The Noise Sensitivity Scale Short Form (NSS-SF), developed in English as a more practical form of the classical Weinstein NSS, has not to date been validated in other cultures, and its validity and reliability have not yet been confirmed. This study aimed to validate NSS-SF in Bulgarian and to demonstrate its applicability. The study comprised test-retest (n = 115) and a field-testing (n = 71) of the newly validated scale. Its construct validity was examined with confirmatory factor analysis, and very good model-fit was observed. Temporal stability was assessed in a test-retest (r = 0.990), convergent validity was examined with single-item susceptibility to the noise scale (r = 0.906) and discriminant validity was confirmed with single-item noise annoyance scale (r = 0.718). The lowest observed McDonald's omega across the studies was 0.923. The cross-cultural validation of NSS-SF was successful but it proved to be somewhat problematic with respect to its annoyance-based items.
Keywords: Confirmatory factor analysis, cross-cultural validation, noise sensitivity questionnaire
|How to cite this article:|
Dzhambov AM, Dimitrova DD. Psychometric properties of the Bulgarian translation of Noise Sensitivity Scale Short Form (NSS-SF): Implementation in the field of noise control. Noise Health 2014;16:361-7
|How to cite this URL:|
Dzhambov AM, Dimitrova DD. Psychometric properties of the Bulgarian translation of Noise Sensitivity Scale Short Form (NSS-SF): Implementation in the field of noise control. Noise Health [serial online] 2014 [cited 2020 Nov 24];16:361-7. Available from: https://www.noiseandhealth.org/text.asp?2014/16/73/361/144409
| Introduction|| |
Noise is a threat to human well-being. In Europe, road traffic noise is the dominant source of noise annoyance.  As a result of modern lifestyle noise pollution is constantly rising,  and it is considered a common cause for psychological distress, cognitive impairment, sleep disturbances, increase in social conflicts, anxiety, nervousness, emotional instability, argumentativeness, etc. ,,,
Psychoacoustics is the science of sound perception, and noise sensitivity (NS) is an important construct in traffic-related psychoacoustics. Some authors define NS as a personality characteristic and a measure of individuals' annoyance caused by the noise around them. , Weinstein defined NS as a personal attribute of sufficient power and generality to permit predictions of reactions to environments encountered for the first time.  Job stated that NS refers to an individual aversion toward and/or reactivity to noise and noisy environments.  It should not be related to hearing acuity, but rather it should reflect a predisposition and evaluative judgment of noise sources,  and some consider it a possible moderator of effect of noise exposure on annoyance. 
The Weinstein Noise Sensitivity Scale (NSS) is a widely used psychometric tool for NS assessment. ,,, However, since in field studies longer questionnaires result in decreased response-rates, Benfield et al. developed the NSS Short Form (NSS-SF), which is psychometrically similar to the longer scale and shows good internal consistency. 
So far noise pollution research in Bulgaria has been primarily focused on noise exposure, disregarding the psychoacoustical dimensions of the problem, or has used single-item NS instruments (for a review see Tzvetkov and Angelova).  Moreover, the psychometrical tools used worldwide are fairly unknown to Bulgarian scholars. To the best of our knowledge, there is no psychometrically sound NS measurement instrument available for use by Bulgarian specialist in environmental hygiene, social medicine, and psychology. On the other hand, NSS-SF has not to-date been validated in other cultures, and its validity and reliability have not yet been confirmed.
Since exposure to noise is a hazard to human well-being, understanding the components of noise and psychological perceptions of noise as they relate to the mentioned above outcomes becomes essential. Based on the findings of Benfield et al.,  we hypothesized that the Bulgarian translation of NSS-SF (BNSS-SF) would prove reliable and valid instrument.
The aim of this study was to validate NSS-SF in Bulgarian and to demonstrate its applicability.
| Methods|| |
This research was carried out using:
- BNSS-SF (available upon request in Bulgarian);
- One-item 11-point numerical scale ("not annoyed at all " to "extremely annoyed ") measuring noise annoyance according to ISO 15 666;  and
- One-item 11-point numerical scale ("not noise-sensitive at all " to "very noise-sensitive ") measuring individual susceptibility to noise, according to Zimmer and Ellermeier. 
Different scaling of the instrument for NS and annoyance intended to minimize the common method bias.
The English version of NSS-SF (courtesy of Dr. Benfield) was adapted and translated according to the widely accepted guidelines for successful translation of instruments in cross-cultural research.  Differences in the original and the back-translated versions were discussed and resolved by joint agreement of both translators and a third independent reviewer. The level of inter-rater agreement was measured with Krippendorff's alpha (α = 0.921), which suggested high inter-rater agreement. 
The five items - expressing attitudes toward noise in general and emotional reactions to environmental sounds encountered in the everyday life - are measured on 6-point bipolar Likert scales ("strongly disagree " to "strongly agree "), where higher scores indicate higher NS. Only direct coded questions were administered.
Three questionnaire studies along with the data processing and linguistic validation took place in the period 17 February to 31 May, 2013. They comprised a test-retest and field-test of BNSS-SF.
Referring to the scope of our studies, we used convenience sample (not representative) for the test-retest study, including people with different socioeconomic status, demographic characteristics, occupation, place of residence (city district), etc. from our circle of friends, relatives, and acquaintances.
For the field-test, we used the noise monitoring system and strategic noise maps of Plovdiv Municipality to determine three neighborhoods with constantly high traffic-generated noise levels and then surveyed 71 residents of those neighborhoods.
When answering the questions, the respondents in the three studies were asked to think of the past 6 months.
It was estimated that in order to detect even small effect size of 0.10, 87 participants were needed (0.80 statistical power, one latent variable, five observed variables and α = 0.05). Nevertheless, the minimum sample size for the test-retest study had to be at least 100. Posthoc statistical power for the field-test justified its sample size. According to Gignac, structural equation modelling is feasible with samples as small as 100 and even smaller samples are not problematic if some statistical criteria are met. 
Administration of the questionnaires
The questionnaires were E-mailed to the participants in the test-retest study after preliminary discussions of their participation. They were asked to report in the response E-mails the date at which they completed the questionnaire, which helped us to synchronize and schedule the retest. For each individual, the retest was in about 5 weeks (35 days) interval after the initial test like Benfield et al. did in their validation of NSS-SF.  After the completion of BNSS-SF five respondents participated in a debriefing session and none of them reported any difficulty in handling the questionnaire.
In the field-test, the interviewees were randomly selected (each third person) and interviewed in the local parks and green spaces of their neighborhood by the first author during their recreational time at randomly selected days of the week (1-23 May, 2013).
Characteristics of the participants for the test-retest and field studies
A total of 115 participants completed BNSS-SF (78% response rate). Their mean age was 45.42 (standard deviation [SD] = 15.88, standard error [SE] = 1.43) years; 51.30% were men; 17.39% had upper secondary, 75.65% master/bachelor and 6.96% doctoral degree; 9.57% were in the lower, 77.39% in the middle and 13.04% in the upper socioeconomic class.
The 71 participants in the field-test of BNSS-SF were residents of three distinct neighborhoods of Plovdiv (Central, Northern, and Western districts) (59% response rate). Men (47.89%) (official statistics: ≈48%) and women (52.11%) were equally represented in the sample. The mean age was 45.41 years (SD = 18.34, SE = 2.13). The distribution of the participants across age ranges was: 16.90% were 18-25 (official statistics: ≈9%); 18.31% were 26-35 (official statistics: ≈20%); 18.31% were 36-45 (official statistics: ≈19%); 14.08% were 46-55 (official statistics: 17%); 14.08% were 56-65 (official statistics: ≈16%) and 18.31% were over 65 years of age (official statistics: ≈19%). The biggest group of participants pertained to the middle socioeconomic class (54.93%), comprised employed citizens (40.85%), married/living with a spouse (61.97%), and with upper secondary or bachelor/master educational degree (92.95%) (official statistics: ≈62%). Given those sociodemographic characteristics, we concluded that the sample in the field-test was somewhat representative of Plovdiv's population after adjustments for the age range 18-65+, when compared to official data from the National Statistical Institute (http://www.nsi.bg/en) and Municipal Plan for Development of Plovdiv 2005-2013 (http://www.webcitation.org/6PEpoVL55). Some of the discrepancies are due to the fact that only adults participated in the study.
The inclusion criteria were for them to have lived at least 10 years in the neighborhood and in 50 m radius from traffic-heavy streets selected after visual inspection of the strategic noise maps of Plovdiv retrieved from the Municipality website.
The study was sociomedical and noninterventional, voluntary and performed after obtaining informed consent and therefore was not subjected to ethical evaluation by the University Committee, which was acceptable.
Data analytic strategy and statistics
The data were screened for univariate and multivariate (Mahalanobis distances) outliers, and they were winsorized. 
Likert scale variables were treated like interval to allow the use of common parametric tests. , Missing values (<2%) were tested for response patterns with Little's missing completely at random test and replaced using expectation-maximization algorithm. The distribution of the items was assessed with graphical analysis, skew and kurtosis and a variant of Small's omnibus test of multivariate normality.  Correlation coefficients were compared statistically.  In order to make inferences without making strong distributional assumptions, bootstrapping method with bias-corrected confidence estimates was used. The criterion for the significance level of P < 0.05 (two-tailed) was used.
A confirmatory factor analysis (CFA) was performed using maximum likelihood estimation (because the data were normally distributed) to test the model fit of the hypothesized one-factorial structure of BNSS-SF. The latent factor was scaled using Item 1 as marker indicator. Absolute close-fit indexes (Chi-squared test, root mean square root of approximation [RMSEA], and standardized root mean square residual [SRMR]) and an incremental close-fit index (comparative fit index [CFI]) are reported with cutoff criteria - low χ2 relative to degrees of freedom with a non-significant p value (P > 0.05),  RMSEA ≤ 0.06, CFI ≥ 0.95, and SRMR ≤ 0.06.  The internal-consistency of the questionnaire was assessed with McDonald's ω coefficient, which is recommended over Cronbach's α. , BNSS-SF is reported as unit-weighted composite score divided by the number of items, in order to interpret it on the same scale as the individual items.
All statistical analyses were performed with SPSS, SPSS AMOS (IBM SPSS Statistics for Windows, Version 21.0, 2012. Armonk, NY, USA), and FACTOR v.9.2. 
| Results|| |
The mean BNSS-SF score was 3.16 (SD = 1.07, SE = 0.099, 95% confidence intervals [CI]: 2.96, 3.35). Omega was 0.979, and the inter-item correlations were in the range 0.787-0.875 which suggested very good internal consistency of the questionnaire. Skew (-0.034 to -0.303) and kurtosis (-0.402 to -0.644) were within the acceptable range for CFA.  A variant of Small's omnibus test of multivariate normality was confirmed (VQ3 (10) = 10.46, P = 0.402). BNSS-SF did not differ across educational degrees, and socioeconomic classes, between men and women and it was not correlated with age.
A CFA suggested very good fit between the model and the observed data: χ2 (5) = 7.55, P = 0.183, RMSEA = 0.067, CFI = 0.996, and SRMR = 0.011. All measurement error was presumed to be uncorrelated. The model was over-identified (5 degree of freedom). The squared multiple correlations ranged from 0.77 to 0.89, indicating good reliability of the observed variables in relationship to the latent construct. The model fit was achieved without any posthoc modifications because of good fit indices and residual statistics. The largest standardized residual covariance was ǀ0.152ǀ suggesting standard normal distribution. All standard errors were below 0.34. The structural model of BNSS-SF is presented in [Figure 1].
|Figure 1: Structural model of the confirmatory factor analysis with Bulgarian translation of Noise Sensitivity Scale Short Form as latent variable and the five items as observed variables. Note: n = 115. All factor loadings (not italicized) are standardized, positive, and significant (P < 0.01); the squared multiple correlations are italicized; BNSS-SF — Bulgarian translation of Noise Sensitivity Scale Short Form; e1-e5 - error-terms|
Click here to view
In order to confirm discriminant validity of BNSS-SF in the retest, we measured individual noise annoyance of the participants (M = 5.10, SD = 1.91, SE = 0.18, 95% CI: 4.75, 5.48). The two concepts are related but are distinguished mainly by the fact that NS should have nonacoustical antecedents contrary to annoyance.
In this study, we tested the temporal consistency of BNSS-SF. The mean BNSS-SF score was 3.36 (SD = 1.18, SE = 0.11, 95% CI: 3.13, 3.57). Omega was 0.960 (r(113) = 0.843-0.946).
In the retest the factorial simple solution was retained. There were strong self-correlations between each item in the pretest and retest [Table 1]. The temporal stability of BNSS-SF in our study [Table 1] (correlation no. 6) was significantly higher (z = 8.329, P < 0.001) than that observed in the study of Benfield et al.  (r = 0.83). The noise annoyance score was as follows: M = 5.30, SD = 1.83, SE = 0.17, 95% CI: 4.97, 5.62. It was positively correlated with BNSS-SF (r(113) = 0.702, P < 0.001, 95% CI: 0.56, 0.80).
Discriminant validity was assessed using the correction for attenuation formula: rxy/√rxx * ryy , where rxy is the correlation between BNSS-SF and noise annoyance (in the retest), rxx is the reliability of BNSS-SF and ryy is the reliability of the 11-point noise annoyance scale.  Because noise annoyance was measured on a one-item scale (omega coefficient could not be computed), the "test-retest" correlations were considered measures of reliability. The corrected correlation between BNSS-SF and noise annoyance was 0.718 (<0.85), indicating good discriminant validity. 
After we had confirmed the high internal-consistency, temporal consistency and stable one-factorial structure of BNSS-SF, we wanted to field-test it in a representative community sample and to establish nomological and convergent validity.
The mean BNSS-SF composite score was 3.75 (SD = 1.11, SE = 0.13, 95% CI: 3.49, 4.02). BNSS-SF was normally distributed. The CFA suggested very good model fit for the one-factorial solution of BNSS-SF: χ2 (5) = 6.171, P = 0.290, RMSEA = 0.058, CFI = 0.996, and SRMR = 0.026. Omega of 0.923 and inter-item correlations in the range 0.556-0.821 indicated high internal consistency of the scale. [Table 2] presents the correlation matrix of the five items of BNSS-SF.
BNSS-SF scores were higher in younger people (r(69) = -0.363, P = 0.002, 95% CI: -0.649, -0.077) and in those with lower educational degree (rs = -0.322, P = 0.006, 95% CI: -0.524, -0.105). However, the education might be a function of age and cannot be interpreted separately in its relation to NS. NS did not differ across occupations and between men and women, but it was associated with the civil status of the participants (F(3,67) = 3.66, P = 0.017, ηp2 = 0.14) and it was significantly higher (P < 0.05) in single people compared to widows/widowers and divorced people.
Convergent validity was tested by correlating BNSS-SF with a relevant to the assessed construct - one-item tool measuring the susceptibility to noise (NS).  Strong positive correlation was found between BNSS-SF and the one-item measure (r(69) = 0.906, P < 0.001, 95% CI: 0.859, 0.940), which was significantly stronger than that found by Zimmer and Ellermeier (z = −5.326, P < 0.001). 
Nomological validity was demonstrated in real-life settings. In general, this is achieved by examining the predictive ability of the scale within the nomological network of antecedent and consequent variables.  Because the antecedents of NS are yet to be confirmed in the literature, in our study we sought the ability of BNSS-SF to predict noise annoyance. BNSS-SF significantly predicted noise annoyance (B = 1.50, SE = 0.125, β = 0.787, t = 10.60, P < 0.001) and the model explained over 60% of the variance in the dependent variable (adjusted R2 = 0.614, F(1,69) = 112.36, P < 0.001). The observed power was 1.00, which suggested adequate sample size.
| Discussion|| |
Overall, the CFA suggests that the hypothesized one-factorial solution of BNSS-SF is reliable psychometric tool, which can be used in appraising people's NS and its relation to noise annoyance. BNSS-SF demonstrated high factor loadings and internal consistency. It was highly correlated with NS measured on a one-item scale providing "finer grain of response alternatives,"  and with noise annoyance that added another level of validity. Our CFA fit-indices - CFI and RMSEA in the field-test and only CFI in the test-retest - were better than those reported by Benfield et al.  according to the qualitative criteria of Chen (ΔRMSEA > -0.015) and Cheung and Rensvold (ΔCFI > +0.01). , In the model of Benfield et al.  CFI was 0.97 and RMSEA was 0.078. Compared to the results of Benfield et al.,  our CFA produced higher loadings and internal consistency measures across the three samples. However, understanding the discrepancies may require other studies implementing BNSS-SF and NSS-SF in both languages in order to draw firm conclusions about the scale's cross-cultural performance above and beyond the random fluctuations due to chance, sample size, common method variance, etc. No mechanical modifications were necessary for the development of NSS-SF,  which reiterates the adequacy of the scale in measuring the latent construct in a cross-cultural context, because we did not modify it either. The comparison of our validation of NSS-SF with other cross-cultural validations of the scale was not possible, because we found no information about NSS-SF's having been validated in languages other than English. Thus, our findings might be considered contribution to the field.
Noise sensitivity was stronger in younger people, which is generally the case in the literature, but only in our field-test, probably because the sample was more representative of the real sociodemographic profile of the population than the test-retest sample. Sex has been controversially linked with NS, so it was not concerning that in our sample BNSS-SF did not differ across sexes.
If we are to understand the buffering potential of urban environments (both natural and built) to attenuate noise annoyance and thus promote citizens' positive health, it will be a fallacy to simply study the relationships of those environmental features with noise annoyance or the subsequent health outcomes, because the correlations will vary from study to study as those buffers most likely have moderating role on various links in the relationships noise exposure - NS-noise annoyance. In order to adequately predict the impact of, say, urban parks and greenery in residential areas on noise annoyance, we have to conceptualize a more complex model. Moreover, we speculate that NS might not only shed some light into the complex psychoacoustical relationships, but it might be modified by changing the environment, and thus, our reactivity to noise could be attenuated.  This is highly warranted since people's self-regulated NS might be important for predicting the rate of annoyance caused by traffic noise. 
The implementation of the European recommendations for noise control in Bulgaria gives scientists the opportunity to become more involved in noise pollution research, for example, Plovdiv Municipality bestows strategic noise maps of the city and access to the yet limited, but very informative noise monitoring system. Collaboration between public health sector and municipality experts could result in most favorable outcomes. We believe that the experts in environmental hygiene, social medicine and psychology should be given the proper research tools to address the multi-dimensional phenomenon of noise pollution. We hope that BNSS-SF will contribute to broadening the current research interest of public health experts. In order to promote and facilitate further research on NS different psychometric instruments need to be validated in order to be cross-checked.
Strengths and limitations
Strengths of our study are the adequate methodological approach and the validation of the scale at different levels (temporal consistency, linguistical validity, nomological, convergent, and discriminant validity). Our study is the first cross-cultural validation of NSS-SF and might serve in promoting its use in field research. The justified choice of statistical analyses is also superior to the commonly used techniques in the validation and development of psychometric instruments in Bulgaria. We chose to perform CFA, test-retest correlations and internal consistency estimation in order to ensure comparability of the results with the original NSS-SF. Nevertheless, correlations are not an optimal procedure for the estimation of reliability, and other alternatives should be considered in future. Cronbach himself vouched for the Generalizability Theory as a superior technique to estimate the different sources of error variance,  and it has been applied to a NS construct by Schütte et al. 
Although some authors consider coefficient omega unreliable in a sample size <1000,  it has been successfully used in similar to our sample sizes. , Omega does not share the assumptions of Cronbach's α leading to more inaccurate internal consistency reliability estimates.  It is interesting that in our field-test sample widows/widowers and divorced people were less noise sensitive than those who were single. It was also beyond the scope of this research to compare BNSS-SF with the original 21-item Weinstein's scale. The test-retest coefficients as reliability measures, which were used in the discriminant validity analysis, do not assess specific factor error.  However, for the single-item scale measuring noise annoyance omega coefficient could not be computed. This should not noticeably affect our results, especially giving the consistent performance of the scale in the other validity diagnostics. The test-retest correlations were extremely high, which is probably due to the fact that the participants could have checked their previous E-mail responses, especially given that they were not blinded to the objectives of the study. Nevertheless, high temporal stability of the NS construct has been reported previously (0.8-0.9). , The response rate in our field-test was lower than that reported by Benfield et al. in study No 4 (67%); the same probably refers to our test-retest convenience sample in comparison to theirs, although they did not report that response rate. 
Finally, the very high correlation between NS and noise annoyance is probably due to the items "I get annoyed when my neighbors are noisy" and "I get mad at people who make noise that keeps me from falling asleep or getting work done", which are to some extent measuring annoyance and may thus be dependent on the actual noise exposure.  Those items represent 40% of NSS-SF and along with another similar annoyance-based item, only 14% of Weinstein's 21-item scale, which could explain our discordance with the literature. Also these two items might be affecting the way people respond to the other three items. This implies the necessity to modify BNSS-SF. Likewise, more sophisticated instruments such as the Noise Sensitivity Questionnaire (NoiSeQ) are being adapted in order to cross-validate them with BNSS-SF and to seek for integration between the brevity of BNSS-SF and the multidimensionality of NoiSeQ. 
| Conclusion|| |
The cross-cultural validation of NSS-SF was successful but it proved to be somewhat problematic with respect to its annoyance-based items. The Bulgarian version of the questionnaire is free and can be of use to specialists in environmental hygiene, social medicine and psychology when addressing the public health problem of noise pollution. We recommend that the structural integrity of BNSS-SF should be assessed each time it is used and that any analyses including BNSS-SF should be common-method-bias-adjusted (preferably at design level) until sufficient data about its performance has been accumulated.
| References|| |
Babisch W. Noise and health. Environ Health Perspect 2005;113:A14-5.
Ohrström E. Longitudinal surveys on effects of changes in road traffic noise-annoyance, activity disturbances, and psycho-social well-being. J Acoust Soc Am 2004;115:719-29.
Goines L, Hagler L. Noise pollution: A modem plague. South Med J 2007;100:287-94.
Hume KI, Brink M, Basner M. Effects of environmental noise on sleep. Noise Health 2012;14:297-302.
Fyhri A, Klaeboe R. Road traffic noise, sensitivity, annoyance and self-reported health - a structural equation model exercise. Environ Int 2009;35:91-7.
Paunovic K, Jakovljevic B, Belojevic G. Predictors of noise annoyance in noisy and quiet urban streets. Sci Total Environ 2009;407:3707-11.
Weinstein ND. Individual differences in reactions to noise: A longitudinal study in a college dormitory. J Appl Psychol 1978;63:458-66.
Job RF. Noise sensitivity as a factor influencing human reaction to noise. Noise Health 1999;1:57-68.
Ellermeier W, Eigenstetter M, Zimmer K. Psychoacoustic correlates of individual noise sensitivity. J Acoust Soc Am 2001;109:1464-73.
Stansfeld SA. Noise, noise sensitivity and psychiatric disorder: Epidemiological and psychophysiological studies. Psychol Med Monogr Suppl 1992;22:1-44.
Miedema HM, Vos H. Noise sensitivity and reactions to noise and other environmental conditions. J Acoust Soc Am 2003;113:1492-504.
Zimmer K, Ellermeier W. Psychometric properties of four measures of noise sensitivity: A comparison. J Environ Psychol 1999;19:295-302.
Benfield JA, Nurse AG, Jakubowski R, Gibson AW, Taff, BD, Newman P, et al
. Testing noise in the field: A brief measure of individual noise sensitivity. Environ Behav 2012;20:1-20.
Tzvetkov D, Angelova M. Infrasound, Ultrasound, Noise and Vibrations. 2 nd
ed. Sofia: Medicina i fizkultura; 2007. [in Bulgarian].
Brislin RW. Back translation for cross-cultural research. J Cross Cult Psychol 1970;1:185-216.
Hayes AF, Krippendorff K. Answering the call for a standard reliability measure for coding data. Commun Methods Meas 2007;1:77-89.
Gignac G. Self-reported emotional intelligence and life satisfaction: Testing incremental predictive validity hypotheses via structural equation modeling (SEM) in a small sample. Pers Individ Dif 2006;40:1569-77.
Hoaglin DC, Iglewicz B. Fine tuning some resistant rules for outlier labeling. J Am Stat Assoc 1987;82:1147-9.
Bowling A. Research Methods in Health. Buckingham: Open University Press; 1997.
Norman G. Likert scales, levels of measurement and the "laws" of statistics. Adv Health Sci Educ Theory Pract 2010;15:625-32.
DeCarlo LT. On the meaning and use of kurtosis. Psychol Methods 1997;2:292-307.
Weaver B, Wuensch KL. SPSS and SAS programs for comparing Pearson correlations and OLS regression coefficients. Behav Res Methods 2013;45:880-95.
Barrett P. Structural equation modelling: Adjudging model fit. Pers Individ Dif 2007;42:815-24.
Hu LT, Bentler PM. Cutoff criteria for fit indexes in covariance structure analysis: Conventional criteria versus new alternatives. Struct Equ Modeling 1999;6:1-55.
Gignac GE. On the inappropriateness of using items to calculate total scale score reliability via coefficient alpha for multidimensional scales. Eur J Psychol 2013. Available from: http://www.drgillesgignac.blogspot.com.au/
. [Last accessed on 2013 Oct 01; Last cited on 2013 Sep 13].
McDonald RP. Test Theory: A Untied Treatment. Mahwah, NJ: L. Erlbaum Associates; 1999.
Lorenzo-Seva U, Ferrando PJ. FACTOR: A computer program to fit the exploratory factor analysis model. Behav Res Methods 2006;38:88-91.
Muthen B, Kaplan D. A comparison of some methodologies for the factor analysis of non-normal Likert variables. Br J Math Stat Psychol 1985;38:171-89.
Spearman C. The proof and measurement of association between two things. Am J Psychol 1904;15:72-101.
Brown T. Confirmatory Factor Analysis for Applied Research. New York: Guilford Press; 2006.
Bhattacherjee A. Individual trust in online firms: Scale development and initial test. J Manag Inf Syst 2002;19:211-41.
Chen FF. Sensitivity of goodness of fit indexes to lack of measurement invariance. Struct Equ Modeling 2007;14:464-504.
Cheung GW, Rensvold RB. Evaluating goodness-of-fit indexes for testing measurement invariance. Struct Equ Modeling 2002;9:233-55.
Dzhambov AM. Neighbourhood quality as a modifier of environmental noise perception: Bringing together environmental hygiene/medicine and environmental psychology. [submitted manuscript].
Raw GJ, Griffiths ID. Individual differences to road traffic noise. J Sound Vib 1988;121:463-71.
Cronbach LJ, Shavelson RJ. My current thoughts on coefficient alpha and successor procedures. Educ Psychol Meas 2004;64:391-418.
Schütte M, Marks A, Wenning E, Griefahn B. The development of the noise sensitivity questionnaire. Noise Health 2007;9:15-24.
Ten Berge JM, Socan G. The greatest lower bound to the reliability of a test and the hypothesis of unidimensionality. Psychometrika 2004;69:613-25.
Schroeders U, Wilhelm O. Testing reasoning ability with handheld computers, notebooks, and paper and pencil. Eur J Psychol Assess 2010;26:284-93.
Rekart KN, Mineka S, Zinbarg RE. Autobiographical memory in dysphoric and non-dysphoric college students using a computerised version of the AMT. Cogn Emot 2006;20:506-15.
Schmidt FL, Le H, Ilies R. Beyond alpha: An empirical examination of the effects of different sources of measurement error on reliability estimates for measures of individual differences constructs. Psychol Methods 2003;8:206-24.
Kishikawa H, Matsui T, Uchiyama I, Miyakawa M, Hiramatsu K, Stansfeld SA. The development of Weinstein′s noise sensitivity scale. Noise Health 2006;8:154-60.
Angel M Dzhambov
MMSc Student, Faculty of Medicine, Medical University of Plovdiv, 15-A Vasil Aprilov Blvd.,
Source of Support: None, Conflict of Interest: None
[Table 1], [Table 2]