Sensory processing refers to a person’s ability to receive and organize sensory information for use in everyday life (Ayres, 1972, 1979/2005; Miller & Lane, 2000). The relationship between sensory processing and behavioral patterns in everyday life has been studied extensively in clinical populations, primarily with children (Koenig & Rudney, 2010). Less understood are the sensory processing patterns of adults and how they influence their everyday behavioral patterns and activity choices.

Researchers who study the relationship between sensory processing and the activity choices of adults have concluded that sensory processing across the life course can affect a person’s quality of life, coping strategies, and activity choices. The few existing studies on this topic have focused on sensory overresponsiveness (SOR), also called sensory defensiveness. SOR is defined as a condition in which responses to innocuous sensory stimuli are characterized by unusually heightened arousal accompanied by feelings of discomfort and anxiety (Parham & Mailloux, in press). Kinnealey, Oliver, and Wilbarger (1995)  used qualitative analysis of interviews to study how activity choices and everyday life patterns are affected by the coping strategies and avoidance mechanisms used by adults with sensory defensiveness. These authors organized their qualitative analysis to reflect SOR within specific sensory systems, that is, tactile, vestibular, auditory, and so forth. Fanchiang’s (1996) qualitative analysis of the case of Dale revealed that this man’s character, sense of self, and selection of occupations across his life course were affected by his learning disability and his sensory processing needs, particularly with regard to tactile processing. Moreover, significant associations have been reported between measures of SOR and health-related quality of life, as well as mental health symptoms of depression and anxiety (Kinnealey, Koenig, & Smith, 2011; Pfeiffer & Kinnealey, 2003).

To date, the most significant contribution to the understanding of sensory processing of adults has been made by Brown and Dunn (2002), who developed the Adolescent/Adult Sensory Profile (AASP), a standardized self-report questionnaire intended for people ages 11–65 yr. The standardization sample consisted of 900 people, 92% White and predominantly living in the midwestern United States (Brown & Dunn, 2002). The 60 items of the AASP measure the respondent’s frequency of responses to specific sensations by means of a Likert scale. AASP scores are interpreted using a model derived from factor analysis to classify responses within four quadrants of sensory processing: low registration, sensory seeking, sensory sensitivity, and sensory avoiding. Responses in each quadrant are scored categorically in relation to distance from the normative mean: much less than most people, less than most people, similar to most people, more than most people, and much more than most people. The quadrants in this model are viewed as emanating from a combination of neurological threshold and self-regulation patterns that may reflect temperament-related traits that affect everyday functioning (Brown & Dunn, 2002; Dunn, 2001; Dunn & Westman, 1997). However, this model does not explicitly address responses within individual sensory systems. AASP users can examine item responses to make inferences about the roles of specific sensory systems in the everyday life of a respondent, but the AASP does not provide standard scores or other guidelines for critically assessing the involvement of specific sensory systems within each quadrant.

The goal of this study was to initiate development of a new instrument, the Adult Sensory Processing Scale (ASPS), to facilitate further research on the relationship between occupational choices and diverse modes of processing within specific sensory systems. In contrast to the AASP, the ASPS was designed to measure different patterns of responsiveness (overresponding and underresponding, as well as sensory seeking) with respect to specific sensory systems (auditory, visual, tactile, vestibular, and proprioceptive). In the ASPS, overresponsiveness is defined as a tendency toward heightened sensitivity or aversion in response to ordinary sensory input that would not bother most people. Underresponsiveness is defined as a tendency toward reduced sensitivity or lack of awareness of sensory input that most people would notice. Sensory seeking is defined as a tendency toward initiating behaviors that generate sensations with greater intensity or frequency than is typical of most people. The emphasis on diverse patterns of processing with respect to distinct sensory systems is consistent with Ayres’s (1972, 1979/2005; Ayres & Tickle, 1980) original theory of sensory integration.

If shown to be reliable and valid, the ASPS may be a valuable research tool for expanding knowledge of the relationship between everyday activity choices of adults and diverse processing patterns within distinct sensory systems. Moreover, the ASPS may also be valuable as a research tool for characterizing the sensory processing features of adults with disabilities, such as autism spectrum disorder or mental illness. Eventually, the ASPS may possibly contribute to the development of new clinical assessment and intervention strategies that address the occupational needs of adults in relation to their individual sensory processing characteristics.

The ASPS is a self-report questionnaire designed to measure behavioral responses that are indicative of sensory processing challenges in five systems: tactile, proprioceptive, vestibular, auditory, and visual. The original version of this questionnaire included 10–14 questions for each system, addressing overresponsiveness and underresponsiveness as well as sensory seeking.

The development of the ASPS was conducted in two phases: item development, which included generation of items and assessment of content validity, and instrument construction, which examined validity using factor analysis and Rasch analysis, as well as internal consistency of items in the final factor solution. This research was approved by the University of Southern California institutional review board.

Item development included item generation and content validity. On the basis of an extensive literature review and clinical expertise, we generated from previous research 59 items thought to be influenced by sensory processing and 12 items focusing on general arousal (Mehrabian, 1995). We later eliminated the arousal items to focus the instrument specifically on sensory responsiveness as it relates to choices in daily life. All items focused on the frequency of specific behavioral or affective responses to sensory experiences, rated on a 5-point Likert scale (5 = always, 4 = often, 3 = sometimes, 2 = rarely, and 1 = never).

Content validity was established using the index of item-objective congruence (IIOC). This statistical procedure, developed by Rovinelli and Hambleton (1977), is best used in test development to evaluate content validity at the item development stage (Thorn & Deitz, 1989). It requires two or more judges to rate each test item relative to a set of objectives or domains. Because the items were designed to measure sensory responsiveness within distinct sensory systems, the IIOC was computed for each item to measure its congruence with specific sensory systems. To compute the IIOC, six experts in sensory integration rated each item for each of the five sensory systems using a 3-point scale: 1 (definite indicator of responsiveness to a particular sensory domain), 0 (undecided), and −1 (not an indicator of responsiveness to specific sensory system).

An item was considered acceptable if the IIOC for that item was >.70 (Rovinelli & Hambleton, 1977). Items with an IIOC between .50 and .69 were examined individually and accepted, revised and reevaluated, or rejected. Of the original 71 items, 64 produced IIOC values ≥.70, and 3 of the sensory systems (tactile, vestibular, and auditory) produced items with strong item–objective congruence.

The 71-item questionnaire with consent form was created using Qualtrics software (Qualtrics, Provo, UT) on a secure and data-encrypted website hosted by the University of Southern California. We distributed the link to the survey using fliers, community venues, Internet postings, and personal communications. Participants between ages 18 and 64 yr who were interested in participating accessed the website. Once the participants affirmed their consent by checking the “agree” box in the consent form, the site directed them to a secure web page where they completed the ASPS. Table 1 shows the demographic information from 491 respondents between age 18 and 64 yr who completed the questionnaire. Data were exported to SAS Version 9.1.3 (SAS Institute, Cary, NC) and Winsteps Version 3.80.1 (Linacre, 2014) for statistical analysis. Sample size was determined on the basis of the suggestion by Comrey and Lee (1992)  that a minimum number of 5 participants per variable is adequate to represent and evaluate the psychometric properties of measures; therefore, with 71 items, a sample of 491 was considered acceptable.

We performed exploratory factor analysis (EFA) to identify latent constructs and the underlying structure of the items. The total number of factors was determined by examining the scree plot as well as the eigenvalues. Following Gorsuch’s (1983)  recommendations, we retained factors with an eigenvalue >1. Sixteen factors that consisted of homogeneous variables originally emerged. To perform data reduction and simplify data structure, we deleted five factors that included no more than two items. Following Comrey and Lee’s (1992)  recommendations, we also deleted items with a factor loading of <.32.

We examined item structure, construct hierarchy, and measurement precision using Rasch analysis (Winsteps Version 3.80.1). Results are summarized in Table 2. All 71 items resulted in a person separation reliability (PSR) of .89 (interpreted similarly to a Cronbach’s α), suggesting the survey had good measurement precision (Bond & Fox, 2007). Six items misfit (infit mean square [MnSq] > 1.4), and the principal components analysis (PCA) indicated an eigenvalue of 6.7, suggesting the instrument showed some multidimensionality (Pesudovs, Gothwal, Wright, & Lamoureux, 2010). Removal of the 12 arousal items resulted in minimal loss of measurement precision (PSR = .87). Although the PCA percentage of variance explained was only about 10%, the eigenvalue of 6.2 suggested multidimensionality. We examined the items separately to identify whether they appeared to reflect overresponsiveness or underresponsiveness to sensory input. The overresponsiveness items had a PSR equivalent to the full scale, and the underresponsiveness items had a somewhat lower PSR of .71, suggesting these items alone did not capture differences in participants as well as the full scale or the overresponsiveness items. Across the analyses, 6 sensory processing items consistently misfit (infit MnSq > 1.4). All arousal items and all 6 of the misfit sensory processing items were then removed before further analyses.

We again performed EFA to determine the stability of factors indicated in the first exploratory analysis. The same evaluative criteria for item inclusion were applied at this stage. The final set of items with exploratory factor analysis loadings is presented in Table 3. Results indicated 11 factors: Overresponsiveness to Vestibular Input (Factor 1), Overresponsiveness to Auditory Input (Factor 2), Overresponsiveness to Visual Input (Factor 3), Overresponsiveness to Social Touch (Factor 4), Proprioceptive Seeking (Factor 5), Overall Underresponsiveness (Factor 6), Underresponsiveness to Proprioceptive–Vestibular Input Affecting Postural–Motor Abilities (Factor 7), Auditory Seeking (Factor 8), Underresponsiveness to Tactile Input (Factor 9), Intolerance to Movement (a type of vestibular overresponsiveness; Factor 10), and Overresponsiveness to Touch Involving Textures (Factor 11).

In summary, as a result of the series of factor and Rasch analyses, we removed 23 items (12 arousal, 4 tactile, 4 proprioceptive, 1 visual, 1 vestibular, and 1 auditory) reducing the scale to 48 items. Table 3 shows the mean and standard deviation of item ratings within each factor for these 48 items.

Internal consistency was established by computing Cronbach’s α reliability coefficient for each factor and all retained items of this questionnaire. The overall internal consistency for the entire set of 48 items was .87, which is considered strong (Kline, 2000). The average α value of the 11 factors was .67, and αs for the items making up each factor ranged from .6 to .8, which is considered to be acceptable. However, two factors, Underresponsiveness to Tactile Input (Factor 9) and Overresponsiveness to Touch Involving Textures (Factor 11) fell below .6 and were subsequently removed. Factor 8 (Auditory Seeking) contained only 2 items and was also removed from the final version. The final version of the scale contains 40 items and 8 factors (Table 4).

The ASPS is a unique instrument for measuring adult sensory processing, because it was designed to measure responsiveness within specific sensory systems. Four key findings indicate that it has the potential to make important contributions to sensory integration theory and practice: It is a valid tool for identifying patterns of sensory responsiveness linked to distinct sensory systems in adults; it identifies specific patterns of underresponsiveness, sensory seeking, and overresponsiveness within sensory systems; it replicates sensory processing patterns previously identified in research on children; and it links specific sensory systems to particular responsiveness patterns in combinations that have not previously been reported.

Findings showed that 5 of the 11 factors identified by the final EFA reflect responses to individual sensory systems (vestibular, tactile, auditory, visual, and proprioceptive). The first four of these factors represent overresponsiveness. Specifically, Factor 1 contains items that indicate overresponsiveness to rotary and linear vestibular input; Factor 2, overresponsiveness to auditory input; Factor 3, overresponsiveness to visual input; and Factor 4, overresponsiveness to tactile experiences that have a social component, such as being touched unexpectedly. The next five factors (5–9) reflect underresponsiveness or sensory seeking. Factor 5 represents proprioceptive seeking; Factor 6, underresponsiveness in several sensory systems, including auditory, proprioceptive, and visual; and Factor 7, underresponsiveness to proprioceptive–vestibular input influencing movement and posture. Factor 8 represents auditory seeking, and Factor 9 indicates underresponsiveness to nonsocial touch. Although these last two factors contained unacceptably few items or indicated poor internal consistency, they are interesting in their specificity to very particular types of sensory responses within particular sensory systems.

The congruence of the factors in this study with some of the sensory integration patterns previously identified in research on children is noteworthy. For example, Ayres (1964, 1965, 1966) identified tactile defensiveness as a specific factor in her early studies and later described postural control problems in children with vestibular and proprioceptive processing difficulties (Ayres, 1972, 1979/2005, 1989). Mailloux et al. (2011)  used the Sensory Integration and Praxis Test (Ayres, 1989) and the Sensory Processing Measure Home Form (Parham & Ecker, 2007) in a factor analysis that identified several patterns that are similar to those revealed in the current study. Specifically, the vestibular–proprioceptive underresponsiveness pattern found by Mailloux et al. (2011)  corresponds to Factor 7 in the current study (Underresponsiveness to Proprioceptive–Vestibular Input Affecting Postural–Motor Abilities), and Mailloux et al.’s Tactile Defensiveness factor corresponds with Factor 4 in the current study (Overresponsiveness to Social Touch). Moreover, Mailloux et al.’s pattern of somatosensory discrimination shows similarities to Factor 9 in the current study (Underresponsiveness to Tactile Input).

The consistency of these patterns across age groups raises the question of whether they persist from childhood into adulthood. Although developmental continuity cannot be presumed from the results of cross-sectional studies, it is plausible that these sensory processing patterns persist over time. If future longitudinal research confirms this suggestion, it would indicate that children do not simply “outgrow” early sensory processing patterns, underscoring the potential role of intervention to prevent or minimize lifelong stress related to sensory characteristics.

Additional findings from the ASPS suggest that meaningful subgroups of sensory processing differences may exist that have not previously been quantified. Specifically, Factors 1 and 10 indicate that different types of vestibular overresponsiveness may exist, that is, one relating primarily to fear or avoidance of movement through space, which Ayres (1979/2005) termed gravitational insecurity, and the other relating to being easily nauseated by movement sensations, which Ayres called intolerance to movement. Similarly, Factors 4 and 11 both address tactile overresponsiveness (often called tactile defensiveness), but this study is the first to present evidence that some people with tactile defensiveness may be particularly reactive to being touched by other people, whereas others may be reactive specifically to nonsocial tactile stimuli.

The ASPS may offer a viable alternative to current instruments that measure sensory processing in adults. Initial findings indicate that it has adequate internal consistency, strong content validity, and acceptable construct validity. Although the sample size is adequate, a limitation is the sample of convenience. Future research examining test–retest reliability and discriminative validity is needed to ascertain whether the psychometric properties of the ASPS are adequate for clinical use.

Implications for occupational therapy practice and research are as follows:

We thank Thomas Decker, Cheryl Ecker, Susan Knox, Zoe Mailloux, Gustavo Reinoso, Aneeta Sagar, and Susanne Smith Roley, who helped with content validity ratings. We also greatly appreciate the assistance of Annika Buck and Denise Duran for computing the index of item-objective congruence for all items. Last, we thank the participants who answered the questionnaire on Qualtrics. Without their help, the study would not have been possible.