ll 1 ‘1' p f 1}) ' ill 'I 'I l [I ”I‘M ‘1 |l H U ‘4" W ~ :I —| _s .mooo I FREQUENCY AND {EQMQUEXTY 63F wavfifi ‘ SGU'NEDS 133* AS FUNC’NONS G?" AN QPT‘EMUM *t’v’ARMNG :$YS:... . ' f " . .. E“. v. :2 4n“ Tnm‘s :3: ma “3.9.56; Cu In; A. "2 «*r ~'*:*+§=:s: 32‘! \J' .5‘3\.’C'.“fd{3z:i.’fx 371m: :2: {5 ’2715r'33‘ THE!!! LIBRARY Michigan State University ABSTRACT FREQUENCY AND COMPLEXITY OF HARMONIC SOUNDS AS FUNCTIONS OF AN OPTIMUM WARNING SYSTEM by Bruce Donald Olsen The purpose of this study was to determine whether any significant differences exist in subjects rating of the alerting potential between harmonic complex tones made up of two, three, four, and fivegxuwetones half an octave apart; and if any significant differences exist in subjects rating of the alerting potential between harmonic tones with frequencies ranging from 700 cps to 4000 cps. The subjects participating in this study were twenty- four students at Michigan State University who were tested and found to have normal hearing thresholds. The subjects were asked to rate eighty-eight different sounds in relation to nine characteristics. Each characteristic was rated along a five point continuum allowing the subject to deter- mine the strength of a sound in relation to each charac- teristic. These sounds were played on a tape recorder and were heard through a pair of earphones. The findings of this study indicate that there is a significant difference in subject ratings of alerting potential between complex harmonic sounds. The more complex the soundthe better it was rated by the subjects as an BRUCE DONALD OLSEN alerting signal. The range between 1800 cps and 2200 cps for the lowest frequency of complex harmonic soundsvmus also shown by subjects rating of alerting potential to be an optimum range for alerting signals. It would appear then, that from the results of sub- jects rating of alerting potential, the complexity and frequency of harmonic sound8pflay a significant role in determining if they are suitable for an optimum warning system. FREQUENCY AND COMPLEXITY OF HARMONIC SOUNDS AS FUNCTIONS OF AN OPTIMUM WARNING SYSTEM By Bruce Donald Olsen A THESIS Submitted to Michigan State University in partial fulfillment of the requirements for the degree of MASTER OF ARTS Department of Speech 196A TABLE OF CONTENTS Page LIST OF TABLES . . . . . . . . . . . . . iii LIST OF FIGURES . . . . . . . . . . . . . iv LIST OF APPENDICES . . . . . . . . . . . . v Chapter I. STATEMENT OF THE PROBLEM 1 Introduction . . . . . . . . . . l Hypotheses . 3 Importance of Study . 3 Statement of Problem and Purpose of Study . . . . . A Definition of Terms . . . A II. REVIEW OF THE LITERATURE . . . 7 Deve10pment of Studies in the Area . . . 7 Studies in Alerting Potential . . 9 Criterion of Observer . . . 10 III. SUBJECTS, EQUIPMENT, AND TESTING PROCEDURES 13 Subjects 13 Equipment. 13 Procedure. 14 IV. RESULTS AND ANALYSES. 21 Introduction. 21 Analyses . . 21 Discussion of Results. 30 Discussion 32 V. SUMMARY AND CONCLUSIONS. 3A Summary 34 Conclusions . 35 Areas for Further Study 36 BIBLIOGRAPHY . . 37 APPENDICES. . 39 ii Table LIST OF TABLES Intensity Curve of Four Randomly Selected Pairs of Earphones at 500 cps, 1000 cps, 2000 cps, 3000 cps, and 4000 cps . Format of ReSponse Sheet Utilized to Determine Relevant Dimensions for Rating Alerting Characteristics of Audio Signals . . . . . . . Computations for Analysis of Variance of Intensity. . . . . . . . . Comparisons of Means of Sound Complexities Measured by Subject Judgment . . . Computation for Analysis of Variance of Lowest Frequency Component . . . . . Differences Between all Pairs of Means Com- puted for Two Component Sounds. iii Page 17 2O 24 27 28 31 Figure LIST OF FIGURES Block Diagram of Instrumentation Utilized for Recording Audio Signals on Magnetic Tape . . . Curve of Best Fit of Four Pairs of Ear— phones. . . . . . . . . . . The Average of Subjects Rating of Alerting Potential of Two, Three, Four, and Five Component Sounds Across Twenty-four Subjects Two, Three, Four, Five Component Sounds of Subjects Rating of the Alerting Potential Averaged Across Twenty-four Subjects, a Five Point Continuum iv on Page 15 l9 25 29 LIST OF APPENDICES Appendix A. Directions to Subjects B. List of Individual Tones Comprising the Forty—Four Complex Sounds. . . . C. Raw Data, Averaged over Eighteen Scales for Each Sound by Subject Page A0 A3 A6 CHAPTER I STATEMENT OF THE PROBLEM Introduction A study was completed at Michigan State University, September, 1963, for the Office of Civil Defense, Contract No. OCD-OS-62-182 by Oyer and Hardick. The title of the project was "Response of Population to Optimum Warning Signal." The project isolated characteristics that were believed to be useful as a warning system by subjects in the experiment. Because of time limitations and the broad focus of the research objectives outlined in the contract, it was not feasible to attempt to explore in depth some areas that required further study. Therefore it is recommended that the audio warning signal selection process be continued within the ranges of frequency, intensity and time Specified in the present investigation.1 One of the suggestions for further investigation was to determine the differential effects of signal complexity and relationships of components comprising the spectrum on 2 judgments of alerting potential of audio signals. It was lHerbert J. Oyer and Edward J. Hardick, "Response of Population to Optimum Warning Signal,” Office of Civil Defense, Contract No. OCD-OS—62—182, September, 1963, p. 7. 21bid. from this suggestion that the present study evolved. This study involves determining frequency and complexity of tones with harmonic relationships that would be most suc- cessful as Optimum warning signals. Areas that should be explored in greater depth are those that involve determination of more refined estimates of signal characteristics based on the ranges determined by the present study.1 The frequency range studied there- fore was limited to the area suggested by research results of the study by Oyer and Hardick. The optimal warning signal should be within the range of 700 cps to #000 cps.2 Since the research findings indicate that isolated pure tones do not have as great an alerting potential as complex signals,3 all sounds in this study were complex sounds being made up of two or more pure tones that were harmonically related. The intensity of the harmonic sounds used was held constant with an output of 70 dB (re: .0002 dyne/cm2) through earphones. Two variables were therefore isolated for study, difference of frequency and difference of complexity of harmonic complex sounds. l1bid., p. 3. 2Ibid., p. u. 31bid. Hypotheses In order to determine results from this study, the following null hypotheses have been proposed. 1. There are no statistically significant differ- ences between the means of subjects rating of the alerting potential of complex harmonic sounds of two, three, four, and five components, as measured on a nine point scale along a five point continuum. 2. There are no statistically significant differ- ences between the means of subjects rating of the alerting potential of different frequencies of harmonic sounds, as measured on a nine point scale along a five point continuum. Importance of Study Little attention has been focused on human reSponses as criteria for determining optimal audio alerting signal characteristics.1 This area has grown in importance as greater emphasis is placed on civil defense. The impetus was the advent of World War II when it was necessary to provide warning in areas far removed from the zone of combat. In addition, it was necessary to provide warning to large numbers of people, many of whom lived in noisy metropolitan areas. It is rather obvious that efficient alarms sounded in time for people to take self-protective measures lIbid., p. 7. 2Ibid., p. 12. are important for the maintenance of high morale, and should decrease the possibility of panic among the public.1 It is for these reasons that studies in this area are important. State of Problem and Purpose of Study This study attempted to determine whether any signifi- cant differences exist in subjects rating of the alerting potential due to changes of the lowest frequency component of complex sounds. In order to investigate possible differences in these areas, subjects attending Michigan State University, between the ages of twenty-two and thirty-four, and found by testing to have normal hearing, participated by judging eighty-eight sounds. The sounds were confined to frequencies of between 700 cps and 4000 cps. The sounds were each heard twice (eighty-eight sounds in total) and were judged on a nine point scale along a five point continuum. Definition of Terms Complex Tones: Two or more tones between 700 cps and A000 cps combined so that they are harmonically related in half octave steps. 1Federal Civil Defense Administration. Final Report: The Effectiveness of Sonic Outdoor Warning Devices. Novem- ber, 1953, p. 17. Optimum Warning Signal: In order to have a warning system it must first be an alerting signal. An alerting signal should have "grabbing power," it should be an I "attention getter,’ should have "distracting ability." Operationally defined, it is used to attract attention. Warning is defined as alerting people to the threat of extraordinary danger and the related effects of disaster. An alerting signal can pro- vide warning to the population only when the sig- nal is associated with the threat of danger and the related effegts of disaster through a condi- tioning process. Scaling: Scaling is defined as the ordering of stimuli along one or more continua on the basis of magnitude of human reSponse.3 Stevens states the assignment of numerals 4 (numbers) to aspects of objects or events creates a scale. Decibel (dB): Although the absolute amount of sound intensity can be measured, it is more common to express it in terms of the ratio between the intensity of the sound being measured and a stan— dard reference intensity, which for audiologic convenience compares closely with the threshold of the human ear. -The most common reference point from which sound intensities are computed is a power of ten to the minus sixteen watt per 8 uare centimeter, or a pressure of .0002 dyne/cm2 two tenthousanths dyne per square dentimeter) . lOyer and Hardick, op. cit., p. 23. 2Ibid., p. 12. 3Ibid., p. 10. 4S. S. Stevens (ed.), Handbook of Experimental Psychology (New York: John Wiley and Sons, Inc., 1951), p’ 230 The ratio between this reference point and any given intensity is expressed in t rms of a log— arithmic unit called the decibel. Harmonics: When the frequencies of the partials are integral multiples of the lowest or fundamental frequency they are called harmonics.2 Equal Loudness: InSpection of the loudness curve shows that there is only a slight difference in loudness between 700 cps and 4000 cps when intensity is held con— stant at 70 dB. For this study, equal loudness was defined as a constant intensity level of 70 dB.3 lHaves A. Newby, Audiology (New York: Appleton—Century— Crofts, Inc., 1958), p. 12. 2Giles w. Gray and Claude M. Wise, The Bases of Speech (New York: Harper and Brothers, 1959), p. 114. 3Harvey Fletcher, Speech and Hearing in Communication (New York: D. Van Nostrand Company, Inc., 1961), p. 188. CHAPTER II REVIEW OF THE LITERATURE Development of Studies in the Area There is very little published research concerning the evaluation of auditory stimuli within the con- text of alerting or warning. Much of the reported research has been concerned with engineering problems related to signal generation and propagation. Little attention has been focused on human responses as cri— teria for determining optimal audio warning signal characteristics.’ Some of the reports reviewed in this chapter are taken from a secondary source since they were not available to this writer. In 1942 Volkmann and Graham reported a designed to determine the audibility sound level requirements of alarm signals in the quest for an adequate signal that would give basic outdoor coverage and supplementary indoor coverage for air raid warnings. As the center frequency was changed, noticeable changes with frequency occurred due to masking characteristics of the noise. The most effective center frequency for noise penetration was around 2000 CpS. Since in the case of local coverage no noticeable loss occurrs for the transmissions of high frequencies in the air, a signal frequency of approximately 2000 cps was found to be desireable. lOyer and Hardick, op. cit , p. 7. 2 n J. E. Volkmann and M. L. Graham, A Survey on Air Raid Alarm Signals,” Journal of the Acoustic Society of America, 14 (1942), p. l. Houston and Walker conducted tests to determine the probability of hearing the various signals in a background of simulated aircraft noise. A variety of noise Spectra were used. A pure tone of 2500 cps, presented intermittently through earphones was found to be the most effective warning 1 signal. A study Sponsored by the Federal Civil Defense Admin- istration stated that a multitone signal showed promise of being effective, eSpecially a signal with a two~tone inhar- 2 monic relationship. This report, according to Oyer and Hardick, did not represent experimentation designed to yield data reflecting differential human response to audi- tory signal characteristics as pertaining to alerting. In a study conducted by Solomon, he reported that: Twenty passive sonar recordings were ranked by fifty subjects in terms of the aurally perceived (psychologi- cal) characteristics on seven different dimensions. Correlations were run between octave band sound pres— sure level measurements of the sounds and their rank orders on seven psychological dimensions. Meaningful relationships were found between ranks on certain psychological dimensions and energy concentration within certain octave bands. Also, an analysis was made of the manner in which the twenty sounds clustered within the Space defined by seven psychological dimensions. Analyses of 1R. C. Houston and R. Y. Walker, "The Evaluation of Auditory Warning Signals for Aircraft,” Air Force Techni— cal Report No. 5762 (June, 1949) as quoted in Oyer and Hardick, op. ci§., p. 13. 2Federal Civid Defense Administration. Final Report: op. cit., p. 19. these sound clusters revealed that the rhythmic beat pattern of stimulus is a principal attribute upon which sonar men base their judgments of similarity. Erlick and Hunt reported that a desirable warning signal should: (1) be easily detectable; (2) hold attention; (3) be quickly and accurately identifi— able; and (4) be infinitely retainable as a function of time with regard to meaning. They indicated that pure tones particularly of frequencies above 1000 CpS have a tendency to interfere with operator effectiveness. They implied that research in this area should center on the evaluation of sounds al- ready used for "attention getting" purposes. These sounds, they stated, employ the fundamental tech- niques used to gain perceptual attention. Studies in Alerting Potential An area of importance to optimum warning is the alerting of the population. Studies in this area will be discussed in relation to this study to serve as a broad foundation from which this study evolved. ryter reported that the annoyance of noise appeared to vary in relation to (1) unexpectedness; (2) interfer- ence with auditory behavior; (3) inappropriatness; (4) intermittency; (5) reverberation; (6) loudness; (7) frequency pattern.3 His earlier results indicated 1L. N. Solomon, ”Search for Physical Correlates to Psychological Dimensions of Sounds,” Journal of Acoustical Society of America, 31, no. 4 (1959), p. 492. 2D. E. Erlick and D. P. Hunt, Evaluation of Audio Warning Displays of Weapons Systems. Wright Air Development Division, Technical Report 60—814 March, 1961, as quoted by Oyer and Hardick, op. cit., pp. 13-15. 3K. D. Kryter, "The Effects of Noise on Behavior: Feelings of Annoyance,” Journal of Speech and Hearing Dis— orders, Monograph Supplement, 1950, pp. 17-18. 10 that annoyance was discriminable from loudness, and that higher frequency bands were more annoying than the lower frequency bands of equal loudness.l Sataloff states that annoyance was related to loudness, pitch, and modulation of pitch and/or loudness. Loudness-~The most important single factor in determining annoyance-judgments is the intensity of the sound. Pitch-—In general, sounds having their energy concentrated among the higher audible frequencies are more annoying than low-frequency noises. Modulation of Loudness and Pitch-~A third impor- tant factor is the modulation which the sound under- goes. Apparently the changes in loudness are more effective than changes in pitch but the individual differences on this point are2too conSpicuous to permit a safe generalization. Oyer and Hardick reported that when all acoustic characteristics other than frequency are held constant. the Optimum warning signal should be within the range of 700 cps and 4000 cps.3 Criterion of Observer A final area which requires attention is that of the observe reaction. Since he is the judge of the sounds being tested, information should be discussed pertinent to variables that must be taken into account by the experimenter in relation to him. lIbid. —-—.—— 2J. Sataloff, Industrial Deafness (New York: McGraw- Hill Book Company, Inc., p. 47. 3Oyer and Hardick, op. cit., p. 4. 11 In a study for the Armed Services Technical Informa- tion Agency, it was reported that: In order to evaluate the performance of an observer, it is necessary to consider the decision function which this observer is attempting to maximize. In the particular case where the observer is attempting to maximize the expected value of his decision concerning the presence or absence of a signal it was seen that the observer must utilize three types of information. First, he must utilize the information which occurs in the observation interval. He must consider the values and costs associated with his decisions as well as the probability of a signal occurring in the observation interval. It is not possible to interpret meaningfully data reflecting on an observer's ability to utilize information presented in an observation interval with— out considering these additional variables. That is to say, the responses of an observer in a psychophysical experiment must be regarded as responses of the total organism andlnot Simply as outputs of the sensory system under study. In a more recent study, Egan and Clarke emphasize that the decision of an observer may depend on (a) the information content of the stimulus (b) the information available to the observer before the presentation of the stimulus, (c) the properties of the sensory analyzer and (d) motivation vari— ables as they relate to the consequences of each decision. Tanner states that psychoacoustician may be interested in studying the auditory equipment of ob- servers in listening to acoustic signals. He may find that analyses he wishes to perform can best be carried out if he uses signals and noises which are atypical to those of everyday environments. This may permit greater agreement between the 1w. P. Tanner, T. G. Birdsall, and F. R. Clarke, "The Concept of the Ideal Observer in Psychophysics,” AD239022 Contract No. AF 19 (604)-2277, (April, 1960), p. 39. 2J. P. Egan and F. R. Clarke, "PsychOphysics and Sig— nal Detection,‘ AD—291450. Washington, D. C.: Ofice Tech— nical Service, Department of Communication_(l963) as quoted in dsh abstracts (Vol. III no. 3, July, 1963), p. 205. 12 physical conditions of the experiment and the assumptions made to permit conputations. The fact that the physical conditions are atypical with re- gard to everyday environments does not necessarily degrade the quality or the usefulness f the answers to the questions he is asking. It may be seen through this review of the literature that there has not been a great deal of work done in the field of an optimum warning signal. This writer substan- tiates Oyer and Hardick in their statement concerning literature in this vital area. Samples of sounds have been small and in most studies results have been general- ized from the laboratory to outside situations. The liter- ature review pointed up the need for a study focused on evaluation of many auditory signals as related to alerting, with relevant judgments made of their effectiveness by groups of subjects representing the population.2 1W. P. Tanner, Jr. "The Theory of Signal Detecta— bility as an Interpretive Tool for Psychophysical Data," AD239022 Contract No. AF 19(604)—2277, (May, 1960), p. 8. 2Oyer and Hardick, 0p. cit., p. 17. CHAPTER III SUBJECTS, EQUIPMENT, AND TESTING PROCEDURES Subjects The twenty-four subjects (14 male and 10 female) who participated in this study were all students at Michigan State University. They ranged from undergradu— ates to PhD candidates and were between the ages of twenty-two and thirty-four. All subjects were screened before they participated in the experiment and were found to have binaural hearing within normal limits. Normal hearing subjects were defined as having binaural thresholds of 15 dB or better at 500 Cps, 1000 cps, 2000 cps, and 4000 cps.l Equipment Forty—four complex tones(lowest frequency and at least one harmonically related tone half an octave above the lowest tone and not over 4000 cps) were recorded twice on a tape in random order. The equipment at the Michigan State University Speech and Hearing Science Laboratory was utilized both for recording of the sounds 1H. Davis and R. Silverman, Hearing and Deafness (New York: Holt, Rinehart, and Winston,*1nc. 1960), p. 245. 13 14 and for carrying out the experiment. Twelve chairs, each equipped with a set of earphones, were used for both sessions of the experiment. The arrangement of equipment for recording is pictured in Figure 1. The equipment used included three audio oscillators (Hewlett—Packard Model 2026, Hewlett— Packard Model 200A, and B and w Model 200); two tape recorders (Ampex Model 601-2, and Ampex Model 601); a mixer (Ampex Model 35); two reels of tape (3m number 111); level recorder (Bruel and Kjaer Model 2305); a Spectro— meter (Bruel and Kjaer Model 2112); voltmeter (Bruel and Kjaer Model 2409); 6cc coupler (artificial ear, Bruel and Kjaer Model 2203). The electronics technician em— ployed by the Michigan State University Speech Department aided with equipment procedures and supervised the use of equipment during the experiment. Procedure In order to record the complex sounds needed for this experiment three audio oscillators, each set at a different half octave interval in relation to the lowest frequency were employed. If only two frequencies were needed for the complex sound, only two audio oscillators were used. If four or five frequencies made up the complex sound all three audio oscillators were used. The lowest frequency and the next two half octave frequencies were fed Tape . Tape Recorder '“ Mixer Recorder 1 o 5 Audio ' 4 Oscillator 3 F Audio 3 Oscillator 2 1. Audio oscillator (Hewlett-Packard Model 2026) 2. Audio oscillator (Hewlett-Packard Audio Model 200A) Oscillator p// 3. Audio oscillator (B and W Model 200) l 4. Tape recorder Ampex Model 601-2) 5. Tape recorder Ampex Model 601) 6. Four channel stereo mixer (Ampex Model Mx—35) 7. Earphones (Telephonics TDH 39) Tape Line Recorder Mixer 5 6 Amp‘irier 7 Figure 1.--Block diagram of instrumentation utilized for recording audio signals on magnetic tape. 16 into the mixer and recorded on one of the tape recorders. The complex sound was about ten seconds in length to allow some flexibility in playback. This tape recorder was played with the first three frequencies recorded on the tape as the final frequency or two frequencies were put into the mixer from two audio oscillators. This combi- nation yielded the four or five frequency sound which was required. The complex sound was then recorded on the second tape recorder for five seconds. Ten seconds of silence between each sound was allowed in order that the subjects have time to make their judgments. The timing for the five Second sound and the ten second interval was measured by a stop watch. A booklet of eighty—eight re- sponse sheets was given to each subject. Each page repre- sented the ratings across nine scales along a five point continuum for a Specific sound. Pencils were also provided. The complex sounds were recorded in the random order in which they were presented to the subjects. In order to keep the subjects from losing their place, since they were to listen to eighty—eight sounds, each sound was cued by number just before it was heard. In order to obtain an output of approximately equal intensity from the twelve sets of earphones of 70 dB (re: .0002 dyne/cme) at all of the testing frequencies, and ad— justment in the input of the intensities of frequencies into the mixer was necessary in order to fit the frequency 17 curve of the earphones. To do this, eight earphones (four pairs) were randomly selected from the twelve pairs used in the experiment. Each receiver was evaluated using an arti- ficial ear, voltmeter, Spectrometer, and level recorder. The intensity was plotted from the 6cc coupler on a graph level recorder for the frequencies 500 Cps, 1000 cps, 2000 cps, 3000 Cps, and 4000 cps. These figures were averaged for each frequency and plotted on a graph on semilogarithmic graph paper. Other intensities were then interpolated from the curve of best fit. The intensity of each frequency to be measured was analyzed using 90 dB of white noise. The results of four randomly selected pairs of earphones are shown in Table 1. Table 1 INTENSITY CURVE OF FOUR RANDOMLY SELECTED PAIRS OF EARPHONES AT 500 CPS, 1000 CPS, 2000 CPS 3000 CPS AND 4000 CPS Earphone Set 500 1000 2000 3000 4000 Pair 1 phone 1 88 dB 93 dB 93 95 99 Phone 2 87 90 93 99 99 Pair 2 phone 1 88 92 94 100 99 phone 2 88 94 93 96 98 Pair 3 phone 1 87 90 95 99 96 phone 2 86 91 94 99 96 Pair 4 phone 1 85 89 86 88 102 phone 2 87 90 89 95 100 Total 698 729 737 771 789 Average 87 91 92 96 99 Reference Point 90 dB SPL. 18 The graph in Figure 2 illustrates the curve of best fit. To state this curve in terms of actual changes of intensity of the above frequencies, the reference point on the curve of best fit was set arbitrarily at 1000 cps and all other frequencies were corrected in terms of intensity to this reference point. During the experiment the sounds were presented through earphones to the subjects for a period of five seconds per sound with ten seconds between sounds. Each sound was heard twice during the eXperiment, for a total of eighty—eight sounds. The subjects were asked to rate the eighty-eight sounds in terms of nine characteristics along a five point continuum. Each sound was numbered and analyzed by the scale shown in Table 2. The results were interpreted for an alerting signal in terms of rating of pitch and complexity of tones. The subjects were given a copy of the directions so that they could follow them as they were read orally by the experimenter before the experiment began. The experiment lasted forty minutes including one five minute break after sound 44, half way through the test. Decibels 19 100 W 90 .. 500 1000 2000 3000 4000 Frequency in Cycles Per Second Figure 2.-—Curve of best fit for four pairs of ear- phones. The horizontal line represents intensity expressed in decibels. The vertical line represents frequency expressed in cycles per second. 20 .:m .Q ..pHo .mo axoapnmm new nozo Hm moaenanoeomsoeo smaszmo noaenanoeoosaco gammao noaenanoeonsoeo oZHQBmaemzoz noaenanoeonnoco ozH9022m£ UCSOm wasp mmoQ Bzmmmb m>m£ Ccsom mafia mmoQ UZHQBmw£ Ccsom mHCp mooQ UZHwOZZ< ®>m£ UCSOm mdflw m®OQ UZHZE<3 m>m£ pcsom maflp mooQ UZHEmmH< ®>m£ UCSOm mHEP mmOQ BZMBmHmZH m>m£ Eczom mafip moom UZHZmm£ ocsom many moom Bzm£ Dcsom was» moom .mamcwam oaosm Mo moapmasopomswno mafipnmaw mafiumn pom mcoammoEHc pcm>oaon ocHEnopop on oomaaaus poozm oncommoh wo meeomnu.m oHQmB CHAPTER IV RESULTS AND ANALYSIS Introduction As indicated in Chapter I, a significant difference in subjects rating of the alerting potential of the lowest frequency and tone complexity would seem to indicate two characteristics that are important in determining an optimum warning system. This study was concerned with determining whether or not a difference exists in terms of change in frequency and change in complexity. The following statis— tical procedures were therefore employed to determine the Significance of the differences and the reliability of the subject ratings of those who participated in this experi- ment. Both of the null hypotheses listed in Chapter I were tested to determine whether or not they should be rejected. Analyses In order to analyze the results of this experiment each of the five Spaces along the continuum was assigned a number. These were numbered from left to right. This would mean that those sounds which achieved low scores by subject judgment would be considered to have characteristics which would lend them to be acceptable for a warning system. '21 22 Those sounds which were rated high by subject judgment were rejected as sounds for an alerting signal. Since each sound was heard by the subjects twice, a reliability estimate could be obtained by averaging the score over the nine characteristics for each sound the first time it was heard by the twenty—four subjects and comparing it with the average score the second time each subject heard it. Because this study was basically dealing with the characteristics of an alerting Signal, it was decided to check the reliability of only the ten sounds that had the lowest scores (most acceptable for an alerting signal) and only the ten highest scores (least acceptable for an alerting signal) as representative of the overall reliability. To determine the reliability the Pearson Product— Moment correlation was employed. It measured, in this case, strength of relationship of the nine characteristics of a sound heard two different times during the experiment. This correlation measures the amount of Spread about the linear least-squares equation. The explanation and the formula was taken directly from Blalock's book, Social Statistics.l lHubert M. Blalock, Social Statistics éNew York: McGraw-Hill Book Company, Inc., 1960), p. 28 . 23 Reliability had a wide range in this study. It ranged from —.0005 to .83 with an average reliability of .53. One reason for the low reliability may have been that only ten seconds were allowed between sounds to judge the sound on nine scales. A second reason may have been that there were no practice sounds given to the subjects. As a result the subjects may not have been able to analyze the first sounds because they lacked internal criteria in comparing the sound to another sound. It is felt, however, that within the range of reliability certain conclusions may be reached. The first hypothesis subjected to a statistical analy- sis was: no statistically significant differences exist between the means of subjects rating of the alerting poten— tial of harmonic tones of two, three, four, and five components, as measured on a nine point scale along a five point continuum. To do this a one way analysis of variance was utilized as disucssed in Blalock's book, Social Sta- tistics.1 From this analysis, Table 3 evolved. Required at the .01 level of significance for 3 and 92 degrees of freedom (df) is 4.04. Since 86.65 is larger than 4.04, the null hypothesis is rejected. This analysis is graphically represented in Figure 3. 1 Ibid., p. 250. 24 TABLE 3 COMPUTATIONS FOR ANALYSIS OF VARIANCE OF TONE COMPLEXITY Sums of Degrees of Estimate of Squares Freedom Variance F* Total 59.78 N-l=95 Between 44.18 k—l= 3 14.73 86.65 Within 15.60 N—k=92 .17 *4.04 F is significant. 25 UL) l tinuum of two, three, four, five component tones. to Average ratings (of subjects of alerting potential) on a five point con- 1 1 I l I 2 3 4 5 Number of soundsin.the complex sound Figure 3.——The average of (subjects rating of alerting poten— tial) two, three, four, and five component sounds across twenty—four subjects. 26 Although an analysis of variance shows there is a statistically significant difference, it does not indicate where the difference or differences are. Therefore, a :7 test was utilized to note where the significant differences occurred in terms ofsnnuklcomplexity. The formula for the tftest was taken directly from Lindquist's book, Design and Analysis of Experiments in Psychology and Education.1 In order to use this formula discussed by Lindquist, an average of subjects rating of the alerting potential of two component sounds across all twenty-four subjects was made. The same procedure was followed for all three com— ponent sounds, all four component sounds and all five com— ponent sounds. The value for p_was found in the pftable for 92 degrees of freedom at the .01 level of signifi- cance for a two—tailed test. This value (d) was then compared with the differences between the averages of the columns. If the difference was greater than the d value it was concluded that there was a significant difference. Table 4 shows a comparison between complex tones. Two, three, and four component sounds are listed horizontally and are compared with three, four, and five component sounds listed across the t0p. Significant differences are shown in all cases except for between the complex 1E. F. Lindquist, Design and Analysis of Experiments in Psychology and Education (Boston: Houghton Mifflin 00., 1956), p. 93. 27 TABLE 4 COMPARISONS OF MEANS OF SOUND COMPLEXITIES MEASURED BY SUBJECT JUDGMENT Number of Number of Components Components 3 4 5 2 .36 1.42 1.60 3 1.06 1.24 .18 28 sounds of four and five component sounds. Graphically the downward trend can still be noted, but it has leveled off to a great extent. The second hypothesis tested was: there are no statistically significant differences between the means of subjects rating of the alerting potential of different frequencies of harmonic sounds as measured on a nine point scale along a five point continuum. The same design for one way analysis of variance was utilized.1 The resulting data are summarized in Table 5, and graphically represented in Figure 4. TABLE 5 COMPUTATION FOR ANALYSIS OF VARIANCE OF LOWEST FREQUENCY COMPONENT Sums of Degrees of Estimate of Squares Freedom Variance F* Total 187.08 N-l = 455 Between 30.71 k-l = 18 1.71 4.78 Within 156.37 N—k = 437 .358 *1.99 is significant. Required at the .01 level of significance for 18 and 455 degrees of freedom (df) is 1.99. Since the F value of 4.78 is greater than 1.99, the null hypothesis is rejected. 1Blalock, op. cit., p. 250. .Edanducoo wagon m>au a so .muomnnsm hsomnhucosu muchom cowmhm>mi tampon wcauhoaa on» no mcdumn uncannun «0V meadow uneconaoo 0>HH annom .moncu .ozaus.: madman vacuum Ava mmaoho ca hocosaonm comm comm owed ommm scam cmom.omma cow omma coma owma owed owns cows ooHH coca ova new ops b - d W 1 u A u u a wagon acocoaaoo 05m / condom 20:09:00 each I o /\ mocaom acmconaoo ounce.l!llll. monsom acocoqaoo 039 29 \ \O» ... 2. r d C . . . . . . . . . . . . H spunos iuauodmoo any; pus ‘Jnog ‘aaaui ‘oMi JO mnnutquoo iutod SAT; a no Sutiaa afieaanv 3O Since the one way analysis of variance did not show where the difference or differences occur, the pftest was again utilized. Table 6 indicates the total results. Some significant differences occur throughout the table but seem to follow no logical pattern except between the frequencies of 1800 cps and 2200 cps, where they Seem significantly different with other frequencies, but not among themselves. Discussion of Results The statistical analyses performed indicate that the first hypothesis could be rejected. The pfscores computed to test the mean differences for the four different com- plexities, showed that the greater the complexity of the sound, the better it seemed to be for an alerting signal. Although there was no significant difference between com- plex sounds made up of four or five components, a notice- able trend can be noted for the more complex sounds to be judged by the subjects as having greater alerting potential. The statistical analysis for the second hypothesis indicates that it too may be rejected. A particular range of frequencies was shown by the pftest of subjects rating of alerting potential to be significantly more alerting than those above or below the range. The significant range was found to be 1800 cps to 2200 cps. It would appear on 31 .Ho>oH Ho. ecu um unmonchHm mH ms. wchoooxm msHm> v mcnot omit any. APPENDIX B 43 LIST OF INDIVIDUAL TONES COMPRISING THE FORTY-FOUR COMPLEX SOUNDS Sound Number Sound Components in cycles per second 1 700, 1050 2 800, 1200 3 900, 1300 4 1000, 1500 5 1100, 1650 6 1200, 1800 7 1300. 1950 8 1400, 2100 9 1500, 2250 10 1600, 2400 11 1700, 2550 12 1800, 2700 13 1900, 2850 14 2000, 3000 15 2100, 3150 16 2200. 3300 17 2400, 3600 18 2500, 3750 19 2600, 3900 20 700, 1050, 1400 21 800, 1200, 1600 22 900, 1350, 1800 23 1000, 1500, 2000 24 1100, 1650, 2200 25 1200, 1800, 2400 26 1300, 1950, 2600 27 1400, 2100, 2800 28 1500, 2250, 3000 29 1600, 2400, 3200 30 1700, 2550, 3400 31 1800, 2700, 3600 32 1900, 2850, 3800 33 2000, 3000, 4000 34 700, 1050, 1400, 2100 35 800, 1200, 1600, 2400 36 900, 1350, 1800, 2700 44 45 LIST OF INDIVIDUAL TONES COMPRISING THE FORTY- FOUR COMPLEX SOUNDS-—Continued Sound Number Sound Components in cycles per second 37 1000, 1500, 2000, 3000 38 1100, 1650, 2200. 3300 39 1200, 1800, 2400, 3600 40 1300, 1950, 2600, 3900 41 700, 1050, 1400, 2100, 2800 42 800, 1200, 1600, 2400, 3600 43 900, 1350, 1800, 2700, 3600 44 1000, 1500, 2000, 3000, 4000 APPENDIX C 46 47 ss.m mw.m mo.m mo.m no.3 :s.m HH.m 6m.m H6.m oo.m 6H.m mo.m so 66.: 66.: 6H.e 6H.s mm.s oH.s oo.m oo.s er.s om.s 6m.: 66.: mm mm.H os.m mm.m mm.m Ne.m sm.H e~.m mm.m s:.m HH.m He.m mo.m mm se.m oo.m oo.m em.m oo.m om.m o6.m oo.m oo.m sm.m om.m om.m Hm mm.m NN.: 6m.: mm.m O6. mm.m 06.: mo.e om.s NN.: mm.: He.s om 66.m ::.m 6m. m:.m oH.: mo.m HH.s 6m.s o6.m 66.m mm.m sm.m mH oo.m sm.m oo.m sm.m oo.m :m.m oo.m em.m oo.m oo.m sm.m oo.m 6H H6.m Hs.m NN.: oH.: sm.e He.m 6m.m NN.: sm.m 6o.: mm.s mm.m AH sm.m mw.m o.m ms.m mm.m rm.m wm.m oo.m 66. Se.m sm.m mo.m 6H mo.m Hm.m mm.m om.m mo.m oo.m om.m sm.m sm.m om.m mm.m 6m.m 6H m:.m em.m em.m oH.m oH.m 6H.m 66. 66.m em.m mm.m o6.m 6H.m eH mm.m mo.m mo.s oH.m mw.s 6m.: mo.s mm.s se.m o6.s mw.m NN.: mH o6.m ss.m o6.m ewm.m He.m o6.m O6.m mm.m mm.m mm.m mm.m mm.m mH sm.m HH.m mo.: 6H.m o6.e em.m mm.: em.m sm.m NN.: sm.m 6m.s HH :m.m om.m mm.m oo.s me.m mm.m as.e oo.m oo.m 66.m 6:.N 6H.m oH oH.m mm.m mm.m mm.m oH.m mm.m mm.m HH.: mm.m mm.m s:.m ss.m m oH.m mm.m He.m mm.m oo.: Hm.s 6m.s mm.e sm.m 66.: no.3 6m.: 6 mm.m sm.m mm.s mm.m m6.m mm.m HH.s mm.e oH.s mm.m es.m oH.m s mm.m es.m se.m sH.m 66.: m:.m mm.m 6H.: mm.m 66.: ee.m oo.: 6 mo.m es.m em.H oH.m HH.m Hs.m O6.m ss.m mm.m HH.m HH.m se.m m oH.m os.m mm.m mm.m sm.m mm.m mm.m ss.m He. or.: ::.e sm.m s sm.m mm.m :m.m mw.m ss.m 6m.m 66.m oo.m me.m mm.m sm.m oH.m m 66.m ::.m oo.m em.m os.m oH.m 6H.m mm.m mm.m 6H.m HH.m :s.m m 6m.m o6.m mm.m es.m HH.: oH.: mm.m mm.m mm.m sm.m oo.s NN.: H mH HH oH m 6 e 6 6 s m m H LOQESZ Ocsom muOOmnsm .HOOmQSm an Ocsom some now mOHmom Compano Ho>o Oweno>m .wpma 36m 48 66.6 e6. 6 6s.6 6o.6 6o.6 o6.6 66. 6 He.6 66.6 3o.6 6H.6 6o.6 36 66.6 66. 6 66.6 36.6 66.H 36.6 O6. 3 H6.6 oo.3 66.6 66.6 66.6 66 66.6 66. 6 6e.6 66.6 e6.H 66.6 36.6 66.6 66.6 33.6 66.H 63.6 66 oo.6 66.6 33.6 HH.6 66.H 66.6 oo.6 66.6 oo.6 oo.6 o6.6 36.6 H6 66.6 66.6 s6.3 33.3 oH.6 6H.3 66.3 s6.6 6o.3 6o.3 63.6 36.6 o6 66.6 66.6 e6.3 oH.6 33.H 66.6 oH.3 66.6 oH.6 66.6 36.6 66.6 6H 66.6 oo.6 oo.6 oo.6 33.6 oo.6 66.6 36.6 oo.6 36.6 66.6 66.6 6H 66.6 6H.6 oo.6 66.6 oo.H H6.3 66.3 66.3 66.3 66.6 66.6 66.6 6H 66.6 6o.6 66.6 66.6 66.6 6o.6 es.6 66.6 66.6 66.6 oo.6 63.6 6H oo.6 66.6 66.6 oo.6 HH.6 O6.6 o6.6 66.6 oo.6 36.6 oo.6 66.6 6H 00.6 06.6 66.6 33. 6 36.6 66.H 66. 6 66.6 66.6 66.6 ee.6 63.6 3H 66.3 66.6 66.3 66. 3 33.6 H6.6 o6. 6 66.6 66.6 63.6 33.6 H6.6 6H 66.6 6o.6 66.6 66. 6 66.6 O6.6 o6. 6 63.6 66.6 63.6 66.6 66.6 6H 66.6 66.6 oo.3 oo.3 66. H 66.6 66.6 oH.6 66.6 66.6 6e.6 6H.6 HH oo.3 oH.6 66.6 o6.6 6o. H 6H.6 oo.3 66.6 o6.6 66.6 6H.3 33.6 oH 66.6 666. 66.6 66.6 66. 6 66.6 66.6 63.6 66.6 33.6 H6.6 33.6 6 oo.3 ee.6 66.3 66.3 oo.H 66.6 H6.6 66.6 33.6 63.6 es.6 66.6 6 56.3 66. 6 H6.3 33.6 36.H 66.6 6e.6 66.6 66.6 66.6 66.6 66.6 6 66.6 36. 6 oo.3 66. 3 66.H 66.6 766.6 66.6 66.6 66.6 66.6 oH.6 6 33.6 666 66.6 66. 6 e6.H 66.H HH.6 66.6 66.H 66.H 66.6 66.H 6 66.H 33.6 66.6 up. 6 ee.H 63.6 6o.3 66.6 66.6 66.6 o6.6 66.6 3 66.6 66.6 H6.6 66.6 66.6 66.6 33. 6 e6.6 6o.6 6e.6 66.6 6o.6 6 oo.6 6666 H6.6 33.6 oo.6 63.6 66. 6 56.6 66.6 es.6 Ne.6 66.6 6 66.H 66.6 oH.6 36.6 6H.H 66.6 63. 6 63.6 66.6 66.6 66.6 66.6 H 36 66 66 H6 o6 6H 6H NH 6H 6H 3H 6H 969852 Ocsom mpOprsm .ooSQHpcoounpoompsm 6p 6:506 £066 906 mOH606 Coopano no>o owmno>6 .mpmm 36m 49 66.H 66.6 36.6 H6.6 o6.6 6H.6 66.6 66.6 66.6 H6.6 H6.6 66.6 36 oo.H 33.6 o6.6 66.6 o3.6 66.6 HH.6 66. 6 66.3 oo.6 o6.6 36.6 66 6H.H 66.H 66.6 66.6 66.6 60.6 66.6 66. H 66.6 66.6 66.6 66.H 66 O6.H o6.H 6H.6 o6.6 66.6 oo.6 06.6 o6. 6 66.6 66.6 66.6 66.6 H6 66.H 66.H o6.6 66.6 66.3 oo. 3 66.3 66.6 66.6 66.6 6H.3 o6.6 O6 66.H 66.6 O6.6 66.6 6o.6 63. 6 66.6 O6.6 o6.6 66. 6 66.6 66.6 6H 33.6 66.6 66. 6 oo.6 oo.6 oo. 6 oo.6 oo.6 oo.6 36. 6 oo.6 oo.6 6H oo.H 66.H 66. 6 66.6 66.6 H6.6 6H.3 66.6 66.6 6H. 6 66.6 33.6 6H oo.6 66.6 66. 6 6o.6 oo.6 oo.6 oo.6 oo.6 66.6 66.6 HH.6 o6.6 6H 33.H 66.H 66.6 o6.6 oo.6 oo.6 6H.6 36.6 H6.6 66.6 o6.6 66.6 6H 66.H 66.6 66.6 o6.6 66.6 66.6 36.6 66.6 66.6 66.6 66.6 33.6 3H 6H.H 33.6 6H.6 36.6 66.6 63.H 66.6 66.6 66.6 66.6 oH.6 66.6 6H 33. 6 66.6 66.6 o6.6 o6.6 66.6 o6.6 H6.6 oH.3 33.6 66.6 o6.6 6H 6H. 6 66.6 oH.6 33.6 66.6 63.6 66.6 33.6 6o.6 66.6 oH.6 33.6 HH oo. H o6.H 66.6 66.6 oH.6 66.6 o6.6 33.6 oH.6 o.6 66.6 o6.6 oH 36.H 66.6 H6.6 66.6 HH.6 oH.6 oH.6 66.6 66.6 o6.6 oo.6 H6.6 6 oo.H 66.6 66.6 66.6 66.6 o3.6 36.6 66.6 33.6 HH.6 oo.6 36.6 6 66.H 66.H o6.6 66.6 66. 6 33.6 HH.6 66. 6 66.6 66.3 H6.3 66.6 6 oo.6 6o.6 63. 6 6H.6 66. 6 o6.6 66.6 66. 6 66. 6 66.6 6H.6 o6.6 6 oo.H 66.H 66. 6 66.H 6T H 66.H 66.H 63. 6 36. H oH.6 66.H 36.H 6 66.H HH.6 oT 6 6o.6 66.6 66.6 66.6 H6.6 66. 6 66.6 66.6 H6.6 3 66.H 66.H 66.6 H6.6 66.6 66.6 66.6 66.6 H6.6 6H.6 63.6 36.6 6 O6.H H6.H oo.6 66.6 oo.6 66.6 66.6 o6.6 6o.6 6o.6 oo.6 oo.6 6 66.H 66.6 60.6 66.6 H6.6 66.6 66.6 H6.6 66.6 66.6 66.6 66.6 H 66 66 36 66 66 H6 o6 66 66 66 66 66 662852 Unsom mHOOwnzm .coschcoonupoonpsm 69 @2506 some now monom coopano Ho>o owmno>6 .6669 36m 50 6H.6 66.6 36.H 66.6 66.6 63.6 66.6 H6.H 36 o6.H o6.H oo.H 66.H 66.H 66.H oo.H oo.H 66 66.H 66.H HH.H 66.H HH.H 66.H HH.H 66.H 66 o6.H oo.6 oo.H o6.H HH.H 66.H oo.H oo.H H6 66.H 66.H 6o.H 66.H 66.H 66.H 66.H 63.H o6 66.H 33.H 66.H 66.6 63.H 6H.6 o6.H o6.H 6H 66.6 33.6 66.6 H6.6 33. 6 66.6 33.6 33. 6 6H 6o.H 66.H o6.H 66.H H6. H 66.6 oo.H oo. H 6H 66.6 6o.6 HH.6 66.6 66. 6 oo.6 66.6 6o. 6 6H 6o.6 oo.H oo.H o6.H o6. H 6H.6 6o.6 66. H 6H 66.H 36.H 6o.6 3T H 66. 6 H6.H 66. H 6o. 6 3H 66.H 66.H 36.H 36. H H6.6 66.6 66. H 6T H 6H 66.6 oH.6 33.6 66.6 o6.6 66.6 6T 6 66.6 6H H6.6 HH.6 oH.6 66.6 63.H H6.H 66.H 63.H HH oo.H o6.H o6.H oo.6 66.H o3.H oo.6 oo.H oH 66.H oH.6 66.H 63.6 66.6 66.6 6o.6 HH.6 6 oo.H oo.H oo.H 66.6 oo.H 33. H .oo.H oo.H 6 33.H 66.H .HH.H oH.6 oo.H 66. H 66.H 66.H 6 o6.H oo.6 66.H 66.6 oo.6 66. H oo.H 66.H 6 6o.H 63.H HH.H 66.H oo.H 66.H oo.H HH.H 6 HH.H 66.H 66.H HH.H HH.H 66.H 66.H HH.H 3 6o.6 66.H oo.H 66.H 66.H 66.H 66.H 66.H 6 36.H o6.H oo.6 33.H O6.H oo.6 oo.6 o6.H 6 oo.H 33.H 6o.H 66.H H6.H 66.H HH H 66.H H 33 63 63 H3 o3 66 66 66 669832 Unsom mpoomnsm .UOSQHHGOOnnuoonQSm 69 venom £066 606 monom CompcwHo so>o owmno>6 .muma 36m ROOM USE ONLY ”'CII'IIIIIIIII[IIIIIIJIIIMIQIII