The Effects of Listening Environment and Earphone Style on Preferred Listening Levels of Normal Hearing Adults Using an MP3

Player
William E. Hodgetts, Jana M. Rieger, and Ryan A. Szarko

Objectives: The main objective of this study was to determine the influence of listening environment and earphone style on the preferred-listening levels (PLLs) measured in users ear canals with a commercially-available MP3 player. It was hypothesized that listeners would prefer higher levels with earbud headphones as opposed to over-the-ear headphones, and that the effects would depend on the environment in which the user was listening. A secondary objective was to use the measured PLLs to determine the permissible listening duration to reach 100% daily noise dose. Design: There were two independent variables in this study. The first, headphone style, had three levels: earbud, over-the-ear, and over-the-ear with noise reduction (the same headphones with a noise reduction circuit). The second, environment, also had 3 levels: quiet, street noise and multi-talker babble. The dependent variable was ear canal A-weighted sound pressure level. A 3 3 within-subjects repeated-measures ANOVA was used to analyze the data. Thirtyeight normal hearing adults were recruited from the Faculty of Rehabilitation Medicine at the University of Alberta. Each subject listened to the same song and adjusted the level until it sounded best to them in each of the 9 conditions. Results: Significant main effects were found for both the headphone style and environment factors. On average, listeners had higher preferred listening levels with the earbud headphones, than with the over-the-ear headphones. When the noise reduction circuit was used with the over-the-ear headphones, the average PLL was even lower. On average, listeners had higher PLLs in street noise than in multi-talker babble and both of these were higher than the PLL for the quiet condition. The interaction between headphone style and environment was also significant. Details of individual contrasts are explored. Overall, PLLs were quite conservative, which would theoretically allow for extended permissible listening durations. Finally, we investigated the maximum output level of the MP3 player in the ear canals of authors 1 and 3 of this paper. Levels were highest with the earbud style, followed by the over-the-ear with noise reduction. The overDepartment of Speech Pathology and Audiology (W.E.H., J.M.R., R.A.S.), University of Alberta; Craniofacial Osseointegration and Maxillofacial Prosthetic Rehabilitation Unit (COMPRU) (W.E.H., J.M.R.), Caritas Health Group, Edmonton, Canada.

the-ear headphone without noise reduction had the lowest maximum output. Conclusions: The majority of MP3 players are sold with the earbud style of headphones. Preferred listening levels are higher with this style of headphone compared to the over-the-ear style. Moreover, as the noise level in the environment increases, earbud users are even more susceptible to background noise and consequently increase the level of the music to overcome this. The result is an increased sound pressure level at the eardrum. However, the levels chosen by our subjects suggest that MP3 listening levels may not be as significant a concern as has been reported recently in the mainstream media.
(Ear & Hearing 2007;28;290297)

Portable radios, cassette players and compact disc players have been popular for many years. Recently, portable MPEGaudio layer 3 (MP3) music players have been miniaturized such that users now are able to carry thousands of songs with them in an unobtrusive device. The portability of MP3 players and the availability of downloadable music have spurred record sales for these devices. The National Product Development (NPD) Group Canada market research firm reported that sales of MP3 players more than tripled in Canada between June 2004 and June 2005 with approximately 40% of Canadian households owning one (NPD Group Canada, 2005). At 40%, the NPD group predicts that this market is still in its early stages of development. While MP3 players may be a great way to pass the time on your morning commute or at the gym, there is growing concern about the potential damage caused by this sort of leisure noise. Zogby International, a polling, market research, and information service was recently commissioned by the American Speech-Language-Hearing Association (ASHA) to survey adults and high school students regarding the use of personal electronic devices and headphones (Zogby, 2006). More than 61% of teens said they owned an MP3 player. Within the teen group, some disturbing trends between statements of hearing loss symptoms and the use of MP3 players were reported. Alarmingly, 51% of the high school students had experienced at least 1 of the 4 symptoms

0196/0202/07/2803-0290/0 Ear & Hearing Copyright 2007 by Lippincott Williams & Wilkins Printed in the U.S.A. 290

EAR & HEARING, VOL. 28 NO. 3 of hearing loss listed in the survey. When respondents were asked about the volume they choose for their devices, the majority of high school students responded loud whereas the majority of adults chose medium. Noise-induced hearing loss (NIHL) is not only a function of level, but also of duration of exposure. As the level of the signal increases, the maximum allowable duration decreases. Here the adult trend created cause for concern. Roughly 43% of adults said they listened to their MP3 player between 1 to 4 hr per session. Nine percent of adults reported listening for more than 4 hr per session. Although the Zogby survey data and media reports have raised concern regarding the potential hazards of MP3 players, objective studies investigating the effects of personal music systems on NIHL have yielded mixed results. While there is certainly a great deal of theoretical risk associated with the output capabilities of personal music systems (Fligor & Cox, 2004), actual damage risk at typical use levels is less transparent with the estimates ranging from 1 in 1500 (Rice, Rossi & Olina, 1987; (Rice, Breslin & Roper, 1987) to as high as 25% (Williams, 2005). There are a number of methodological complications with studying the effects of MP3 players on NIHL. First, noise standards, such as the National Institute for Occupational Safety and Health (NIOSH; 1998) and Occupational Safety and Health Administration (OSHA, 1981; 1983) that can be used to estimate damage risk, assume occupational noise is used in the calculations. Spectral, temporal and dynamic variations in music tend to be less predictable than occupational noise. Careful consideration of the type of music stimulus (a song with minor spectral, temporal and dynamic variation) and how it is measured needs to be made for these calculations to be relevant. Second, each device has its own output capabilities and the headphones supplied are not standardized. Third, the user-preferred listening level is likely to change as a function of environment and the noise reduction capabilities of the earphones used. This study addressed the issues of headphone type and listening environment by measuring the user-preferred output levels of a song (calculated to have long-term intensity fluctuations of only 10 dB) in the ear canal after the output stage of the headphone. The objectives of this study were to determine the influence of listening environment (quiet versus street noise versus multi-talker babble) and earphone style (over-the-ear versus in-theear) on the preferred-listening levels (output levels) measured in a persons ear canal while listening to music with a commercially-available MP3 player. The following research questions were of interest:

291 1. Are preferred-listening levels significantly higher in street noise than in multi-talker babble or in quiet (main effect)? 2. Are preferred-listening levels significantly higher with the earbud style of headphone compared to the over-the-ear style with and without noise reduction capabilities (main effect)? 3. Are there any significant interactions between preferred-listening levels and environment by earphone style (interaction effects)? 4. What is the maximum allowable listening duration to reach 100% daily-noise dose with each headphone style and each environment based on the output levels obtained in this study?

MATERIALS
Subjects

AND

METHODS

Thirty-eight subjects (15 males, 23 females) with a mean age of 27.5 yr (20.7 to 36.9 yr; SD 4.18) were recruited from the Faculty of Rehabilitation Medicine at the University of Alberta. This study was approved by the Health Research Ethics Board at the University of Alberta and informed consent was obtained from all subjects. All subjects had pure tone hearing levels better than 20 dB HL at 500, 1000, 2000, and 4000 Hz. Instrumentation MP3 player and song A Creative MuVo N200 MP3 player (Milpitas, CA, USA) was used for the study. The song No Reason to Cry Out Your Eyes by Hawksley Workman was converted from CD into MP3 format at a bit rate of 192 kbps and then uploaded from the computer to the MuVo. This song was chosen for two reasons. First, it was popular in Canada in 2004 to 2005. Second, the song has a steady driving beat and a small dynamic range throughout and therefore provided listeners with a consistent level for making adjustments to the volume of the MuVo. Figure 1A shows the average amplitude spectrum of the song as well as the levels exceeded 70% and 1% of the time. On average, the difference between the level exceeded 70% of the time and the level exceeded 1% of the time was only 12.4 dB. A song with a larger dynamic range would likely have resulted in the user needing to make many adjustments throughout the song to accommodate the variability in amplitude. It is important to note that the level on Figure 1A would have changed depending on ear listeners PLL. Environments Assessment of preferred listening levels (PLLs) occurred in 3 environments: quiet, street noise and multi-talker babble. The quiet condition served as a baseline comparison for the effects

292

EAR & HEARING / JUNE 2007

Fig. 1. Spectra of the song used in the experiment (A), the street noise (B) and multi-talker babble (C). For each graph the three curves represent average (dotted), 30th percentile (dashed) and 99th percentile (solid) levels.

of noise. The street noise was a 10 sec repeating pattern routed from a compact disc recording through the audiometer that varied in intensity between 70 and 80 dB A-weighted overall SPL. The noise was presented from a soundfield speaker at 1 meter from the subject at zero degrees azimuth. Similarly, the multi-talker babble was presented at an overall level of 70 dB A. The calibration levels for these noises were originally determined in the presence of the third author with a sound level meter at his left ear. With each subject in the experiment, the levels of the noises were monitored by the reference microphone of the probe microphone system (see below for description) to ensure that the calibrated levels were actually being delivered to each subject in Aweighted SPL. Figures 1B and 1C show the average level in 1/12th octave bands as well as the levels exceeded 70% and 1% of the time for both the street noise and the multi-talker babble respectively. It is important to note that the band SPL for a broadband signal is lower than the overall SPL, which is why

the levels in Figure 1B and 1C appear to be lower than the overall level of 70 to 80 dB A reported above. Footsteps, cars and a siren were discernable in the street noise. Users were given as much time as they needed to complete this task. However, once users established the repeating pattern they found it easy to set the volume of the MP3 player in this condition. It was anticipated that these noises would be similar to the kinds of noises experienced commuting on foot (street noise) or in a subway (multitalker babble). Headphones There were 2 styles of headphones used for this study. The earbud style was the default headphone sold with the MuVo. The other style was a supra-aural, over-the-ear headphone with a noisecancelling option (Cyber Acoustics ACM-800, Vancouver, WA, USA). Figure 2 shows the spectra for the over-the-ear headphones with and without noise reduction. These curves were derived by delivering a 70 dB pink noise to a user wearing the headphones with a probe microphone analyzing the spectrum of

EAR & HEARING, VOL. 28 NO. 3

293 microphone was seated in the subjects canal, they were provided the following instructions: The purpose of this test is to find your preferred listening level in different environments with different headphones. There are 3 environments and 3 types of headphones for a total of 9 tests. You will hear the same song for each test. All you have to do is adjust the volume control on the MP3 player up or down until the song sounds best to you. The order of the headphones and the environments was randomized for each subject. After each subject found his/her PLL in each listening condition, they handed the MP3 player to the experimenter. The experimenter went to a set location in the song that the user had heard for each condition and took a probe microphone reading at that location (the word eyes in the chorus of the song). Eyes was 1282 ms in length and had a consistent level for this duration. Thus, the Verifit had time to update the level 3 times within the measurement window. This ensured that the level obtained from the Verifit was the true level for that word for each subject. To minimize the possibility of experimenter bias, the reading was taken by both the experimenter and an assistant on the first 30 trials. The experimenter, listening with monitoring headphones through a y-splitter from the MuVo, would point to the assistant when the appropriate point in the song was reached. The assistant and the experimenter agreed to within 1 dB for all checked trials.

Fig. 2. Real ear spectra for the over-the-ear headphones with and without noise reduction.

the incoming signal near the eardrum in 1/12th octave intervals. The noise-reduction (NR) circuit on these headphones reduces the output for frequencies below 1000 Hz, while providing a boost to frequencies above 1000 Hz. Real ear measurement A probe microphone was inserted into the left ear canal of each user in order to assess the output levels associated with each users PLL. The probe tube from the probe microphone was marked 30 mm from the tip for adult males and 28 mm from the tip for adult females. The probe was inserted until the mark rested in ear listeners inter-tragal notch. Otoscopy was performed to ensure that the probe tube tip was within 5 mm of the tympanic membrane for all subjects (Audioscan, 2005). Once in place, we secured the probe microphone with a piece of tape just below the inter-tragal notch. The probe microphone was connected to an Audioscan Verifit (Dorchester, Ontario, Canada) real-ear analyzer. At the start of each test session the probe microphone was calibrated according to manufacturers specifications. Due to the probe tube resonances, the probe microphone will not have a flat spectrum. Therefore this microphone needs to be calibrated with a broadband signal against the flat response of the real ear reference microphone. The Verifit stores this calibration correction and applies it to all probe microphone measures. The Verifit was set to function as a real ear sound level meter that analyses the input spectrum (in our case in dB A-weighted SPL) every 384 ms (Audioscan, 2005). Procedures Prior to the PLL task, a hearing screening at 20 dB HL was completed for subjects who had not had such completed in the past year. Once the probe

Statistical Design and Analyses A 3 3 within-subjects repeated-measures ANOVA was used to analyze the data. There were two independent variables, headphone style, with 3 levels (earbud, over-the-ear and over-the-ear with NR) and environment, with 3 levels (quiet, street noise and multi-talker babble). The dependent variable was dB A-weighted ear canal SPL. The ANOVA was used to determine the main effects of headphone style and environment. In addition, the ANOVA tested for any interaction between these two independent variables. Based on the research questions listed earlier, 18 planned pairwise comparisons were of interest. To minimize the likelihood of a type-1 error for these multiple contrasts, a Bonferonni correction was applied to the experiment-wise p-value. We divided 0.05 by 18, the number of planned contrasts. This meant that for any one pairwise t-test of means to be significantly different, the p-value associated with that contrast had to be lower than 0.002778. All statistical analyses were performed using SPSS (Version 14.0, 2005).

294
TABLE 1. Mean A-weighted ear canal sound pressure levels for each of the 9 conditions. Standard deviations are in parentheses Quiet In-the-ear Over-the-ear Over-the-ear/NR 77.8 (7.7) 75.2 (9.5) 75.2 (9.1) Street noise 88.8 (3.9) 84.5 (4.8) 83.0 (5.4) Multi-talker babble 86.7 (4.3) 82.9 (5.5) 81.6 (5.6)

EAR & HEARING / JUNE 2007 The ANOVA also revealed a significant 2-way interaction between headphone style and environment (F(3.036, 112.341) 6.830, p .001; eta squared .10). Mean and standard deviation values for each of the 9 conditions are presented in Table 1. Figure 3 shows the means plus the 95% confidence intervals for each condition. Table 2 lists the 18 pairwise contrasts and their associated significance level. Fifteen of the eighteen contrasts were significant. The only comparisons that failed to reach significance were the earbuds versus over-the-ear in quiet, over-the-ear versus overthe ear with NR in quiet, and the over-the-ear versus over-the-ear with NR in street noise. The over-the-ear versus over-the ear with NR in quiet contrast was not significant. Recall that Figure 2 showed the spectral differences with and without noise reduction. We tested the difference in dB A for a fixed volume setting in the real ear of several subjects. There was a consistent 6.5 dB A increase in ear canal SPL with the NR activated. However, the difference virtually disappears in the quiet condition. Presumably, subjects noticed the increased output and adjusted the level to the same PLL in quiet for both over-the-ear headphone conditions. This suggests that the changes in frequency spectra did not have a significant effect on the PLL of users. Instead, differences between the over-the-ear with and without noise reduction (e.g., in multi-talker babble) are more likely a result of the NR circuit rather than simply a difference in frequency spectra. The mean PLLs obtained in each of the 9 conditions were used to calculate the permissible listen-

RESULTS
The repeated-measures ANOVA revealed that the main effects of SPL in relation to headphone style and environment were significant, though the assumption of sphericity was not met. To assess the sensitivity of these models to the assumption of sphericity, a mixed model approach with unstructured covariance was also fit. These models yielded similar results, so the RM ANOVA is reported here. A Greenhouse-Geisser epsilon correction was applied to both headphone style (F(1.385, 51.244) 42.224, p 0.001; eta squared 0.12) and environment (F(1.199, 44.378) 114.553, p 0.001; eta squared 0.54). When collapsed across all environments, use of the in-the-ear headphones resulted in significantly higher PLLs (M 84.4, 95% CI 82.8 to 86.0) compared to the over-the-ear (M 80.9, 95% CI 78.8 to 82.9) and the over-the-ear with NR (M 79.9, 95% CI 77.8 to 82.1). When collapsed across all headphone styles, the street noise (M 85.4, 95% CI 84.0 to 86.8) resulted in significantly higher PLLs than either the multitalker babble (M 83.7, 95% CI 82.1 to 85.3) or the quiet condition (M 76.0, 95% CI 73.2 to 78.8).

Fig. 3. Interaction for headphone style showing the mean ear canal level for each headphone environment condition plus the 95% confidence intervals for each.

EAR & HEARING, VOL. 28 NO. 3

295

TABLE 2. Mean differences (in dB), 95% confidence intervals and 2-tailed significance for the planned pairwise comparisons on the significant headphone environment interaction. Bolded pairs failed to reach statistical significance 95% CI Pair 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 ITE quiet ITE street noise ITE quiet ITE babble ITE street noise ITE babble OTE quiet OTE street noise OTE quiet OTE babble OTE street noise OTE babble OTE NR quiet OTE NR street noise OTE NR quiet OTE NR babble OTE NR street noise OTE NR babble ITE quiet OTE quiet ITE quiet OTE NR quiet OTE quiet OTE NR quiet ITE street noise OTE street noise ITE street noise OTE NR street noise OTE street noise OTE NR street noise ITE babble OTE babble ITE babble OTE NR babble OTE babble OTE NR babble Mean 11.01 8.84 2.17 9.29 7.76 1.53 7.84 6.47 1.37 2.64 2.66 0.01 4.37 5.83 1.46 3.72 5.03 1.30 Lower bound 12.95 10.57 1.29 11.19 9.57 0.64 9.44 8.11 0.56 0.86 1.10 1.00 3.04 4.43 0.53 2.48 3.73 0.68 Upper bound 9.08 7.11 3.06 7.38 5.96 2.41 6.25 4.84 2.18 4.42 4.22 1.03 5.70 7.22 2.39 4.97 6.32 1.93 Sig.(2-tailed) 0.001 0.001 0.001 0.001 0.001 0.001 0.001 0.001 0.001 0.005 0.001 0.979 0.001 0.001 0.003 0.001 0.001 0.001

ITE, in-the-ear (earbuds); OTE, over-the-ear; OTE NR, over-the-ear with noise reduction.

ing duration to reach 100% daily noise dose according to the following equation (NIOSH; OSHA; American Academy of Audiology, 2003): T (min) 480/2(L-85)/3 (1) Where T number of minutes that can be spent at that level of exposure (L) before reaching 100% daily noise dose. Table 3 lists the means, 95% confidence intervals and number of hours to reach maximum noise dose. Durations to reach maximum noise dose ranged from 3.3 hr for the earbuds in street noise to as much as 77 hr for the over-the-ear headphones in quiet. We also tested the maximum sound pressure levels achievable in the ears of authors 1 and 3 of this paper. The earbuds reached 110 dB A, the
TABLE 3. Mean ear canal level in dB A and 95% confidence intervals. Maximum permissible noise dose in hours for each headphone and noise condition also presented 95% CI Condition ITE quiet ITE street noise ITE babble OTE quiet ITE street noise OTE babble OTE NR quiet OTE NR street noise OTE NR babble Mean Lower bound Upper bound Hours 77.82 88.83 86.66 75.17 84.46 82.93 75.16 83.00 81.63 75.29 87.54 85.23 72.04 82.87 81.12 72.16 81.21 79.79 80.33 90.12 88.08 78.31 86.05 84.75 78.16 84.79 83.47 42.07 3.30 5.45 77.51 9.06 12.89 77.74 12.70 17.42

over-the-ear with NR reached 106.5 dB A, while the over-the-ear without noise reduction reached 100 dB A. By substituting these values into equation 1, with this MP3 player and headphone combinations the maximum daily noise dose can theoretically be reached in a matter of a few minutes (see Table 4).

DISCUSSION
The ever-increasing popularity of MP3 players has lead to a media blitz over the potential hearing hazards of these devices. At the time of writing, Apple inc. was facing a class action lawsuit from a Louisiana man over the potential hazard caused by the maximum output capabilities of the iPod. Perhaps in response to this, Apple released software that could be uploaded to the iPod to limit the maximum output capabilities of the device. We tested this in our lab using a 2-cc coupler and the same song used for this study. We found that by adjusting the volume limiter to the maximum setting the output of our iPod Nano at full settings could be reduced by approximately 50 dB.
TABLE 4. Maximum ear canal level in dB A and the maximum permissible noise dose in minutes for each style of headphone Headphone style ITE OTE OTE NR Mean 110.5 100.0 106.6 Minutes 1.33 15.00 3.34

ITE, in-the-ear (earbuds); OTE, over-the-ear; OTE NR, over-the-ear with noise reduction.

ITE, in-the-ear (earbuds); OTE, over-the-ear; OTE NR, over-the-ear with noise reduction.

296 It is important to emphasize that these maximum levels, while theoretically harmful (Fligor & Cox, 2005), are only relevant to users who choose to listen at those levels (see Table 4). For the majority of the public, what matters to risk assessment is the actual usethe preferred listening levels and the length of time users choose to listen at those levels. The survey data from Zogby (2006) regarding MP3 player use levels and duration may seem troubling, especially given the percentages of users who listen at loud levels and for extended periods of time. However, the loudness data are subjective and what seems loud to one user may be considered medium or extremely loud to another. The actual objective ear canal level in dB A for a preferred listening level is needed to help audiologists and consumers understand whether or not these survey data are indeed troubling. Based on recent literature, it could be expected that preferred listening levels would be affected by the type of headphone that a listener uses and the environment in which they are listening. For example, a recent study by Williams (2005) showed that noisy environments were likely contributors to the overall listening levels in his study. However, use levels were not assessed in quiet, making it difficult to quantify the actual effects of noise on listening levels. Further, the device and headphone combinations used by the subjects were not specified. Studies reporting user listening levels have relied on a KEMAR mannequin to estimate the actual users ear canal levels (Rice et al., 1997; Williams, 2005). While KEMAR has ear canal properties that approximate the average adult ear canal, it is well-established that no two ears are alike. Saunders and Morgan (2004) have shown that individual ear canal acoustics can result in as much as a 40 dB difference between ears in actual measured SPL for the same input stimulus. For this reason, we felt the use of a real-ear probe microphone was most relevant for the PLL measurements in this study, since it would capture the A-weighted SPL with each users real ear acoustics contained in the measure. The results of the present study revealed that the majority of the variance was accounted for in the main effects of headphone style and environment. For headphone style, the earbuds resulted in the highest PLL, while the over-the-ear style resulted in a significantly lower PLL. The difference between over-the-ear with and without noise reduction was significant but the effect was small. The listening environment also accounted for much of the variance in the data. Perhaps due to its spectral, temporal and level variability, the street noise was more

EAR & HEARING / JUNE 2007 disruptive to the user than the multi-talker babble. It is also possible that the active noise reduction circuit in the over-the-ear headphones was better able to filter out the low frequencies of the multitalker babble. On the whole, both noise conditions led to higher PLLs than the quiet condition. MP3 users who have finished using their device in a noisy environment (e.g., the gym) and then turned it on in a quiet environment (e.g., at home) will have no difficulty absorbing this point. The interaction between headphone style and environment, while significant, accounted for a smaller proportion of the variance associated with the ANOVA. Many of the comparisons were statistically significant; however, the effects were small. This is visually apparent by the overlap in confidence intervals in Figure 3. We recommend cautious interpretation of the interaction effects in terms of their real-world significance as some comparisons of less than 2 dB reached statistical significance. Although it was not the main focus of this investigation, we did explore permissible listening duration to reach 100% daily noise dose. There are 2 limitations that need to be addressed with the approach taken in this study. First, the formula used assumes occupational noise was inserted into the calculation. Occupational noise often has much more predictable spectral and temporal properties than music. In this study, we chose a song with very few changes in overall level that would make it easy for the listener to set the MP3 volume and also make it easy for the experimenters to measure the ear canal level. Consequently, it could be argued that the generalizability of these results may extend only to pop music with a consistent level. However, recording engineers routinely apply compression in the final mastering of music so that overall levels within and between songs are not tremendously variable. If this were not the case, people listening to the radio or their MP3 player would be forced to constantly adjust the volume control even in a quiet environment. The second issue is the validity of the ear canal A-weighted SPL used in the calculations. Damagerisk criteria are based on sound levels in the free field (outside the ear), not in the near field (ear canal). The ear canal is known to provide a boost to the incoming signal especially in the mid to high frequencies ((Shaw, 1974a). It is likely that the A-weighted levels we measured would have a slight positive bias and would need to be adjusted down a few dB before being used in the calculation. However, with the exception of the earbuds in noise, the uncorrected ear canal PLLs were at levels that

EAR & HEARING, VOL. 28 NO. 3 would be considered relatively safe, in occupational terms (i.e., 9 hr permissible duration). Overall, the PLLs chosen by the subjects in this study were quite conservative. The study participants were adults, only a small percentage of whom owned or used an MP3 player on a regular basis. Additionally, with all of the media attention directed at the hazards of MP3 players, it is difficult to imagine the subjects not determining what the experiment was attempting to address. Each subject was carefully instructed to adjust the music to the point where it sounded best and to imagine they were in the real world and not an audiology experiment. However, it is conceivable that the overall PLLs may have been conservatively set due to the subjects motivation to not alarm the audiologist in the room. Finally, there were more women than men in this study. Williams (2005) showed that, in a field trial, men were listening at a significantly higher level than women. A secondary 2 3 3 mixed ANOVA on our data with gender as a betweensubjects variable was not significant. However, there did appear to be a trend emerging for men to listen at a higher level than women (though mostly in the quiet conditions). In light of the survey data reported by Zogby (2006) and the fact that the subjects in this investigation are not necessarily representative of the entire listening public, more research is needed to determine if the same patterns of use found in this study apply to the following groups: People who routinely use MP3 players MP3 users of different ages (high school students versus adults) Male versus female MP3 users

297 members in the faculty of rehabilitation medicine) suggest that MP3 listening levels may not be as significant a concern for professional adults, as has been noted recently in the mainstream media. This may be especially true if users opt for overthe-ear style of headphones.
Address for correspondence: William E. Hodgetts, Department of Speech Pathology and Audiology, University of Alberta, Edmonton, Alberta, Canada, T6G 2G4. E-mail: bill.hodgetts@ualberta.ca Received April 10, 2006; accepted November 29, 2006.

REFERENCES
American Academy of Audiology. Position Statement: Preventing noise-induced hearing occupational hearing loss. October, 2003. Available at: http://www.audiology.org/professional/positions/ niohlprevention.pdf. Accessed February, 15, 2006. Audioscan. (2005). Verifit VF-1 real-ear hearing aid analyzer users guide: Version 2.4.1. Dorchester, Ontario. Fligor, B. J., Cox, L. C. (2004). Output levels of commercially available portable compact disc players and the potential risk to hearing. Ear & Hearing, 25, 513527. National Institute for Occupational Safety and Health (NIOSH). (1998). Criteria for a recommended standard: Occupational noise exposure, revised criteria. Pub. No. 98-126. Cincinatti: NIOSH. NPD Group Canada. MP3 Players Top the consumer Electronic Market. August 19, 2005. Available at: http://www.npd.com/ dynamic/releases/press_050818.html. Accessed, November 16, 2005. Occupational Safety and Health Administration. (1981). Occupational noise exposure: Hearing conservation amendment. Federal Register, 46, 40784179. Occupational Safety and Health Administration. (1983). Occupational noise exposure: Hearing conservation amendment; Final rule. Federal Register, 48, 97389785. Rice, C. G., Breslin, M., Roper, R. G. (1987a). Sound levels from personal cassette players. British Journal of Audiology, 21, 273278. Rice, C. G., Rossi, G., Olina, M. (1987b). Damage risk from personal cassette players. British Journal of Audiology, 21, 279288. Saunders, G. H., Morgan, D. E. (2003). Impact on hearing aid targets of measuring thresholds in dB HL versus dB SPL. International Journal of Audiology, 42, 319326. Shaw, E. A. G. (1974a). The External Ear. In: Keidel WD, Neff WD (eds). Handbook of Sensory Physiology (Vol. 5, pp, 455 590). New York: Springer-Verlag Publisher. Williams, W. (2005). Noise exposure levels from personal stereo use. International Journal of Audiology, 44, 231236. Zogby International. Survey of Teens and Adults about the Use of Personal Electronic Devices and Head Phones. March 14, 2006. Available at: http://www.asha.org/NR/rdonlyres/10B67FA1002C-4C7B-BA0B-1C0A3AF98A63/0/zogby_survey2006.pdf. Accessed March 20, 2006.

CONCLUSIONS
Headphone style and environment have significant effects on the PLLs of normal hearing adults. Use of the most commonly sold headphone style (earbud) was associated with the highest ear canal levels. The over-the-ear headphones appear to block out more noise, which results in lower ear canal levels for the same PLL. Also, subjects choose higher ear canal levels as the amount of background noise increases for all headphone styles. Overall, levels chosen by our subjects (student