Abstract
Noise, or unwanted sound, exposure has been shown to have a wide range of negative physical and psychological effects. Although situational context, sound characteristics, and individual expectation affect the experience of noise and its related outcomes, the personality trait of noise sensitivity also plays a critical role in assessing noise impacts. As the most widely used 21-item Noise Sensitivity Scale measure of sensitivity is often too long to administer in time-sensitive field settings, the authors conducted five studies to create and validate a shortened, field friendly version of the original, longer measure of noise sensitivity. The resulting five-item measure of noise sensitivity was shown to be internally consistent, temporally stable, highly correlated with the original measure, and predictive of noise-related outcomes such as attitudes toward specific noise, acceptability ratings of noise events, and motivations for visiting quiet locations. The applied value of the scale and implications for facilitating future research are discussed.
Unwanted sound, or noise, is considered one of the most ubiquitous of ambient environmental stressors. Noise generated by transportation, workplace technology, and other humans has been shown to be detrimental in a number of ways. For example, the presence of unwanted sounds has been related to higher stress levels (e.g., Ising & Kruppa, 2004), increased cognitive errors (e.g., Benfield, Bell, Troup, & Soderstrom, 2010b; Stansfeld et al., 2005), reduced health and well-being (e.g., Stansfeld & Matheson, 2003), disturbances in sleep cycles (e.g., Muzet, 2007), negative changes in affect (e.g., Benfield, Bell, Troup, & Soderstrom, 2010a; Mace, Bell, & Loomis, 1999), and lower amounts of prosocial behaviors (e.g., Moser, 1988). Increasingly, scientists are recognizing the physical health consequences of noise (e.g., Babisch, 2006; Fritschi, Brown, Kim, Schwela, & Kephalopoulos, 2011) and consequently adding “harmful” to its definition (e.g., “unwanted or harmful sound”). In doing so, they embrace beliefs that noise can be measured not just as a subjective annoyance but as an objective harm quantifiable by its health consequences across populations or by a “reasonable person” standard of assessment. In short, exposure to noise is a problem physiologically and psychosocially, so understanding the components of noise and psychological perceptions of noise as they relate to these outcomes is important.
Different theoretical perspectives have been used to explain the process underlying these negative effects, but most rely heavily on the interaction between the physical sound stimuli and the perceiver of the sound. For instance, arousal and stress perspectives posit that noxious sound increases negative affect, generalized anxiety, and frustration in the listener, which in turn causes stress and/or arousal with subsequent changes in blood pressure, sleep patterns, immune functioning, and hormone levels (Cohen, Evans, Stokols, & Krantz, 1986). In these models, the individual’s reaction to the sound (e.g., annoyed, frustrated) drives the negative consequence rather than the sound itself. Similarly, sounds that are perceived as unnecessary or detrimental to something the listener values are more annoying and psychologically damaging. The same is true of noises that are considered by the perceiver to be hazardous to his or her health (Green & Fidell, 1991). Noise generated by someone seemingly unconcerned with the listener’s well-being is also evaluated negatively (Miedema & Vos, 1999). In short, people are generally motivated to avoid noise, but individual differences in listeners and their perceptions of the stimuli contribute heavily to noise-induced outcomes and experiential reports of noisy environments.
Examples of these individual differences in sound and noise perception are numerous. For instance, individuals trained in music or singing respond differently when compared with novices (Williams, 2005). Similarly, military combat veterans can be hypersensitive to the holiday sights and sounds of New Year’s eve (e.g., flashing lights, loud explosions) because similar stimuli have become associated with mortal danger (Chemtob, Roitblat, Hamada, Carlson, & Twentyman, 1988). More generally, the rumbling of a thunderstorm can be an exciting and pleasant experience to some but terrifying or depressing to another. Likewise, a parent trying to lull a newborn to sleep or a night shift worker trying to rest during the day perceives bird chirps, garbage trucks, and telephone rings differently from those who are currently less motivated for quiet conditions.
Given the definition of noise (i.e., unwanted or harmful sound), an emphasis on the perceiver and his or her individual perceptions or traits makes sense if researchers want to truly differentiate between effects caused by the sound, effects caused by the listener, and effects that depend on the interaction between the two. Consequently, individual differences are often tested and controlled for in sound-related research to partial out which effects are universal and which are more person dependent.
Individual Differences in Noise Sensitivity
For example, the Big Five personality characteristics of extraversion and neuroticism have been shown to affect participant perceptions of and performance during exposure to noise. Extraversion—the tendency to be gregarious, be assertive, and seek out social stimulation—has been shown to be related to preference for greater sound intensity (Geen, 1984) and reduced distraction by extraneous noise (Belojevic, Slepcevic, & Jakovljevic, 2001; Daoussis & McKelvie, 1986; Furnham, Gunter, & Peterson, 1994). Likewise, neuroticism—the tendency to be anxious, emotionally negative, and less emotionally stable—has also been shown to be positively correlated with annoyance during noise exposure (Öhrström, Bjorkman, & Rylander, 1988) and with noise-induced sleep disturbances (Jakovljevic, Belojevic, Paunovic, & Stojanov, 2006). Neurotics when compared with more emotionally stable individuals also perform worse on semantic retrieval tasks (von Wright & Vauras, 1980) and prose memorization tasks (Nurmi & von Wright, 1983) under noisy conditions.
Aside from the previously mentioned personality variables, one individual difference has been particularly salient in research on noise-related outcomes. Noise sensitivity refers to an individual aversion toward and/or reactivity to noise and noisy environments (Job, 1999; Weinstein, 1978). Research on individual noise sensitivity has shown that it is conceptually distinct from extraversion or neuroticism, uncorrelated with noise exposure, and largely independent of auditory acuity or function (Ellermeier, Eigenstetter, & Zimmer, 2001; Job, 1988; Miedema & Vos, 2003; Stansfeld, Clark, Turpion, Jenkins, & Tarnopolsky, 1985). Higher noise sensitivity is related to higher skin conductance responses in the presence of noise (Stansfeld & Shine, 1993), greater reported annoyance (Belojevic & Jakovljevic, 2001), elevated heart rate (Abel, 1990), sleep loss (Jakovljevic et al., 2006), psychiatric disturbance (Stansfeld, Clark, Jenkins, & Tarnopolsky, 1985), and slower habituation in the presence of noise (Stansfeld & Shine, 1993). As made obvious by those findings, the inclusion of some form of measurement for noise sensitivity is paramount to most noise- or sound-related studies. Although this is accomplished using several different strategies ranging from single-item indicators to standardized scales to noise complaint frequency, some measures have been favored more than others.
The Weinstein (1978) Noise Sensitivity Scale (NSS), developed with a college student sample, is considered the most well-researched and widely used formal measure of noise sensitivity (Job, 1999; Miedema & Vos, 2003; Zimmer & Ellermeier, 1999). It has been shown to be psychometrically valid (Ekehammar & Dornic, 1990; Zimmer & Ellermeier, 1999) as well as temporally stable (Stansfeld, 1988; Zimmer & Ellermeier, 1999), and the majority of the research cited previously used this 21-item instrument.
The Current Studies
As stated previously, noise is an ever-present environmental stressor linked to a host of adverse effects that manifest themselves to varying degrees depending on individual differences in the perceiver of the sound. As a result, characteristics of the listener need to be included in sound research to allow for clear distinctions to be made between effects caused by the stimuli and effects facilitated by the individual. Because noise sensitivity as measured by Weinstein’s (1978) NSS has been shown to be a critical individual difference factor in the perception of noise and the evaluation of noise-related outcomes, it should probably be measured in most human-based noise research projects. Exploring noise effects in applied research contexts such as museums and national parks is becoming more common (e.g., Jakubowski, 2011; Pilcher, Newman, & Manning, 2009). As the compliance rate for agreeing to intercept interviews with participants in these applied settings declines with the length of questionnaires employed (Loomis, 1987), the 21-item version of the NSS can make it less likely that potential participants will complete any questionnaire. Accordingly, we report herein the development of a short, field-useful version of the NSS that is psychometrically reliable, valid, and representative of the original. This was accomplished across five studies.
Study 1
Using exploratory factor analysis on three separate samples (college students, adults), the first study was designed to isolate five to seven items that best represented the construct of noise sensitivity. Those items would serve as the NSS short form (NSS-SF) used for later studies on reliability, validity, and confirmatory factor structure.
Method
Participants
Three separate samples of participants were recruited through either laboratory research projects (two student samples; n = 390 and 308) or an online survey community (adult sample; n = 202). Demographics for each sample are given in Table 1.
Sample Demographics Across the Five Studies.
Materials and Measures
Participants were given the Weinstein (1978) NSS that consists of 21 items measured along a 6-point scale ranging from “disagree strongly” to “agree strongly.” This measure has been shown to have strong internal consistency (α = .86) and to be psychometrically robust (Ekehammar & Dornic, 1990; Weinstein, 1978; Zimmer & Ellermeier, 1999).
In addition, participants in Sample 3 (adults) completed the Big Five Inventory–54 item (BFI-54; John & Srivastava, 1999) during the survey. The BFI-54 requires participants to rate agreement with personally descriptive items using a 5-point scale and provides subscale scores for extraversion (9 items; α = .84), agreeableness (9 items; α = .82), conscientiousness (9 items; α = .85), emotional stability (9 items; α = .89), and openness (18 items; α = .88). This measure was included based on previous literature showing a relationship between noise sensitivity, and the extraversion and neuroticism factors of the Big Five (Dornic & Ekehammar, 1990).
Procedure
Participants completed the questionnaire(s) as either part of a larger laboratory research study involving noise (student samples) or as part of an online survey created exclusively for this project (adult sample). All three samples also completed a brief demographic questionnaire.
Results and Discussion
Three separate exploratory factor analyses were run on the different samples to determine which items on the NSS consistently displayed the highest factor loadings. Such consistency was interpreted to be an indication that those items were regularly demonstrating a strong connection with the larger construct being measured across multiple samples. Results showed that 5 items consistently had the highest factor loadings (.467-.788) across all three samples. Internal consistency among those items was shown to be acceptable, especially given the small number of items (α = .73, .78, and .82, respectively). Moreover, the 5-item scales were highly correlated (r = .86-.91) with the 21-item original NSS. As a result, those 5 items were chosen to act as the initial base for the NSS-SF. Table 2 displays the 5 items retained for the short form along with factor loadings, reliability information, and correlations with the NSS across samples.
Item Factor Loadings, Internal Consistency, and Correlation With Original NSS for Study 1, 2, 4, and 5 Samples.
Note: NSS = Noise Sensitivity Scale.
NSS Item 8 is reverse scored.
*p < .000.
After determining the initial items for the NSS-SF, correlations were run between the NSS-SF score and the Big Five personality traits measured in the adult sample. As described elsewhere, previous research has shown that extraversion and neuroticism (the opposite end of emotional stability) had contrasting relationships to noise annoyance perception and outcomes. Extraversion was expected to negatively correlate with NSS-SF scores; emotional stability (the opposite of neuroticism) was also expected to negatively correlate with NSS-SF. As seen in Table 3, both hypothesized relationships were demonstrated, with extraversion and neuroticism significantly correlating with NSS-SF scores in the hypothesized direction (r = −.15 and −.29, respectively).
Correlations Between NSS-SF and Noise-Related Outcomes (Across Studies).
Note: NSS-SF = Noise Sensitivity Scale short form.
p < .08. **p < .05. ***p < .01.
With five items selected to represent the NSS-SF and internal item consistency being acceptable, a solid foundation to build on was established in Study 1. In addition, initial tests of convergent validity showed relationships that were consistent in terms of direction and magnitude to previous research. However, several additional pieces of information related to structure, overall stability, and validity were necessary to demonstrate that the scale was acceptable for use as a research instrument.
Study 2
The original NSS represented a single unified factor, and the five items used in the short form should perform similarly. Before attempting to further validate the NSS-SF, a confirmation of its factor structure was carried out using a new sample and confirmatory factor analysis procedures.
Method
Participants
Some 438 participants were recruited through a university research pool to complete the NSS-SF. Sample demographics are shown in Table 1.
Materials, Measures, and Procedure
Participants were recruited for a Mass Testing Survey to partially fulfill a course research requirement. Participants completed the five-item NSS-SF along with a brief demographic questionnaire and several other measures unrelated to this report.
Results and Discussion
Confirmatory factor analysis was conducted on the five-item NSS-SF to ensure that it conformed to the unified structure demonstrated in the original NSS within acceptable boundaries related to model fit. That was indeed the case with fit indices being acceptable for the anticipated single-factor structure (comparative fit index [CFI] = .97; root mean square error approximation [RMSEA] = .078). Factor loadings and internal consistency (α) were similar to Study 1 findings and are reported in Table 2.
With a consistent factor structure and replicated evidence of internal consistency, the five-item NSS-SF appeared to be operating in much the same way as the original NSS. However, additional tests of validation were necessary to ensure that the measure performed similarly to its longer counterpart.
Study 3
The first two studies showed that the NSS-SF has the same factor structure of the original scale and is internally consistent. One of the remaining questions regarding temporal consistency was tested in Study 3 using a student sample and a 5-week (35-day) delay between scale administrations.
Method
Participants
Undergraduate students (n = 49) completed the NSS-SF for course research credit. Sample demographics are listed in Table 1.
Measurement and Procedure
Participants completed the NSS-SF on two separate occasions spaced 5 weeks (35 days) apart. Participants were debriefed following each measurement occasion.
Results and Discussion
A correlation between the two separate NSS-SF measurement occasions as a test of temporal stability (i.e., test–retest reliability) was conducted. Results of that analysis showed that individual NSS-SF scores were internally consistent (α = .79 and .81 at Times 1 and 2, respectively), temporally stable (r = .83), and in line with the previously reported stability for the original 21-item NSS across a similar time period (r = .87 across 4 weeks; Zimmer & Ellermeier, 1999).
Given these findings, in conjunction with the first two studies, it was concluded that the NSS-SF is psychometrically similar to the longer, well-established NSS. The NSS-SF had an identical factor structure as well as acceptable internal and temporal consistency; it was also highly correlated with the original scale and converged appropriately with other personality constructs (i.e., extraversion and neuroticism). However, given the larger purpose of the research project (i.e., to develop a short version of the NSS to facilitate the inclusion of individual differences into noise research), additional tests of the NSS-SF in field settings were deemed necessary.
Study 4
Visitors to national parks are often surveyed by researchers for a wide range of purposes including satisfaction with or acceptability of conditions. Noise levels in parks are a source of conflict (e.g., Bell, Mace, & Benfield, 2009-2010) and are under ongoing study as a result of legislative mandates (e.g., National Park Air Tour Management Act of 2000). Consequently, the NSS-SF should be of potential usefulness to such research. Study 4 used the NSS-SF in a noise-related study taking place in Rocky Mountain National Park located near Estes Park, Colorado.
Method
Participants
Sample demographics are reported in Table 1.
Materials and Measures
Participants completed the NSS-SF, a brief demographic questionnaire, and a park visitor experience questionnaire. This questionnaire contained several items related directly to noise, including one set of item asking participants to rate along a 5-point range the degree to which different sound sources (i.e., auto traffic, park operations/construction, visitors talking, aircraft overflights, and personal electronic devices) were problematic to the visit. Participants also rated the acceptability of hearing aircraft overflights during their visit based on varying amounts of temporal frequency (i.e., every 5, 15, 30, or 60 min).
Procedure
Participants were recruited from the Bear Lake trailhead in Colorado’s Rocky Mountain National Park by selecting every nth person (dependent on daily visitor volume). These participants were asked to complete the survey packet containing the demographics, NSS-SF, and park visitor questionnaire. After completing the study, participants were debriefed regarding the purposes of the study. All surveying took place between August 9 and August 13, 2010; 67% of potential participants agreed to take part in the research.
Results and Discussion
Correlations were run between the NSS-SF score, and the ratings for the problematic noises and the aircraft noise frequency acceptability. Those relationships are given in Table 3. Generally, higher NSS-SF scores related to lower ratings of frequent aircraft overflight acceptability and higher ratings of most sounds as being problematic to visitation.
However, these findings are not necessarily based on exposure to actual noise. Visitors to the national park reported certain sounds as being less acceptable or more problematic, but this could reflect attitudinal preferences more than the individual response to the actual problematic presence of those sounds. Nonetheless, the theoretically consistent relationship between noise sensitivity and individual attitudes toward different sounds in context does provide evidence that the NSS-SF performs in a manner consistent with theory and its longer counterpart.
Study 5
A fifth and final study was run to test the relationship NSS-SF score has with responses to actual sound exposure in a different sample of national park visitors. It was hoped that surveying park visitor responses to a controlled set of auditory stimuli while including the NSS-SF would provide sufficient evidence when combined with the other studies to support the scale’s use in future noise research.
Method
Participants
Sample demographics are reported in Table 1.
Materials and Measures
Sound clips
A set of five audio clips was created using natural and aircraft sounds recorded within the national parks system. Each sound clip was equivalent in length (48 s) and depicted a natural environment during an air tour overflight. The clips differed in the level of noise energy created by the overflight, with each clip representing a gain of 7 dBA based on Leq (ranging from 36 to 64 dBA). Leq is a measure of sound pressure level that is averaged over a time period to provide a summary of the general loudness of a particular location. In the context of this research, an Leq of 36 dB represents an environment that is on average quieter than an office when no one is talking. An Leq of 64 dB represents an environment that is similar to a room where many people are talking at once or a lecturer is projecting his or her voice in a large lecture hall.
Acceptability scale and visitor survey
Participants were asked to rate each audio clip based on the acceptability of the event occurring during a visit to the park. These acceptability ratings were measured using a single item ranging from −4 = very unacceptable to +4 = very acceptable that has been used extensively within the acceptability literature (see Vaske & Shelby, 2008, for a review). Participants also completed a general park visitation survey that contained the NSS-SF and demographic questions as well as a set of items concerning the importance of different visitation motives, such as “enjoying time with family,” “experiencing peace and quiet,” or “learning about Native American culture and history.” Those items were rated using a 0 to 4 scale (not at all important to extremely important).
Procedure
Participants were recruited from the Tuolumne Meadows region of California’s Yosemite National Park between July 10, 2010 and August 6, 2010. Those agreeing to participate were given the visitor survey containing the NSS-SF, demographics, and the importance of visitation motives questions. They were then given an MP3 player to listen to the five audio clips and the acceptability rating sheet. Three separate orders of the audio clips were used to prevent order effects; participants were randomly assigned to the different orders. After completing the entire visitor packet and the audio recording rating task, participants were debriefed and thanked for their time.
Results and Discussion
Correlations were run between the NSS-SF and the responses to the visitor motives questions. As shown in Table 3, NSS-SF score positively correlated with the importance of sound-related motives, such as getting away from the crowds (r = .26, p = .001), experiencing solitude (r = .22, p = .005), and hearing the sounds of nature (r = .19, p = .019). NSS-SF score was not correlated with other motives such as getting exercise, viewing the night sky without light pollution, or learning about Native American culture/history.
To test the relationship between NSS-SF score and sound clip acceptability ratings, five multiple regressions were run with demographic variables (age, sex) and NSS-SF score used as predictors of each clip’s acceptability rating. As shown in Table 4, NSS-SF score significantly related to decreased acceptability ratings of sound clips, but only at the higher volume conditions (i.e., 57 and 64 dBA). The general trend of the data suggests that as clip volume increased, the importance of NSS-SF also increased. As the final two audio clips were also the only clips that had acceptability ratings in the “unacceptable” range (i.e., −4 to −1), the results suggest that the predictive value of the NSS-SF is particularly meaningful when unwanted sounds (the definition of noise) are in the unacceptable range, but not when the sounds are acceptable (i.e., are not considered noise). Such findings support the construct validity of the NSS-SF.
Regression Results for NSS-SF Predicting Audio Clip Acceptability (Study 5; n = 165).
Note: NSS-SF = Noise Sensitivity Scale short form.
p < .05. **p < .01.
When taken together, the results of this study provide additional evidence of the NSS-SF’s validity and usability in research contexts. Results showed that the scale relates to more positive visitor attitudes toward sound-specific visitation motives but not sound-unrelated motives. In addition, this study showed that the NSS-SF is predictive for noise conditions (i.e., for unwanted/unacceptable sounds), but not for acceptable sounds (i.e., nonnoise) in a field setting, which is what one would want for a noise sensitivity measure.
General Discussion
Across five studies and seven different samples that included students, adults, and national park visitors, this project created a five-item version of one of the most widely used measures of noise sensitivity—Weinstein’s NSS. This field friendly version of the NSS is structurally identical to the original with comparable levels of internal consistency and temporal stability. It is highly correlated with the original NSS as well as being associated with constructs such as extraversion and neuroticism. It also relates to a host of noise-related attitudes and outcomes such as more negative attitudes toward human-caused noise as well as more positive attitudes toward sound-related motives for visiting national parks. In short, the NSS-SF represents the original measure remarkably well and improves on its research value by being more practical in item-limited, real-world applications.
In a research context, a shortened version with such fidelity with the original scale has several advantages. Most obvious are the possibilities opened to applied researchers. Community noise surveys, visitor intercept interviews, and fieldwork on sleep disturbances would benefit from the addition of an individual difference measure of noise sensitivity but cannot always include the additional items necessary for such a measure. Oftentimes, such situations necessitate the use of a single-item noise sensitivity question, but such measures have been shown to lack reliability (e.g., Zimmer & Ellermeier, 1999). The NSS-SF confers all the advantages of the longer, original NSS without adding the same burden of items; it could serve as a compromise between questionnaire brevity and the need for good measurement. Taken a step further, it would not be unreasonable for the NSS-SF to be used in laboratory situations in place of the longer original. After all, the two scales are structurally identical, highly correlated, and have comparable reliability. Why measure something with 21 items if you can accomplish the same goals with only 5?
This is not to say that the NSS-SF should be heralded as the definitive measure of noise sensitivity. The original NSS is highly valued within noise research and is often the exemplar in discussion related to noise sensitivity (e.g., Job, 1999; Miedema & Vos, 2003; Zimmer & Ellermeier, 1999). As such, it was an obvious choice for the current research purpose. However, other measures such as Zimmer and Ellermeier’s (1998, 1999) Noise Sensitivity Questionnaire (LEF) are equally valuable in noise research and may even be more appropriate in some contexts. Unfortunately, the LEF contains 52 items making it impractical for use in brief surveys. Balancing the importance of questionnaire length, research specific goals, and different measurement options is something that individual researchers will have to decide on themselves, but future research involving the NSS-SF and other measures of noise sensitivity can help with those decisions. Perhaps shortened versions of other scales could be developed to allow for use in field situations to complement the versatility and advantages given by a 5-item NSS-SF.
In the end, a new option for noise researchers has been created that allows for survey brevity and measurement validity. It is hoped that such an instrument allows for new research avenues to be opened in field settings or while addressing applied problems; it is also hoped that such an instrument facilitates the inclusion of individual noise sensitivity into noise research generally. More broadly, the NSS-SF might be useful in studies of and interventions for the serious health consequences of noise, which include among other things hypertension, ischemic heart disease, cognitive impairment in children, tinnitus, and sleep disturbance (e.g., Fritschi et al., 2011). Should the NSS-SF be predictive of who is more susceptible to these noise health consequences, it could also be used to indicate who might benefit more from interventions to curtail noise exposure.
Footnotes
Declaration of Conflicting Interests
The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.
Funding
The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: This research was supported by Cooperative Agreement No. H2380040002, Metrics of Human Responses to Natural Sound Environments, from the National Park Service. Grantees undertaking projects under government sponsorship are encouraged to express freely their findings and conclusions. Points of view or opinions do not, therefore, necessarily represent official National Park Service policy.