The Impact of Presentation Level on SCAN A Test Performance

Transcription

1 Benjamin J. Lovett Theodore L. Johnson Elmira College, Elmira, NY The Impact of Presentation Level on SCAN A Test Performance ABSTRACT: Purpose: The SCAN A (Keith, 1994) is a popular test that is used to screen for auditory processing disorder (APD) in adolescents and adults. It can be administered on a portable CD player, but this leaves the presentation level uncontrolled. This study examined the influence of presentation level on participants performance. Method: The SCAN A was administered through an audiometer at 2 presentation levels (50 and 60 db HL) to 63 college students in a between-subjects experimental design. The students each passed a standard hearing screening; none had a diagnosis of APD. Results: Participants who received the test at 60 db significantly outperformed those who received it at 50 db on each of the 4 SCAN A subtests, and there was a substantial difference in the 2 groups average composite scores. Conclusion: Presentation level appears to affect performance on the types of auditory processing tasks (monaural low-redundancy speech and dichotic listening) that are measured by the SCAN A. It is recommended that SCAN A users administer the test through an audiometer at 50 db HL. Future research is needed to determine whether presentation level also affects other kinds of auditory processing tasks (e.g., binaural integration). KEY WORDS: auditory processing disorder, diagnostic assessment The SCAN A (Keith, 1994) is a test that is used to screen adolescents and adults for auditory processing disorder (APD; also called central au- ditory processing disorder, or CAPD). APD, defined as the defective processing of auditory information in spite of normal auditory thresholds (Jerger & Musiek, 2000, p. 467), is an increasingly recognized condition. Assessment tools for APD include behavioral tests in which clients listen to standardized speech and nonspeech stimuli, usually repeating speech as directed and making judgments about the nonspeech stimuli (e.g., Bellis, 2003). In this brief research note, we report on the effects of varying presentation levels when administering behavioral subtests of the SCAN A. According to surveys of practicing audiologists (e.g., Emanuel, 2002; Hind, 2006), the SCAN series of tests (now in their third edition, with versions for children as well as adolescents and adults) is a very popular assessment tool for APD. The SCAN A consists of four subtests assessing multiple auditory processing functions. In the Filtered Words subtest, clients are asked to repeat monaurally presented words that are low-pass filtered (half of the words are presented to each ear). In the Auditory Figure-Ground subtest, clients are asked to repeat monaurally presented words against a background of speech babble noise of the same intensity (0 db signal-to-noise ratio). In the Competing Words subtest, clients are asked to repeat each of two words presented simultaneously to the ears (i.e., dichotic listening). Finally, in the Competing Sentences subtest, clients hear dichotically presented sentences and are asked to repeat either the sentence heard in the left or the right ear, as directed. One of the features contributing to the SCAN tests popularity is likely their flexibility in administration procedures in terms of equipment, settings, and professional qualifications. First, the tests can be administered using nothing more than a portable CD player and a set of headphones rather than an audiometer. Second, the tests can be administered in any quiet room rather than requiring an audiometric booth. Finally, the tests can be administered by a variety of professionals, including audiologists, speech-language pathologists, school psychologists, neuropsychologists, and educational diagnosticians (Keith, 2009, p. 6). Indeed, Ross-Swain CONTEMPORARY ISSUES IN COMMUNICATION SCIENCE AND DISORDERS Lovett Volume Johnson: Presentation Fall 2011 Level NSSLHA and SCAN A Performance /11/

2 (2007) recommended the SCAN as one of the tests that speech-language pathologists (SLPs) should consider using when assessing students for possible APD. Unfortunately, this flexibility in administration may lead to SCAN results being affected by a variety of extraneous factors. Of particular concern is presentation level. The SCAN A manual (Keith, 1994, p. 11) states that the volume control should be adjusted to the subject s most comfortable listening level, but no procedures are provided for doing this, and given the variety of professionals qualified to administer the test, examiners are likely to follow the only other guidance provided: to make sure that the recording is loud enough for the subject to hear easily (Keith, 1994, p. 11). Of course, a wide range of presentation levels will be loud enough to hear easily, and hear easily cannot mean easy to interpret because the stimuli are designed to be challenging to interpret. This means that the SCAN A is likely to be administered at a variety of presentation levels, none of which can be noted with precision if a portable CD player is used rather than an audiometer. Although the effect of presentation level on SCAN-like tasks has yet to be examined, there are several reasons to hypothesize such an effect. First, the SCAN tasks have relatively low test retest reliability. For instance, the SCAN A manual (Keith, 1994) reports a test retest reliability coefficient of.69, meaning that a substantial portion of the variability in examinees scores is attributable to the particular time the test happened to be given. One of the factors likely to vary across multiple administrations is presentation level, and so varying presentation levels may account for this relatively poor test retest reliability. At the very least, the reliability coefficient suggests that SCAN A scores are affected significantly by extraneous factors that vary across administrations. A second source of evidence, although indirect, for the possible effect of presentation level on SCAN A performance comes from the APD intervention literature. One prominent class of management strategies for APD in educational settings involves amplifying important sounds through the use of personal FM systems or classroom sound-field amplification (e.g., Geffner, 2007; Medwetsky, Riddle, & Katz, 2009). Research has repeatedly shown the effects of amplification on speech recognition skills as well as more general listening and attention skills (Crandall, Smaldino, & Flexner, 2005). For amplification to be effective, the presentation level of the stimuli must affect auditory processing, suggesting that presentation level may also affect performance on more standardized auditory processing tasks. In particular, increasing the presentation level (when it is already well above threshold) appears to aid auditory processing. Although this has not been examined systematically, researchers working with auditory processing measures have noted these effects. For instance, Musiek, Geurkink, and Kietel (1982) administered a binaural fusion task (in which different frequency portions of the same word were administered to different ears simultaneously) at 20 db above its typical administration level (50 db SL instead of 30 db SL) and found that their participants obtained scores that were considerably higher than those reported by other researchers. In another study, Christopherson and Humes (1992) found that their participants (all with normal hearing) performed better on several auditory processing tasks (e.g., frequency discrimination; speech-in-noise, etc.) when the presentation level was raised above 55 db SPL; one of the tasks showed optimal performance at 85 db SPL! The few systematic investigations of the effects of presentation level on diagnostic auditory processing test performance have generally involved gap detection tasks using nonspeech stimuli. In the most recent of these studies, Weihing, Musiek, and Shinn (2007) administered the Gaps-In-Noise (GIN) test to 10 normal-hearing adults at eight different presentation levels ranging from 5 to 50 db SL. Participants were asked to press a button each time they heard a gap in noise, and the gaps ranged from 2 to 20 ms. These investigators found that participants ability to detect gaps in the noise was strongly related to the presentation level, with the proportion of gaps correctly detected increasing with each increase in presentation level. The investigators concluded that as long as the GIN test was administered at 35 db SL or higher, clients scores would not suffer substantially, but the relationship between presentation level and performance was nonetheless marked. It is unclear, though, whether these effects of presentation level extend to auditory processing tasks with more complex stimuli. Typically, auditory processing tasks require clients to identify stimuli that are either degraded in some way, presented against a background of noise, or presented with competing stimuli. Increasing the presentation level would not make the signal any less degraded, nor would it change the signal-to-noise ratio, and so it is possible that no effect on performance would be found. Moreover, auditory processing tasks are typically presented at a stimulus intensity that is well above clients auditory thresholds. Therefore, it is possible that lower intensity presentations could harm performance, but higher intensity presentations would not help, as Weihing et al. (2007) found with the GIN test. In the present study, we sought to determine whether presentation level affected performance on a common measure of auditory processing within the range of administration conditions that are typically encountered in field settings. Therefore, we chose 50 db HL and 60 db HL as presentation levels that were within the range of comfortable speech (unlikely to elicit complaint from clients in field settings) and that were approximately as far above participants hearing thresholds as was the effective presentation level (i.e., 35 db SL) in the Weihing et al. (2007) study. For our instrument, we chose to use the SCAN A, which is a popular measure that is used by a variety of professionals to screen for APD. Based on our review of prior research, we hypothesized that the 60-dB HL group would perform significantly better than the 50-dB HL group on the SCAN A subtests. METHOD Participants The participants were 66 students at a small private college in the northeast. Participants ranged in age from CONTEMPORARY ISSUES IN COMMUNICATION SCIENCE AND DISORDERS Volume Fall 2011

3 to 27 (M = 19.9 years, SD = 1.68). They were recruited from undergraduate classes and given a small amount of extra credit in their respective courses as an incentive to participate. Otoscopy ruled out any cerumen blockages of the external auditory canal. Two participants were excluded for failing a pure-tone hearing screening, and one participant was excluded because the participant s native language was not English, resulting in a final sample size of 63 (55 females, 8 males). Materials and Equipment The SCAN A is commonly used to screen for the presence of APD in adolescents and adults. Its four subtests include two monaural low-redundancy speech measures (Filtered Words, Auditory Figure-Ground) and two dichotic listening tasks (Competing Words and Competing Sentences). The test requires minimal experimenter participant interaction because instructions are delivered along with test stimuli on a CD recording. The test takes a total time of ~20 min. Pure-tone air-conduction hearing screenings were conducted with an Interacoustics AC 40 clinical audiometer. The auditory processing tasks were administered via CD through the Interacoustics audiometer using TDH-39 headphones. Participants were seated in an IAC audiometric booth. All audiometric equipment was calibrated to current ANSI standards. Procedure Participants were provided with informed consent documents for review and signature and were reminded that they could cease participation, without penalty, at any time during the screening. After completing the informed consent procedure, participants were taken to the audiometric booth, where they were given verbal instructions regarding the otoscopic examination and hearing screening procedure. Participants, under headphones, were then provided with verbal instructions for completing the pure-tone screening and were seated facing away from the examiner. Instructions consisted of requesting the participant to raise a hand indicating that he or she heard, or thought he or she heard, a tone or whistle-like sound. A 20-dB warble tone stimulus, lasting ~2 3 s, at frequencies of 500 Hz, 1000 Hz, 2000 Hz, and 4000 Hz, was presented to each ear. Upon completion of the hearing screening, participants were told to follow recorded instructions for the SCAN A that were delivered via headphones. The four subtests of the SCAN A were presented at either 50 db HL or 60 db HL, with each participant randomly assigned to receive the test at one presentation level or the other. The SCAN A was presented at 50 db HL for 32 participants and 60 db HL for 31 participants. The total time taken for each participant to complete the entire study procedure was ~30 min. All procedures were conducted by three speech and hearing students who had previously completed audiology coursework and a minimum of two semesters of audiology clinical practicum. Data were entered electronically to an encrypted Microsoft Excel spreadsheet by the American Speech-Language-Hearing Association-certified audiologist, who also supervised the data collection. Statistical Analysis To examine the effect of presentation level on SCAN A performance, we compared the 50-dB and 60-dB groups on each of five dependent variables: raw scores from the four subtests of the SCAN A and the computed auditory processing composite (overall) score. For each of these five variables, we performed t tests, computing a test statistic, a p value (to determine if each difference between the 50- db and 60-dB groups was statistically significant), and the effect size statistic d, which describes how many standard deviations apart the two groups score distributions were (Cohen, 1988). RESULTS Table 1 presents the comparison of the two groups performance on each of the four SCAN A subtests as well as the auditory processing composite score. Raw scores were used for the subtests to illustrate the effect of presentation level on the actual number of correct items. (Scaled scores were also calculated but did not change the inferential statistical tests substantially.) Independent-groups t tests were used to compare the 50-dB and 60-dB groups on each of the five variables. Students in the 60-dB group performed significantly better than students in the 50-dB group on the Filtered Words subtest, t(61) = 2.15, p =.035, d =.54, and the Auditory Figure-Ground subtest, t(61) = 2.49, p =.016, d =.63. On the two dichotic listening tasks, students in the 60-dB group also showed significantly higher performance: for the Competing Words subtest, t(61) = 4.32, p <.001, d = 1.09, and for the Competing Sentences subtest, t(61) = 2.88, p =.006, d =.73. Finally, students in the 60-dB group showed significantly higher auditory processing composite scores, t(61) = 4.67, p <.001, d = Cohen (1988) offered what are now widely accepted conventions for interpreting the d effect size statistic, with a d of.20 indicating a small effect, a d of.50 indicating a medium effect, and a d of.80 indicating a large effect of one variable on another. By these conventions, all of our effects were of at least medium size, and some were well past the large standard of d =.80. DISCUSSION Our hypothesis that a higher presentation level would result in improved performance on auditory processing tasks was supported by the study results. Presenting the SCAN A at 60 db rather than 50 db (both levels well within the range of comfortable speech) led to subtest scores that were, on average, at least 1 /2 SD higher, and to composite scores that were more than 1 SD higher. To see the potential practical impact of these effects, consider that the SCAN A manual (Keith, 1994) views scores that are 1 SD below the mean to be questionable performance, and scores that are 2 SDs below the mean to be considered disordered. Effects the same size as the ones found in the present study, then, Lovett Johnson: Presentation Level and SCAN A Performance 137

4 Table 1. Comparison of study participants performance on the four subtests of the SCAN A (Keith, 1994) at 50-dB and 60-dB presentation levels. 50-dB group 60-dB group SCAN A subtest M SD M SD t d Filtered Words *.54 Auditory Figure-Ground *.63 Competing Words *** 1.09 Competing Sentences **.73 Composite score *** 1.18 *p <.05. **p <.01. ***p <.001. could easily mean the difference between a normal and a questionable score, or between a questionable and a disordered score, at least in a clinical sample. Although presentation level had a statistically significant effect on performance on each of the four subtests, it is interesting to note that the effects were larger on the two measures of dichotic listening than on the two measures of monaural low-redundancy speech. It may be that participants benefited more from increased stimulus intensity on dichotic tests because the target stimulus was presented in a more acoustically complete form. When the speech stimuli are altered in some way (as in the Filtered Words and Auditory Figure-Ground subtests), increased intensity may be less valuable. Overall, these results support our hypothesis (that participants receiving the higher presentation level would benefit) and are consistent with findings by other researchers (Christopherson & Humes, 1992; Musiek et al., 1982) that even when stimuli from auditory processing tasks are already being presented well above a participant s hearing threshold, performance can be improved by further increasing the stimuli s intensity. The underlying mechanism for this improvement may be the mobilization of a participant s attention; an intriguing paper by Baldwin and Struckman- Johnson (2002) argued that individuals who are already cognitively occupied are most likely to benefit from more intense presentations of auditory stimuli. Implications for Clinical Practice These results place users of the SCAN A in a difficult position. If the normative data had been collected using a standard presentation level, we could recommend that clinicians only administer the test at that presentation level. However, according to the developer s description of the standardization process (Keith, 1995), the data were collected at 21 different sites, sometimes using a portable audiocassette player and sometimes using an audiometer. In light of this, the results of the present study lead to two recommendations. First, because presentation level was found to affect participants performance, it will be important to always note the actual presentation level along with an individual s performance. Therefore, we recommend that clinicians use an audiometer to administer the SCAN A. If a typical portable CD player (or audiocassette player) is used, the presentation level cannot be determined, and so clinicians cannot interpret the scores with the presentation level in mind. Second, we recommend that clinicians use 50 db HL as the default presentation level. In the present study, we found that a 50-dB administration of the SCAN A led to a mean auditory processing composite score (100.70) that was almost exactly the same as the official scaled mean of the norm sample in the manual (i.e., 100). Because none of the students in our sample had an APD diagnosis, we would expect to obtain a mean that is approximately that of the norm sample, and the 50-dB presentation level led to such a mean. Given the popularity of the SCAN test series (see e.g., Emanuel, 2002; Hind, 2006), the implications of this study for clinical practice are considerable. We conclude that scores from the SCAN A should be interpreted cautiously in the context of the presentation level used during administration of the SCAN A. Moreover, although there is a new version of the SCAN A available (the SCAN 3:A; Keith, 2009), it includes the same subtests as the 1994 version of the SCAN A (as well as additional new subtests), and so unless further research shows otherwise, we must assume that presentation level will also affect scores on this new version of the test. Limitations and Directions for Future Research There were several limitations of the present study, each of which lead naturally to future research opportunities. First, our sample was a group of typical college students who were participating solely in research, rather than a clinically referred sample. Therefore, we do not know if, in a clinically referred sample, the difference between 50-dB and 60-dB administrations would affect whether someone was actually diagnosed with APD. Moreover, we do not know at what presentation level clinicians typically administer the SCAN A; a survey would be helpful in determining this. Second, our results raise the question of whether presentation level affects performance on other kinds of auditory processing tasks. The SCAN A s four subtests only include measures of monaural low-redundancy speech and dichotic listening. Other common auditory processing tasks include measures of temporal processing and binaural interaction 138 CONTEMPORARY ISSUES IN COMMUNICATION SCIENCE AND DISORDERS Volume Fall 2011

5 (Baran, 2006), and it is important to know how presentation level affects these as well. Finally, we only examined two presentation levels (50 and 60 db HL), and the examinations were conducted in a sound-treated audiometric booth. Future research might examine a wider range of administration conditions, including more variable presentation levels as well as settings with varying levels of ambient noise. Because the SCAN tests are frequently administered in field settings on portable CD players, the interaction between presentation level and ambient noise would be especially interesting to study. The present study, then, should be viewed as an initial examination of the generalization of the SCAN A across conditions (here, we examined presentation-level conditions). Although much research remains to be done, our finding that presentation level substantially affects participants performance on the SCAN A should lead to more cautious interpretation of SCAN A scores, and we hope that it will inspire future studies as well. REFERENCES Baldwin, C. L., & Struckman-Johnson, D. (2002). Impact of speech presentation level on cognitive task performance: Implications for auditory display design. Ergonomics, 45, Baran, J. A. (2006). Test battery considerations. In G. D. Chermak & F. E. Musiek (Eds.), Handbook of central auditory processing disorder (Vol. 1, pp ). San Diego, CA: Plural. Bellis, T. J. (2003). Assessment and management of central auditory processing disorders in the educational setting: From science to practice (2nd ed.). Clifton Park, NY: Delmar Learning. Christopherson, L. A., & Humes, L. E. (1992). Some psychometric properties of the Test of Basic Auditory Capabilities (TBAC). Journal of Speech and Hearing Research, 35, Cohen, J. (1988). Statistical power analysis for the behavioral sciences (2nd ed.). Hillsdale, NJ: Erlbaum. Crandall, C. C., Smaldino, J. J., & Flexner, C. (2005). Sound field amplification: Applications to speech perception and classroom acoustics (2nd ed.). Clifton Park, NY: Delmar Learning. Emanuel, D. C. (2002). The auditory processing battery: Survey of common practices. Journal of the American Academy of Audiology, 13, Geffner, D. (2007). Management strategies. In D. Geffner & D. Ross-Swain (Eds.), Auditory processing disorders: Assessment, management, and treatment (pp ). San Diego, CA: Plural. Hind, S. (2006). Survey of care pathway for auditory processing disorder. Audiological Medicine, 4, Jerger, J., & Musiek, F. (2000). Report of the consensus conference on the diagnosis of auditory processing disorders in schoolaged children. Journal of the American Academy of Audiology, 11, Keith, R. W. (1994). SCAN A: A Test for Auditory Processing Disorders in Adolescents and Adults. San Antonio, TX: The Psychological Corporation. Keith, R. W. (1995). Development and standardization of SCAN A: Test of Auditory Processing Disorders in Adolescents and Adults. Journal of the American Academy of Audiology, 6, Keith, R. W. (2009). SCAN 3 for Adolescents and Adults: Tests for Auditory Processing Disorders. San Antonio, TX: Pearson. Medwetsky, L., Riddle, L., & Katz, J. (2009). Management of central auditory processing disorders. In J. Katz, L. Medwetsky, R. Burkard, & L. J. Hood (Eds.), Handbook of clinical audiology (6th ed., pp ). Philadelphia, PA: Williams & Wilkins. Musiek, F. E., Geurkink, N. A., & Kietel, S. A. (1982). Test battery assessment of auditory perceptual dysfunction in children. Laryngoscope, 92, Ross-Swain, D. (2007). The speech-language pathologist s assessment of auditory processing disorders. In D. Geffner & D. Ross-Swain (Eds.), Auditory processing disorders: Assessment, management, and treatment (pp ). San Diego, CA: Plural. Weihing, J. A., Musiek, F. E., & Shinn, J. B. (2007). The effect of presentation level on the Gaps-In-Noise (GIN) test. Journal of the American Academy of Audiology, 18, Contact author: Benjamin J. Lovett, Department of Psychology, Elmira College, 1 Park Place, Elmira, NY blovett@ elmira.edu Lovett Johnson: Presentation Level and SCAN A Performance 139