterion time range, but with one additional study session per week. This meant that the per session demands were the same as Phase 1 but the demands per week increased by 33%. After four sessions, P2 voiced reluctance to increase the criterion range so it was agreed to maintain the same range. P2 matched or exceeded the lower criterion in this range, remaining within the set study range for all but one of the 10 scheduled study sessions. The lower limit was missed by 6 min in the session third from last in Phase 2, but it was decided to continue to Phase 3 given the relative stability across the rest of Phase 2. The study range criterion for Phase 3 was set at L35 -U60-min’s and for Phase 4 was set at L45-min; U60-min. P2’s average duration of the independent study increased from 21 min at baseline to 43 min over the last seven data points of the intervention. The final phase of this experiment was a reversal phase. A bi-directional change was applied in order to test experimental control. Study criterion range was set at L25-min, U40-min. P2 met this criterion and demonstrated a reversal in performance resembling that of the baseline.

Figure 2. Session by session data for P2 indicating number of minutes of independent study duration. Baseline (BSL) and study phases are indicated by the vertical lines and upper (U) and lower (L) bounds of the range criterion are indicated by the broken horizontal lines. Baseline began in mid-February with the intervention running until its conclusion in mid-April.

Generalisation

Both participants demonstrated generalization to three new modules each. Following 0 min of study time outside of the lecturing environment in baseline there was an immediate step change in responding by P1 (Figure 3), this was consistently within his agreed criterion range (L40 – U60-min). P2’s independent study time for each module immediately increased from a baseline of 0 min (Figure 4) and chiefly remained within his criterion range (L30 – U60-min); however, a more variable pattern of responding was seen in all modules.

Figure 3. Generalisation data for P1 shows number of minutes of independent study duration across three differing modules. Upper and lower bounds of the range criterion are indicated by broken horizontal lines. Generalisation coincided and complemented revision time; mid-April to mid-May.

Figure 4. Generalisation data for P2 shows number of minutes of independent study duration across three differing modules. Upper and lower bounds for range criterion are indicated by the broken horizontal lines. Generalisation coincided and complemented revision time; mid-April to mid-May.

Social validity

A social validity questionnaire was administered to both participants before and after the intervention and responses were compared. The scale consisted of 13 items and evaluated three themes: social significance of intervention goals, acceptability of procedures, and social importance of effects. For P1 at pre-intervention (Figure 5) show ratings for statements 1–13 (Appendix A) were either scored at 4 or 5 (agree; strongly agree). Post-intervention, these ratings either remained the same (Q.11; rating 4) or increased from a rating of 4 to 5 (Q’s.7 & 12). These data suggest that for this participant, intervention goals were socially significant, used appropriate recording procedures, and positively impacted their academic life. Furthermore, P1 strongly agreed with statement 5 which asked, if they would recommend this project to others with similar support needs? For P2 (Figure 3), the reverse was true. For all statements, with the exception of statement 3, ratings reduced from 5 (strongly agree) at pre-intervention to 4 (agree) at post-intervention. Agreement on statement 3 reduced from 5 (strongly agree) to 2 (disagree). This signals that study goals selected were not in keeping with this participant’s preferences throughout, which may represent fatigue or a drift in aspirations from the outset to the conclusion of the study.

Figure 5. Summarised participant agreement ratings to social validity items (x-axis) administered pre- and post- intervention. Ratings (y-axis) were: 1 - strongly disagree, 2 - disagree, 3 - neutral, 4 - agree, 5 - strongly agree.

Semi-structured interview

Table 2 presents all of the ratings and comments given by each participant to the nine questions asked as part of the post-intervention interview. In Q.1 monetary vouchers were valued particularly highly by P2 but seen as neither important nor unimportant by P1, neither participant disagreed that monetary vouchers were a positive intervention component, with P2 stating “There’s probably no better incentive than money in any situation.” In Q.2 both participants rated the use of SMART goals highly in terms of importance. In Q.3 the agreement by both participants that recording of their performance was either very important (P1) or important (P2), with both reporting this as relating to accountability was of particular note. Divergence between the ratings was seen on Q4. The feedback graphs used were reported to be important by P1 and unimportant by P2. For Q.5, both participants rated having an independent person to monitor performance and set goals a 5 (very important), P1 added “is there an option 6? This was really important.” On Q.6, both participants agreed (P1) and strongly agreed (P2) that the smartwatch was discreet when prompting them. For Q.7, the smartwatch was rated as good (P1) and very good (P2).

Discussion

Higher education institutions have reported an increased number of autistic students that are in attendance (Gurbuz et al., 2019). While this is to be welcomed, high drop-out rates and low-grade attainment have been reported in the past, with approximately 40% completing their studies successfully (Newman et al., 2011). Research is needed in order to determine if dropout rates have decreased in more recent years. Research has begun to elucidate the challenges faced by autistic students in higher education; however, limited research showing the kinds of practices that are effective at providing the necessary support to this group, in this setting, currently exist (Gurbuz et al., 2019; Kuder & Accardo, 2018). Given this, it is timely that we now pay considerable attention to the identification of effective practices capable of supporting autistic students at university. The present study sought to use a combined self-management (goal setting, self-monitoring & recording) and smartwatch/smartphone intervention package with positive reinforcement to increase duration of independent study for autistic university students. It did this by delivering pre-agreed study goals, programmed using a freely available calendar, to each participant’s wrist via the smartwatch. Participants then recorded their adherence to these study goals using a smartphone widget called IFTTT. Weekly monetary rewards were provided. Prior to the commencement of this study, both participants reported significant problems organizing their time to include independent study and consequently seldom engaged in this behavior. This resulted in P2 repeating the final year of the degree.

Following the introduction of the intervention, both participants increased in their respective levels of independent study duration to within the criterion range. This increase was in-line with incrementally increased target study session criterion ranges set across adjacent intervention phases. Interestingly, at pre-assessment P2 reported not having engaged in any independent study for the target module; however, baseline data contradicted this report showing variable levels of independent study to have occurred. This is likely due to reactivity, which refers to the participant’s observed behavior not being representative of their natural behavior and can happen when recording one’s own behavior even before the intervention has been implemented (Paradis & Sutkin, 2017; Seward & Redner, 2023). Despite this, our findings do show that for both participants this intervention proved to be effective at increasing the duration of time spent engaging in independent study. This effect is visible in the repeated co-variation of the onset of the intervention and change in dependent measures for both. As noted previously, our intervention followed the structure of many self-management interventions (Cooper et al., 2019; Miltenberger, 2016) in that it is a package with multiple components. With a package intervention such as this, it is not clear which elements of the intervention may have been responsible for the behavior change seen and, indeed, if all components were needed. Future research could examine elements of the intervention such as goal setting, self-monitoring and recording, prompting, and reinforcement, in order to see if any elements are expendable. For example, would simply breaking the task of studying down into components and setting goals be sufficient to bring about behavior change?

Research on the effectiveness of intervention practices of this type is extremely scarce, thus limiting our ability to compare outcomes. However, Wright (2016) used a combined self-management and wearable smartwatch and smartphone intervention to increase adherence to scheduled tasks. They taught autistic college students to self-manage college-related tasks by prompting their initiation and completion using smartwatch and smartphone-based prompts. Wright (2016) tailored their intervention to the needs of their participants whilst combining AT with a behavioral self-management intervention to support autistic students at higher education. The current study focused on independent study duration rather than initiation and completion, but in keeping with Wright (2016), our technology delivered self-management intervention was effective. Hare et al. (2016) sought to provide some preliminary feasibility data for the use of a mobile device as part of an intervention for autistic individuals targeting anxiety. Both the present study and Hare et al. provide some preliminary evidence for the effectiveness of this approach to behavior change beyond a clinical or experimental context. One thing that should be taken into consideration, however, is that participants may not always have attended to prompts provided by the PDA to complete the questionnaire and then to engage in recommended relaxation techniques as they were not always situationally appropriate (e.g., whilst driving), as a consequence. Irregular contact with an interventionist contributed to this non-adherence. This highlights that, although technology may be useful in bridging the physical gap between experimenter and participant, the importance of the therapeutic alliance must be considered in the development of digital AT and therapies (Tremain et al., 2020). When asked how important it was to have an independent person monitor performance and help set goals (Q.5) both participants gave this the highest rating (rating 5) adding, “Is there an option 6? This was really important” and “I’m confident I could set goals myself. Seeing them through would require an independent monitor.” The check-ins may have served as both reinforcement for studying and self-monitoring behaviors over the past week but also as a prompt for those same behaviors over the coming week.

Autistic third-level students have reported poor organizational, planning, and time management skills to be a factor thought to contribute negatively to the higher education experience and to low-grade attainment and dropout. Similarly, Jansen et al. (2017) and Van Hees et al. (2015) interviewed college students with-and-without autism, and student counselors, on their college experience and its associated challenges. Although a number of challenges have been identified and reported on, a consistent theme to emerge pointed to difficulty handling the lack of predictability, organization, structure, and routine commonly found in life at college or university. This is not surprising given that routine and predictability are important features to be considered in autism (American Psychiatric Association, 2013) and, therefore, should form part of an intervention package designed to support these students. Indeed, this was the case with the current intervention which provided clear goals and structure for independent study, which may otherwise have appeared unstructured and daunting. In the current study, participants met with the first author once per week for approximately 1 h. During this time intervention goals were compared against actual behavior. This provided an opportunity for participants to check-in and problem solve any issues that may have arisen the previous week.

Higher education institutions are often unable to provide the tailored support necessary to meet the unique needs of these students beyond that which is provided to students with other diagnoses (Barnhill, 2016; Kuder & Accardo, 2018). The need to identify effective supports that result in a more organized, predictable, and time-efficient study week for students provided impetus for the creation of the current intervention package. Disability and Learning Support Services have finite resources, but even if resources were plentiful there exists very little in the way of empirically validated support packages for autistic students. Kuder and Accardo (2018) emphasized the importance of utilizing supports that are individualized to the unique needs of the students. They raised the question regarding the feasibility of providing such support to these students. But with AT being so adaptable, inclusion of this component provides what could form part of the answer on this issue. This intervention package was shaped directly by the unique characteristics of both participants enrolled. An individualized intervention is a strong feature of this study and is an approach recommended by others as most likely to contribute to success overall (Kuder & Accardo, 2018).

The current study is novel in its use of AT combined with behavioral change procedures to bring about meaningful change for autistic adults. This combination has seldom been reported on, least of all in a university setting (Aljadeff-Abergel et al., 2015). This study contributes to the literature by demonstrating the effective goal setting and self-monitoring intervention that used wearable smartwatch technology to increase the duration of time participants engaged in independent study, outside of the lecture environment. Individuals from this group are often capable of thriving academically; but in many cases non-academic challenges, such as poor planning and organization skills, contribute to poor academic outcomes (Anderson et al., 2017). Affinity for technology by autistic teens is high and is supported by figures that show teens from this group to engage in approximately 5-h screen time each day (Kuo et al., 2014). This affinity for technology does support the notion that technology-facilitated interventions are something that should be investigated more – especially when “buy in” from this group is likely to be high. The use of mobile devices, and wearables in particular, as part of an intervention brings with it the potential to reduce dependence on caregivers, therapists, and instructors and in so doing increase independence. This autonomy has the potential to improve opportunities for future employment and well-being overall for those that use it (Ayres et al., 2013) with calls for practitioners to identify efficient evidence-based practices that use mobile devices to support this group (Shepley et al., 2017).

While the current study was novel and effective in its approach and outcomes, it is not without its limitations. First, with a sample of only two, the findings are preliminary and further research is needed to determine if AT driven self-management interventions are of wide utility. Another difficulty is that while students logged their independent study time by pressing buttons on their smartphone, there was no independent observation of their study behavior. This may be considered a limitation by some; however, independent observation in this case was simply not feasible nor ethical, and we call on the ecological momentary assessment and ecological momentary intervention literature, of which this may be considered a part, in order to demonstrate how self-reported behaviors may be considered valid and a useful addition to the literature (Mulvaney et al., 2013; Wenze et al., 2014). The benefits of the ecological validity of the behavior occurring in its natural setting, without an observer, may outweigh the limitations of no independent measurement. Furthermore, in training self-management, it appears counterintuitive to supervise or observe the behaviors in question. The use of monetary rewards designed to function as positive reinforcement for the target behaviors also calls into question whether or not the participants were really engaging in self-management rather than having their behavior managed by an external contingency. However, the use of reinforcement in self-management interventions is not unusual (Otero & Haut, 2016) and along the continuum of self-management interventions individuals have more or less control (Cooper et al., 2019). Future research could look further at the concept of self-reinforcement in self-management interventions with autistic students. Finally, social validity measures indicate some decreases for P2 over time. This participant was anxious not to repeat final year again and was determined to study more. Therefore, he engaged in 58 separate study sessions throughout the generalization (revision) phase but reported feeling fatigued at the conclusion of the study. This could account for his reduced rating scores. However, it is worth pointing out that although post-intervention ratings reduced for P2, overall his ratings remained high (4 = agree).

Conclusion

The current application of off-the-shelf technology to provide customizable self-management AT for autistic university students is a considerable addition to our knowledge of effective interventions for this under-served population. Small incremental changes in study duration were important as they took the participant from low variable levels or zero levels of independent study and increased this to a criterion and did so in a way that allowed enough time for these changes in behavior to occur. It is important to note the important role of the experimenter in this process as both participants acknowledged that weekly check-ins were critical to their successes. Future research should explore the feasibility of scaling up such interventions, taking particular note of ways of effectively merging the computing technology and behavior change technology with the need for regular in-person contact. Researchers may also wish to evaluate the intervention package for components of greatest effectiveness to determine if all aspects of the package are needed for success.

Ethics

Ethical approval was granted by Dublin City University. Both participants gave informed consent to participate.

Acknowledgements

We would like to take this opportunity to acknowledge the support provided by The Student Development and Support Service involved in this study and to thank them directly for the help they provided particularly in regard to the recruitment of participants.

Disclosure statement

No potential conflict of interest was reported by the author(s).

Additional information

Funding

This work was supported by funding from the People Programme (Marie Curie Actions) of the European Union’s Seventh Framework Programme [FP7/2007-2013] and the charity RESPECT under the REA grant agreement no. [PCOFUND-GA-2013-608728].

Notes

¹ In this paper, identity-first language is used in keeping with the preferences of members of the autism community that we collaborate with in our research. We recognize that the issue of person first versus identify first terminology can be a contentious one, and the decision that we have made here is one made through collaboration.

Appendix A

Appendix B