EFFECT? Oi? EXPERSMEN'FER BFA§ UPQN
WEE-TONE ANS} SPEECH
AwfiEQMfimiC TE§T RE§ULT$

Thesis {3cm {he Degree 01" m“; D.
{\‘i‘zCE-‘EEGAR‘ STATE UKEVEQSET

Nichoias M. His-skim};
£968

[Hints

 

n

L

L I B R A R Y 3
Nlichzgan State 3
University 1

___ (

 

This is to certify that the

thesis entitled

EFFECTS OF EXP?" t{IMWT

,3 BIAS UPON
FUEL-T0133 AND SPEECH

TEST RES? LTS

presented by
Nicholas M. Hipskind

has been accepted towards fulfillment
of the requirements for

Mdegree in and Speech
Sciences

am‘bwm

Major p{ofessor

DateflV' 7/. /952

0-169

 

 

 

,fi' ‘.-0-

.__.-

M
e57

918%

 

 

LP ': if.

 

2
L .
-

Lé—of-Lem,

.tlfi‘.w ; '
" 111011;” J. 1.1
‘ I

..-i
L

C. "J.
‘l‘ .
-v c

A
l_l. '
...‘r .
T ‘._. l-
. . -\'- —.

 

 

 

 

 

 

 

ABSTRACT

EFFECTS OF EXPERIMENTER BIAS UPON PURE-TONE
AND SPEECH AUDIOMETRIC TEST RESULTS

by Nicholas M. Hipskind

The purpose of this investigation was to determine
whether certain types of previous audiometric information
causes the tester to influence pure-tone and Speech audio-
metric results. Specifically, this study was concerned
with the possibility of testers' expectancy effects on
audiometric responses obtained from normal hearing adults,
hard-of—hearing adults having either conductive or sen—
sorineural hearing impairments, and normal hearing child-
ren having either normal or defective articulation.

Two groups of testers were used that varied con-
siderably in terms of amount of training in clinical audi—
ology skills. One of the groups consisted of four pre-
doctoral students majoring in audiology at Michigan State
University and the other group was composed of four
undergraduate students who were majoring in audiology and
speech sciences at Michigan State University.

The study was composed of four experiments that

were designed to study the phenomena of experimenter

 

-;.—-‘

 

Nicholas M. Hipskind

(tester) bias and effect as related to pure-tone and
speech reception threshold measurement and speech dis—
crimination. The variables were as follows: age of
subjects, audiometric tests employed, articulation of
subjects, status of subjects' hearing mechanism, sophisti—
cation of testers, and types of previous audiometric

information.

Experiment I

In this experiment four sophisticated clinical
audiologists were employed as testers to study experi-
menter bias and effect upon pure-tone air—conduction
thresholds and speech reception thresholds of eight

normal hearing adults.

Experiment II
In this experiment the same four testers were
utilized to explore the bias and effect phenomena as re-
lated to pure-tone air and bone-conduction thresholds and
speech audiometry scores (speech reception thresholds and
speech discrimination scores) of eight adults with sensor—
ineural hearing disorders and eight adults with conductive

hearing impairments.

 

— ~ -‘M'

 

 

Nicholas M. Hipskind

Experiment III

In this experiment the same four testers were
used to investigate the bias and effect phenomena as re-
lated to pure-tone air-conduction thresholds and speech
audiometry scores (speech reception thresholds and speech
discrimination scores) of sixteen normal hearing children.
Half of the children had defective articulation and the

remaining eight had normal articulation.

Experiment IV

In this experiment four unsophisticated testers
were used to study experimenter bias and effect as related
to pure-tone air-and bone-conduction thresholds of sixteen
hard-of—hearing adults. Half of these subjects had a
sensorineural hearing loss and half exhibited a conductive

hearing disorder.

The clinical and statistical results of this study
indicated that there were no significant differences be-
tween audiometric scores obtained by testers as a result
of having either correct or erroneous previous test in-
formation. The results showed that each tester under
each experimental condition (actual test results, better
than actual test results, poorer than actual test results,
and no test results) obtained audiometric results that

were approximately the same. The results also indicated

 

 

Nicholas M. Hipskind

    
  
   

.’:- ;§previous audiometric information was no more
”fligfulential in eliciting an observable bias for speech
Ajgd§hdiometry than for pure-tone audiometry. Finally, the
I lgmount of sophistication of testers did not significant1y~

. {affect the test results.

 

 

V A.» 5
. . “mm--.“

EFFECTS OF EXEERIMENTER BIAS UPON PURE-TONE
AND SPEECH AUDIOMETRIC TEST RESULTS

BY
w

L

Nicholas M§3Hipskind

A THESIS

Submitted to
Michigan State University
in partial fulfillment of the requirements
for the degree of

DOCTOR OF PHILOSOPHY
Department of Audiology and Speech Sciences

1968

 

 

 

 

 

ACKNOWLEDGMENTS

This research was supported in part by a fellowship

  
 
 
  
  
  
  

from the Vbcational Rehabilitation Administration.
Special gratitude is given to Dr. William F.
Rintelmann, Associate Professor of Audiology and Speech
Sciences, for the interest he displayed and the guidance
he gave as thesis advisor.

A very special thanks to my wife, Sandy, and
daughter, Heidi, for their incomparable understanding,

encouragement, and faith during the past three years.

ii

 

__ —--—fi_

 

 

r.,r_

 

TABLE OF CONTENTS
Page
ACKNOWLEDGMENTS . . . . . . . . . . . . . . . . . . 11
LIST OF TABLES . . . . . . . . . . . . . . . . . . V
LIST OF FIGURES . . . . . . . . . . . . . . . . . . ix
Chapter
I. INTRODUCTION . . . . . . . . . . . . . . . 1
Purpose of Study . . . . . . . . . . . . 1
Definition of Terms . . . . . . . . . . . 8
II. REVIEW OF THE LITERATURE . . . . . . . . . 10
Threshold Measurements . . . . . . . . . 10
Reliability of Audiometric Tests . . . . l4
Phenomena of Experimenter Bias and
Effect I I I C I O I I I O I I I I I I I 24
III. PROCEDURES I I I I I I I I I I I I I O I I 45
Overview of Experiments . . . . . . . . . 45
Experimental Conditions . . . . . . . . . 47
Testers I I I O I I I I I O I I I I I O I 50
Subjects . . . . . . . . . . . . . . . 52
Experimental Design . . . . . . . . . . . 52
Equipment . . . . . . . . . . . 56
Recorded Speech Materials . . . . . . . . 61
Test Procedures . . . . . . . . . . . . . 64
Summary of Experiments . . . . . . . . . 73
IV. RESULTS AND DISCUSSION . . . . . . . . . . 75
Descriptive and Statistical Results of
Experiment I . . . . . . . . . . 75
Descriptive and Statistical Results of
Experiment II . . . . . . . . . . . 37
Descriptive and Statistical Results of
Experiment III . . . . . . . . . 107
Descriptive and Statistical Results of
Experiment IV . . . . . . . . . . . . 122
iii

 

b. Page
General Discussion 0 a") I o o I 0’. c o o 138
V. SUMMARY AND CONCLUSION, . . . . . . . . , . g '. 141
summBIY' o o o o a o u o o u o a o o o o o 141

Conclusions . . . . . . . . . . . . . . . 145
Recommendations for Further Research . . . 149

     

BIBLIOGRAPHY I ' I I I I I I ’ I I I I I I I I I I I I I 150

APPENDICES o'} o a o o o o o o o o a c o o‘o o o n a 156

 

.2 iv

’
’ -:
.»« “t
“h J
s,r_“ 7'7 -
V I

 

Table

II.

III.

IV.

VI.

LIST OF TABLES

Speech Audiometric Results with Respect to
the Conditions of Actual, Better, and
Poorer Test Results . . . . . . . . . . . .

Means (M), Standard Deviations (SD),
Standard Errors of Measurement (SEm), and
Ranges of Pure-Tone Air—Conduction
Thresholds Obtained for Eight Normal
Hearing Adults by the Investigator and
Four Pre-Doctoral Students Under Four
Conditions . . . . . . . . . . . . . . . .

Means (M), Standard Deviations (SD),
Standard Errors of Measurement (SEm), and
Ranges of Speech Reception Threshold Ob-
tained for Eight Normal Hearing Adults by
the Investigator and Four Pre-Doctoral
Students Under Four Conditions . . . . . .

Means (M), Standard Deviations (SD),
Standard Errors of Measurement (SEm), and
Ranges of Pure—Tone Air—Conduction
Thresholds Obtained for Eight Sensori-
neural (SN) Adult Subjects and Eight Con-
ductive (Cond.) Adult Subjects by the In-
vestigator and Four Pre—Doctoral Students
Under Four Conditions . . . . . . . . . . .

Means (M), Standard Deviations (SD),
Standard Errors of Measurement (SEm), and
Ranges of Pure—Tone Bone-Conduction
Thresholds Obtained for Eight Sensorineural
(SN) Adult Subjects and Eight Conductive
(Cond.) Adult Subjects by the Investigator
and Four Pre-Doctoral Students Under Four
Conditions . . . . . . . . . . . . . . . .

Means (M), Standard Deviations (SD), Stan-
dard Errors of Measurement (SEm), and
Ranges of Speech Reception Thresholds Ob-
tained for Eight Sensorineural (SN) Adult
Subjects and Eight Conductive (Cond.) Adult
Subjects by the Investigator and Four Pre-
Doctoral Students Under Four Conditions . .

V

Page

50

77

80

88

92

96

 

VIII.

Ix.

XI.

XII.

Means (M), Standard Deviations (SD),
Standard Errors of Measurement (SEm), and
Ranges of Speech Discrimination Scores Ob-
tained for Eight Sensorineural (SN) Adult
Subjects and Eight Conductive (Cond.) Adult
Subjects by the Investigator and Four Pre-
Doctoral Students Under Four Conditions . .

Means (M), Standard Deviations (SD),
Standard Errors of Measurement (SEm), and
Ranges of Pure-Tone Air-Conduction
Thresholds Obtained for Sixteen Normal
Hearing Children——Eight with Normal Articu—
lation (N.A.) and Eight with Defective
Articulation (D.A.)--by the Investigator
and Four Pre-Doctoral Students Under Four
Conditions . . . . . . . . . . . . . . . .

Means (M), Standard Deviations (SD),
Standard Errors of Measurement (SEm), and
Ranges of Speech Reception Thresholds for
Sixteen Normal Hearing Children-—Eight with
Normal Articulation (N.A.) and Eight with
Defective Articulation (D.A.)--by the In-
vestigator and Four Pre-Doctoral Students
Under Four Conditions . . . . . . . . . . .

Means (M), Standard Deviations (SD),
Standard Errors of Measurement (SEm), and
Ranges of Speech Discrimination Scores for
Sixteen Normal Hearing Children——Eight with
Normal Articulation (N.A.) and Eight with
Defective Articulation (D.A.)--by the In-
vestigator and Four Pre-Doctoral Students
Under Four Conditions . . . . . . . . . . .

Means (M), Standard Deviations (SD),
Standard Errors of Measurement (SEm), and
Ranges of Pure-Tone Air-Conduction
Thresholds Obtained for Eight Sensori-
neural Adult Subjects by the Investigator
and Four Undergraduate Students Under
Four Conditions . . . . . . . . . . . . . .

Means (M), Standard Deviations (SD),
Standard Errors of Measurement (SEm), and
Ranges of Pure-Tone Air-Conduction
Thresholds Obtained for Eight Conductive
Adult Subjects by the Investigator and
Four Undergraduate Students Under Four
Conditions . . . . . . . . . . . . . . . .

vi

Page

97

109

112

114

123

125

Table Page

XIII. Means (M), Standard Deviations (SD),
Standard Errors of Measurement (SEm), and
Ranges of Pure-Tone Bone-Conduction
Thresholds Obtained for Eight Sensori—
neural Adult Subjects by the Investigator
and Four Undergraduate Students Under
Four Conditions . . . . . . . . . . . . . . 129

XIV. Means (M), Standard Deviations (SD),
Standard Errors of Measurement (SEm), and
Ranges of Pure-Tone Bone-Conduction
Thresholds Obtained for Eight Conductive
Adult Subjects by the Investigator and
Four Undergraduate Students Under Four
Conditions . . . . . . . . . . . . . . . . 130

XV. Summary of Analysis of Variance Comparing
Differences of Testers (A), Conditions
(B), and Frequency (C) for Eight Normal
Hearing Adults by Pure—Tone Air-Conduc-
tion Audiometry . . . . . . . . . . . . . . 157

XVI. Summary of Analysis of Variance Comparing
Differences of Testers (A) and Conditions
(B) for Eight Normal Hearing Adults for
Speech Reception Thresholds . . . . . . . . 153

XVII. Summary of Analysis of Variance Comparing
Differences of Testers (A), Conditions
(B), Pathologies (C), and Frequency (D)
for Sixteen Hart-of-Hearing Adults by
Pure-Tone Air Conduction Audiometry . . . . 159

XVIII. Summary of Analysis of Variance Comparing
Differences of Testers (A), Conditions
(B), Pathologies (C), and Frequency (D)
for Sixteen Hard-of-Hearing Adults by
Pure-Tone Bone-Conduction Audiometry . . . 160

XIX. Summary of Analysis of Variance Comparing
Differences of Testers (A), Conditions
(B), and Pathologies (C) for Sixteen Hard-
of-Hearing Adults for Speech Reception
Thresholds . . . . . . . . . . . . . . . . 161

XX. Summary of Analysis of Variance Comparing
Differences of Testers (A), Conditions
(B), and Pathologies (C) for Sixteen Hard—
of—Hearing Adults for Speech Discrimina—
tion Scores . . . . . . . . . . . . . . . . 162

Table

XXI.

XXIII.

XXIV.

XXV.

Summary of Analysis of Variance Comparing
Differences of Testers (A), Conditions
(B), Articulation (C), and Frequency (D)
for Sixteen Normal Hearing Children by
Pure-Tone Air—Conduction Audiometry . . .

Summary of Analysis of Variance Comparing
Differences of Testers (A), Conditions
(B), and Articulation (C) for Sixteen
Normal Hearing Children for Speech
Reception Thresholds . . . . . . . . . .

Summary of Analysis of Variance Comparing
Differences of Tester (A), Conditions
(B), and Articulation (C) for Sixteen
Normal Hearing Children for Speech Dis-
crimination Scores . . . . . . . . . . .

Summary of Analysis of Variance Comparing
Differences of Testers (A), Conditions
(B), Frequency (C), and Pathologies (D)
for Sixteen Hard—of-Hearing Adults by
Pure-Tone Air-Conduction Audiometry . . .

Summary of Analysis of Variance Comparing
Differences of Testers (A), Conditions
(B), Frequency (C), and Pathologies (D)
for Sixteen Hard—of—Hearing Adults by
Pure-Tone Bone-Conduction Audiometry . .

viii

Page

163

164

165

166

 

LIST OF FIGURES
Figure Page

1. Illustrative Audiogram Containing the
Conditions of (A) Actual Test Results,
(B) Better Test Results, and (P) Poorer
Test Results for the Right Ear Air-

Conduction Thresholds for One Subject . . . 50
2. Block Diagram of the Audiometric Equip-

ment Used to Make Audiometric Assess-

ments in This Study . . . . . . . . . . . . 59

3. Mean Air—Conduction Thresholds for
Eight Normal Hearing Adults Obtained
by the Investigator and Four Pre-Doctoral
Students Under Four Conditions . . . . . . 79

4. Mean Air—Conduction Thresholds for Eight
Adults with a Sensorineural Hearing Dis-
order Obtained by the Investigator and
Four Pre—Doctoral Students Under Four
Conditions . . . . . . . . . . . . . . . . 90

5. Mean Air—Conduction Thresholds for Eight
Adults with a Conductive Hearing Impair—
ment Obtained by the Investigator and
Four Pre-Doctoral Students Under Four
Conditions . . . . . . . . . . . . . . . . 91

6. Mean Bone-Conduction Thresholds for Eight
Adults with a Sensorineural Hearing Dis-
order Obtained by the Investigator and
Four Pre-Doctoral Students Under Four
Conditions . . . . . . . . . . . . . . . . 94

7. Mean Bone—Conduction Threaholds for Eight
Adults with a Conductive Hearing Impair—
ment Obtained by the Investigator and
Four Pre-Doctoral Students Under Four
Conditions . . . . . . . . . . . . . . . . 95

ix

Figure

8.

10.

ll.

12.

13.

Mean Air—Conduction Thresholds for Eight
Normal Hearing Children with Normal Arti—
culation Obtained by the Investigator and
Four Pre-Doctoral Students Under Four
Conditions . . . . . . . . . . . . . . . .

Mean Air—Conduction Thresholds for Eight
Normal Hearing Children with Defective
Articulation Obtained by the Investigator
and Four Pre—Doctoral Students Under Four
Conditions . . . . . . . . . . . . . . . .

Mean Air—Conduction Thresholds for Eight
Adults with a Sensorineural Hearing Dis—
order Obtained by the Investigator and
Four Pre—Doctoral Students Under Four
Conditions . . . . . . . . . . . . . . . .

Mean Air—Conduction Thresholds for Eight
Adults with a Conductive Hearing Dis—
order Obtained by the Investigator and
Four Undergraduate Students Under Four
Conditions . . . . . . . . . . . . . . . .

Mean Bone-Conduction Thresholds for Eight
Adults with a Sensorineural Hearing Dis-
order Obtained by the Investigator and
Four Undergraduate Students Under Four
Conditions . . . . . . . . . . . . . . . .

Mean Bone-Conduction Thresholds for Eight
Adults with a Conductive Hearing Dis—
order Obtained by the Investigator and
Four Undergraduate Students Under Four
Conditions . . . . . . . . . . . . . . . .

Page

110

111

126

127

132

133

Appendix

AI

B.

C.

D.

LIST

OF APPENDICES

Statistical Results of Experiment I .

Statistical Results of Experiment II

Statistical Results of Experiment III

Statistical Results of Experiment IV

Summary Showing
in Experiment

Summary Showing
in Experiment

Summary Showing
in Experiment

Summary Showing
in Experiment

Data for Each Subject
I . . . . . . . . . .

Data for Each Subject
II . . . . . . . . .

Data for Each Subject
III I I I I I I I I I

Data for Each Subject
IV . . . . . . . . .

A Copy of the Letter Sent to Each Adult
Subject Who Participated in This

Study . . . .

xi

Page
157
159
163

166

168

171

177

183

193

. -p‘-

CHAPTER I

INTRODUCTION

In the field of audiology, research has been
conducted regarding reliability and validity of audio—
metric tests. Research on threshold reliability has
been concerned with inherent stability of electronic
equipment used in the measurement of auditory thresholds
or the inherent stability of the human auditory threshold
itself. There has been very little experimental evidence
reported regarding the reliability and validity of speech
audiometry. Also, there are no reported studies related
to how the experimenter or audiologist may affect the
reliability and validity of audiometric tests. A few
articles, however, suggest that a certain amount of error
may be attributed to preconceived expectations of the
tester. Thus, according to the literature reviewed, there
is a dearth of information concerning experimenter biases

and effects on clinical audiometric results.

Purpgse of the Study

 

The major purpose of this investigation was to
determine whether certain types of previous audiometric

information cause the tester to influence obtained

pure-tone and speech audiometric results. Specifically
this study was concerned with the possibility of testers'
expectancy effects on audiometric responses obtained from
normal hearing adults, hard-of—hearing adults having
either conductive or sensorineural hearing impairments,
and normal hearing children having either normal or defec-
tive articulation.

Two groups of testers were used that varied con—
siderably in terms of amount of training in clinical
audiology skills. They were pre-doctoral students in
audiology and undergraduate students who were taking an
introductory course in audiology. The basic question
posed prior to the investigation was as follows: does
previous audiometric test information cause the tester to
influence significantly audiometric scores as a function
of the following variables:

1. degree of sophistication or amount of training
of the tester

2. the auditory sensitivity of the subject
(normal hearing, conductive hearing loss or
sensorineural hearing loss)

3. pure-tone air-conduction versus bone-conduc—
tion thresholds

4. speech reception thresholds
5. speech discrimination scores

6. pure-tone threshold versus speech reception
thresholds

7. defective articulation versus normal articu-
lation on speech audiometric scores

8. adults versus children subjects

Based on the variables listed above the following
research questions were posed.

1. Does previous audiometric test information
(true or erroneous) influence significantly sophisticated
audiologists when obtaining pure-tone air-conduction
thresholds from normal hearing adults?

2. Does previous audiometric test information
(true or erroneous) influence significantly sophisticated
audiologists when obtaining pure-tone air-conduction
thresholds from adults having a sensorineural hearing
disorder?

3. Does previous audiometric test information
(true or erroneous) influence significantly sophisticated
audiolists when obtaining pure-tone air—conduction
thresholds from adults having a conductive hearing impair—
ment?

4. Does previous audiometric test information
(true or erroneous) influence significantly sophisticated
audiologists when obtaining pure-tone air—conduction
thresholds from normal hearing children?

5. Does previous audiometric test information
(true or erroneous) influence significantly SOphisticated
audiologists when obtaining pure-tone bone-conduction
thresholds from adults having a sensorineural hearing

disorder?

6. Does previous audiometric test information
(true or erroneous) influence significantly sophisticated
audiologists when obtaining pure-tone bone—conduction
thresholds from adults having a conductive hearing impair-
ment?

7. Does previous audiometric test information
(true or erroneous) influence significantly sophisticated
audiologists when obtaining speech reception thresholds
from normal hearing adults?

8. Does previous audiometric test information
(true or erroneous) influence significantly sophisticated
audiologists when obtaining speech reception thresholds
from adults having a sensorineural hearing disorder?

9. Does previous audiometric test information
(true or erroneous) influence significantly sophisticated
audiologists when obtaining speech reception thresholds
from adults having a conductive hearing impairment?

10. Does previous audiometric test information
(true or erroneous) influence significantly sophisticated
audiologists when obtaining speech reception thresholds
from normal hearing children having normal speech articu-
lation?

11. Does previous audiometric test information
(true or erroneous) influence significantly sophisticated
audiologists when obtaining speech reception thresholds
from normal hearing children having defective speech

articulation?

 

12. Does previous audiometric test information
(true or erroneous) influence significantly sophisticated
audiologists when obtaining speech discrimination scores
from adults having a sensorineural hearing disorder?

13. Does previous audiometric test inférmation
(true or erroneous) influence significantly sophisticated
audiologists when obtaining speech discrimination scores
from adults having a conductive hearing impairment?

14. Does previous audiometric test information
(true or erroneous) influence significantly sophisticated
audiologists when obtaining speech discrimination scores
from normal hearing children having normal speech articu-
lation?

15. Does previous audiometric test information
(true or erroneous) influence significantly sophisticated
audiologists when obtaining speech discrimination scores
from normal hearing children having defective speech
articulation?

16. Does previous audiometric test information
(true or erroneous) influence significantly unskilled
testers when obtaining pure-tone air-conduction thresholds
from adults having a sensorineural hearing disorder?

17. Does previous audiometric test information
(true or erroneous) influence significantly unskilled
testers when obtaining pure-tone air-conduction thresholds

from adults having a conductive hearing impairment?

 

18. Does previous audiometric test information
(true or erroneous) influence significantly unskilled
testers obtaining pure-tone bone-conduction thresholds
from adults having a sensorineural hearing disorder?

19. Does previous audiometric test information
(true or erroneous) influence significantly unskilled
testers obtaining pure—tone bone-conduction thresholds
from adults having a conductive hearing impairment?

The primary intent of this research was to deter—
mine whether in a clinical audiometric evaluation previous
audiometric information contributes significatnly to
obtaining erroneous audiometric configurations and there—
fore, misdiagnosis. As stated above, the literature shows
that researchers in audiology have ignored, experimentally,
the influences of experimenter biases in recording audio—
metric test results. Several articles make reference to
the fact that auditory scores are sometimes erroneously
‘reported because of the otologist's past experiences, but
these statements are not based upon experimental evidence.
According to Sataloffl, a serious mistake made when ob-
taining audiometric results is for the audiologist or
otologist to have preconceived ideas of the patient's

auditory sensitivity at various frequencies and conclude

 

lJoseph Sataloff, "Pitfalls in Routine Hearing
Tests," Archives of Otolaryngology, 73 (1961), pp. 717-726.

 

that a specific audiometric configuration always means a
specific pathology. A variety of variables such as physio-
logical, psychological, and methodological are considered
to be the major sources of error in audiometry.l

About forty years ago, psychologists began to in—
vestigate experimentally the personal effects that
examiners had on the test scores obtained from their
clients.2 The literature reveals that studies on experi-
menter bias and experimenter effects have been conducted
in many different disciplines. However, none of the
reported investigations were conducted involving the
sense of hearing nor have any of these studies been con-
cerned with threshold measurement. As previously stated,
however, studies have been designed in the field of
audiology to determine the major sources of errors
encountered in audiometry. Factors contributing to errors
include faulty calibration of equipment, ambient noise,
and the use of different psychophysical methods to obtain

thresholds.

 

lWallace S. High, Aram Glorig, and James Nixon,
"Estimating the Reliability of Auditory Threshold Measure-
ments," Journal of Auditory Research, 1 (1961), pp. 247-
262.

 

2Robert Rosenthal, "Experimenter Attributes as
Determinants of Subjects' Responses," Journal of Projective
Techniques, 27 (1963), pp. 324-331.

In the interest of accurate audiometric
assessment, it is important to know whether previous
audiometric information does cause the audiologist to
influence significantly the patients' audiometric scores.
Because of the paucity of information in this area of

audiology, this study was proposed.

Definition of Terms

 

The following terms have been used extensively
in this investigation. For clarification and convenience,
they are defined below.

Experimenter Bias.-—Occurs when the experimenter

 

obtains results from the subject that he expects to obtain.1

Experimenter Effect.-—Occurs when different ex—

 

perimenters obtain different data from the same subjects.2

Normal Hearing.--For the purpose of this study,
this term means that the individual's pure-tone air and
bone-conduction thresholds are interweaving (i 5 dB) and
are no poorer than 25 dB (ISO 1964 standard) for the
octaves between 250 and 8000 Hz.

Conductive Hearing Loss.—-For the purpose of this

 

study, this term means that the hearing loss is due to an

 

1Neil Freidman, Daniel Kurland, and Robert Rosenthal,
"Experimenter Behavior as an Unintended Determinant of
Experimental Results," Journal of Projective Techniques,
29 (1965), pp. 479—490.

 

21bid.

impairment in the outer or middle ear or both in the
presence of a normal inner ear. The condition is observed
clinically when bone conduction thresholds are normal and
air-conduction thresholds are depressed. A person was
considered to have a conductive hearing loss whenever

his bone-conduction thresholds for 250, 500, 1000, 2000,
and 4000 Hz were at least 10 dB more sensitive than the
same frequencies tested by air—conduction. Also, the
sensitivity of the bone-conduction thresholds could be

no poorer than 25 dB (ISO 1964 standard).

Sensorineural Hearing Loss.--For the purpose of

 

this study, this term means that the hearing loss is
attributed to malfunction of the inner ear. The condition
is observed clinically when air and bone-conduction
thresholds are interweaving (i 5 dB) and a hearing loss

is present. A person was considered to have a sensori—
neural hearing loss when thresholds were poorer than 25

dB (ISO 1964 standard) for at least the frequencies of

2000, 4000, and 8000 Hz.

CHAPTER II

REVIEW OF THE LITERATURE

This chapter is concerned with the reliability of
pure—tone and speech audiometric thresholds and speech
discrimination scores. It is also concerned with research
conducted regarding the phenomena of experimenter bias and
experimenter effect. Specifically, the chapter reveals
the diversified areas in which these phenomena have been
studied and the dearth of information existing in the area

of audition.

Threshold Measurements

 

Man has been interested in the measurement of
hearing sensitivity for the past four centuries.1 ‘However,
this interest long preceded the development of standard-
ized test methods and equipment that adequately and
systematically measured auditory sensitivity.

Electronic equipment has been developed that is
thought to measure precisely auditory thresholds via

pure-tone and Speech stimuli. Scientists have not,

 

1John J. O'Neill and Herbert J. Oyer, A lied
Audiometry (New York: Dodd, Mead & Company, Inc., 1966),
pp. 36—51.

10

 

11

however, developed an instrument to determine auditory
thresholds without having a human relate instructions,
manipulate dials, interpret results, or make recommenda-
tions. Even auditory thresholds obtained by so—called
automatic audiometry cannot be isolated from some type of
human influence. Thus, when attempting to define auditory
thresholds, a number of variables, including the human
variable, must be taken into account.

Dixon Wardl pointed out auditory threshold is not
a simply defined phenomenon. According to Ward, the long
accepted definition that threshold is a constant energy
barrier (". . .if the signal energy exceeds this level,
it will be perceived: if the energy is less than this
critical value, it will not“) is no longer adhered to by
authorities of threshold measurement. Persons sophisti—
cated in areas pertaining to auditory threshold measure—
ment are well aware of the inadequacy of this definition.
It is presently known that numerous variables must be
controlledPWhen attempting to define or measure the human
auditory threshold. Some of these variables include:
the spectrum and duration of the stimulus, the psycho-
physical method employed when conducting the measurement,

practice effects, instructions given to the subject, the

 

1Dixon Ward, "Auditory Fatigue and Masking," ed.
James Jerger, Modern Develgpments in Audiology (New York:
Academic Press, 1963), PP. 242- 243.

 

 

12

emotional state of the subject, age of the subject,
listener fatigue, medical history of the subject, trans-
ducer placement, calibration and stability of the equip—
ment, sophistication of the examiner, and ambient noise
in the test environment.

Authorities in audiology and otology are in agree-
ment with Ward in believing that auditory thresholds can—
not be defined as a static phenomenon that is only
dependent on the stimulus being presented. Many of the
above variables that influence auditory thresholds have
been enumerated by High, Glorig, and Nixon.1

Sataloff2 also has listed various pitfalls common
in audiology and otology that influence auditory thresholds.
One of the most consequential is that the audiologist or
otologist or both have a biased expectancy of what a
threshold ought to be and then consciously or unconsciously
record and interpret thresholds as a function of these
expectancies.

Rintelmann and Harford3 explained that in Bekesy

audiometry, where the listener traces his threshold

 

lHigh, Glorig, and Nixon, "Estimating Reliability
of Thresholds," pp. 247-262.

2Sataloff, "Pitfalls in Routine Hearing Tests,"

3William F. Rintelmann and Earl R. Harford, "Type
V Bekesy Pattern: Interpretation and Clinical Utility,"
Journal of Speech and Hearing Research, 10 (1967), pp.
733-744.

 

13

independent of the tester presenting the stimulus, errors
still arise in the classification of such thresholds.
These authors pointed out explicitly how Bekesy type
classification can be influenced significantly by errone—
ous interpretations of the separation between continuous
and pulsed stimuli. Further, according to Stream and
McConnelll different thresholds are obtained when slight
differences in subject instruction are given for the
adjustment of a pure—tone in automatic audiometry.

Even when the aforementioned variables are con-
trolled there is still appreciable fluctuation in the
human auditory threshold. Investigations have shown that
at a given time a specific stimulus of determined magni-
tude may elicit a response, and at another moment this
same stimulus with increased intensity will not be per-
ceived by the same listener.2 Menzel3 has attempted to
explain the fluctuation of auditory thresholds and the

variability related to threshold measurement. He stated:

 

1Richard W. Stream and Freeman McConnell, "A
Comparison of Two Methods of Administration in Bekesy—Type
Audiometry," Journal of Auditory Research, 4 (1961), pp.
263-271.

 

2

Ward, "Auditory Fatigue and Masking, pp. 242—243.

3Otto J. Menzel, "Error in Audiometry," Eye Ear
Nose and Throat Mgnthly, 42 (1963), p.

 

l4

". . .in nearly all instances something other than the
intended stimulus has determined the overt responses upon
which the auditory measurement is based.“

From the literature reviewed, it is quite evident
that the auditory threshold must be defined according to
measurement technique for each research project in which
threshold values are obtained. This is particularly
necessary when defining the extent to which the variability
of measurement can deviate on repeated measures. In other
words, threshold can be defined theoretically as a parti—
cular point with a i 2 decibel of variability, or it can
be defined as a specific point with a i 5 dB margin of
variability. Often, the amount of variability is depen—
dent upon the size of the attenuator step (2 dB versus

5 dB) employed in the threshold measurement.

Reliability of Audiometric Tests

 

Whenever an experimenter or clinician administers
a test or a series of tests he is usually interested in
the validity and the reliability of the instrument or
instruments he uses. The validity of a test is the degree
to which the instrument measures what it purports to
measure. The reliability of a test is how consistently
the instrument measures a specific parameter.

Since the primary purpose of this investigation

was to determine how the consistency of various audiometric

15

tests was affected by a number of testers receiving
different audiological information, it was necessary to
determine the reliability of the tests used in this re—
search project.
Jergerl defines reliability as applied to audio-
logical tests in the following manner:
Reliability is repeatability of test scores. Abso—
lute consistency refers to the absolute variability
in performance from test to retest. It is concerned
with basic precision of measurement irrespective of
the extent to which rank order is preserved from test
to retest. It is defined by the standard error of
measurement. It may be concretely visualized as the
standard deviation of random error of measurement--
symbolized Se.
Following are the test-retest results obtained by

various investigators concerning the audiometric stimuli

used in this study.

Pure-Tone Air—Conduction Stimuli

In 1939 Witting and Hughson2 conducted a study
to determine the errors in pure—tone air—conduction
thresholds of hard-of—hearing and normal hearing indivi-
duals. These authors found that on repeated threshold

measures the smallest variability occurred at 1000 Hz with

 

1James F. Jerger, "Comparative Evaluation of Some
Auditory Measures," Journal of Speech and Hearing Research,
5 (1962), pp. 3-17.

2E. G. Witting and Walter Hughson, "Inherent
Accuracy of a Series of Repeated Clinical Audiograms,"

Laryngoscope, 50 (1940), pp. 259—269.

 

16

125 and 8000 Hz showing the greatest variability. The
major findings of Witting and Hughson are enumerated
below.

1. Hard-of—hearing individuals exhibit an inher-
ent error of less than 5 dB for pure-tone air—conduction
stimuli. The least variability occurred at 1000 Hz.
Normal hearers were more consistent and exhibited less
variability at each of the frequencies tested.

2. Several large deviations occurred during the
testing that were not explained by the authors; however,
all of the large deviations were found with the hard-of—
hearing subjects.

3. Subjects tended to demonstrate better thresh-
olds on repeated measures even when these repetitions
were made two years after the initial testing.

4. Variations greater than reported by these
authors should be considered erroneous and related to
factors other than the listener's hearing sensitivity.

Carhart and Hayes1 confirmed Witting and Hughson's
findings that the standard error of measurement of air-
conduction audiometry is a function of frequency. They

found that 1000 Hz was the least variable frequency

 

lRaymond Carhart and Claude Hayes, "Clinical Re-
liability of Bone Conduction Audiometry," Laryngoscope,
50 (1940), pp. 1084—1101.

17

5.9 dB). These authors reported the following standard
errors of measurement: 6.5 dB at 250 Hz, 5.9 dB for 1000
Hz, and 6.8 dB for 4000 Hz.

Jergerl found the absolute variability for air-
conduction audiometry to be 5 dB at 250 Hz, 6 dB at 1000
Hz, and 4.7 dB at 4000 Hz. These variations are compar-
able to the findings of Carhart and Hayes.

In an attempt to standardize clinical procedures
for obtaining pure-tone air and bone-conduction thresholds,
Carhart and Jerger2 found that when presenting pure-tone
stimuli in an ascending manner, the test-retest repeat—
ability for 250 Hz was 4 dB, for 1000 Hz was 4.6 dB, and
for 4000 Hz was 7.2 dB.

Harris and Myers3 found that the inherent stability
of pure-tone air-conduction thresholds for normal hearers
was less than i 5 dB. These authors found that when using
a 1 dB step attenuator, this variability or inherent

stability were reduced from i 5 dB to i 2 dB.

 

lJames Jerger, "Comparative Evaluation of Some
Auditory Measures," pp. 3-17.

2Raymond Carhart and James Jerger, "Preferred
Method of Determination of Thresholds," Journal of Speech
and Hearing Disorders, 24 (1959), pp. 330-345.

 

 

3J. Donald Harris and Cecil Myers, "Experiments
on Fluctuations of Auditory Acuity," Medical Research
Laboratory U. S. Naval Submarine Base, New London, Conn-
ecticut, Volume XI No. 13 Report No. 196, 22 June 1952,
Project NM 003 041.21.08.

18

Pure-Tone Bone—Conduction Stimuli

Carhart and Hayes1 measured the absolute
consistency of pure—tone bone—conduction stimuli and were
interested in determining the relationship between air—
conduction and bone—conduction threshold stability. They
found the test—retest reliability for bone—conduction
thresholds to be equally or more reliable than for air—
conduction. They concluded that when bone-conduction
audiometry is administered in an appropriate manner, its
accuracy is as good as that of air-conduction audiometry.
As was found for air—conduction stimuli, the least
variability occurred at 1000 Hz (5.4 dB). This variability
increased as the frequencies became higher. The authors
reported absolute variability of 5.4 dB for 250 Hz and
6.6 dB for 4000 Hz.

Jerger2 also concluded that test-retest reliability
for bone-conduction audiometry is comparable to conven-

tional air—conduction audiometry.
Spondaic Word Stimuli

Although audiologists have notions of test-retest

consistency based upon empirical clinical evidence, there

 

1Carhart and Hayes, "Clinical Reliability of Bone
Conduction Audiometry," pp. 1084-1101.

2Jerger, “Comparative Evaluation of Some Auditory
Measures," pp. 3-17.

19

is a dearth of laboratory information on the test-retest
consistency of speech audiometry.1
Research indicates that the average value for the
pure-tone air-conduction thresholds of 500, 1000, and
2000 Hz correlate positively with speech reception
thresholds obtained by using Spondaic words.2’3’4’5
Hudgins and his associates6 developed two lists

of Spondaic words which were named Auditory Test No. 9.

The authors determined that the absolute test-retest

 

lIbid.

2Raymond Carhart, "Speech Reception in Relation
to Pattern of Pure—Tone Loss," Journal of Speech Disorders,
11 (1946), pp. 97-108.

 

3Raymond Carhart, "Individual Differences in
Hearing for Speech," Annals of Otolo , Rhinolo , and
Laryngology, 55 (1946), pp. 223—267.

4James F. Jerger, Raymond Carhart, and Tom W.
Tillman, "Some Relations Between Normal Hearing for Pure
Tones and for Speech," Journal of Speech and Hearing, 2
(1959), pp. 126-140.

5Harvey Fletcher, "A Method of Calculating Hearing
Loss for Speech from an Audiogram," Journal of the Acoust—
ical Society of America, 22 (1950), pp. 1-5.

 

 

6C. V. Hudgins, J. F. Hawkins, J. E. Karlin and
S. S. Stevens, "The Development of Recorded Auditory Tests
for Measuring Hearing Loss for Speech," Laryngoscope, 57
(1947), PP. 57-89.

20

reliability of these words was 2.8 dB. Hirsh and his
associates1 developed the CID Auditory Test W-l (based on
Auditory Test No. 9) to measure speech thresholds. How-
ever, these authors did not publish test-retest values
for the W-l test.

In a study comparing 2 and 5 dB methods of deter-
mining spondee thresholds (speech reception thresholds),
Chaiklin and Ventry2 found that when using a 2 dB-step
attenuator, ninety—three percent of their twenty-seven
subjects had test-retest differences from 0 to i 6 dB.

No subject had a test-retest difference greater than 8 dB.

Chaiklin and Ventry3 reported that Barrett found
that ninety-six percent of his normal hearing subjects
repeated speech reception thresholds within 1 6 dB and
that a joint study conducted by the National Institute of
Health in cooperation with the Veterans Administration
reported that ninety-three percent of their subjects had

test-retest scores of r 6 dB on spondee words.

 

1Ira Hirsh, Hallowell Davis, S. Richard Silverman,
Elizabeth G. Reynolds, Elizabeth Eldert, and Robert Benson,
"Development of Materials for Speech Audiometry," Journal
of Speech and Hearing Disorders, 17 (1952), pp. 321-337.

 

2Joseph B. Chaiklin and Ira M. Ventry, "Spondee
Threshold Measurement: A Comparison of 2- and 5- dB
Methods," Journal of Speech and Hearing Disorders, 29
(1964), 47-59

 

3Joseph B. Chaiklin and Ira M. Ventry, "Functional
Hearing Loss," ed. James Jerger, Modern Developments in
Audiology (New York: Academic Press, 1963), pp. 76-125.

 

 

 

 

 

 

 

pi

5pe<
(196

Test
flbrds
Hero!

Air p

 

21

The results of studies conducted by Rhum and
Carhart1 and Tillman and Jerger2 have shown that test-
retest variability should not exceed 1 6 dB for spondee

words.

Speech Discrimination Stimuli

For the past twenty-five years monosyllabic words
have been used to measure a listener's ability to dis-
criminate speech. The first monosyllabic word lists were
the twenty PB-50 word lists constructed at Harvard during
World War II. From these original lists several mono-
syllabic, phonetically balanced word lists have been
develoPed. All of these lists consist of fifty one-
syllable words that are relatively familiar to most
listeners.3

Tillman and Carhart4 developed four lists of fifty

phonetically balanced monosyllabic words, Northwestern

 

1Howard B. Rhum and Raymond Carhart, "Objective
Speech Audiometry," Journal of Speech and Hearing Research,
1 (1958), pp. 169-178.

2Tom W. Tillman and James F. Jerger, "Some Factors
Affecting the Spondee Threshold in Normal-Hearing Subjects,"
Journal of Speech and Hearing Research, 2 (1959), pp. 141-
I46.

 

 

3Raymond Carhart, "Problems in the Measurement of
Speech Discrimination," Archives of Otolaryngology, 82
(1965), pp. 253-260.

4Tom W. Tillman and Raymond Carhart, "An Expanded
Test for Speech Discrimination Utilizing CNC Monosyllabic
Words Northwestern University Test No. 6, USAF School of
Aerospace Medicine Areospace Medical Division (AFSC) Brooks
Air Force Base, Texas (1966), pp. 1-12.

 

 

22

University Auditory Test No. 6, to measure speech
discrimination. These lists were found to be equivalent
and yielded good test-retest reliability. At a 24 dB
sensation level the absolute variability of these words
for normal hearers was 2% and it was 3.9% for individuals
with sensorineural hearing impairments. In a study con-

ducted by Rintelmann and Jetty1

at Michigan State Univers-
ity it was found that the standard error of measurement
for Northwestern University Auditory Test No. 6 was 3.58%
when presented to normal hearers at a 24 dB sensation
level.

Haskins2 developed four lists of fifty phonetically
balanced words for testing children. These were named PBK-
50's. Haskins stated: ". . .the retest presentation used
during the present study appears to be unsuitable as a
means of estimating the reliability of the materials and
the procedures employed." Nevertheless on PBK-List l,

which was the List used in the present investigation,

Haskins reported test-retest values of 2.48%.

 

1William F. Rintelmann and Albert J. Jetty,
Reliability of Speech Discrimination Testing using CNC
Monosyllabic Words" (Unpublished Study, Michigan State
University, 1968).

2Harriet Haskins, "A Phonetically Balanced Test
of Speech Discrimination for Children," (Unpublished
Master's Thesis, Northwestern University, 1949).

 

 

 

23

Carhartl stated:

In general as long as the test items are meaning-
ful monosyllables for the patient and their phonetic
distribution is appropriately diversified, one 50-
word compilation is relatively equivalent to another.

Therefore, based on Carhart's statement, it is assumed
that the Phonetically Balanced Kindergarten Word Lists

have approximately the same test-retest repeatability as

the Northwestern University Auditory Test No. 6.

Summary

It has been adequately established that for pure-
tone air and bone—conduction thresholds, the test-retest
repeatability should not exceed 1 5 dB for the octave
frequencies between 250 and 8000 Hz. In other words,
test-retest variability for conventional pure-tone audio-
metry should be no greater than i 5 dB. When employing
spondee words to establish speech reception thresholds,
it has been found that test-retest variability should not
exceed 1'6 dB. Also, there is a positive correlation
between pure-tone air-conduction thresholds of 500, 1000,

and 2000 Hz and speech reception thresholds.

 

1Carhart, "Problems in the Measurement of Speech
Discrimination," p. 254. .

 

 

 

 

 

 

 

 

re:

the

the

 

24

Although more s0phisticated research is needed
concerning test-retest reliability of phonetically balanced
monosyllabic words, it has been illustrated that these
words have an absolute consistency on the order of i 4%.

The Phenomena of EXperimenter Bias
and Experimenter Effect

 

The placebo effect in medical practice illustrates
that humans can be readily influenced. Shapiro gave a
detailed description of the placebo effect.

This is the psychological, physiological, and
psych0physiological effect of any medication or pro-
cedure given with therapeutic intent, which is inde-
pendent of or minimally related to the pharmacological
effect of the medication or to the specific effects
of the procedure, and which Operates through the
psychological mechanism. ‘Although the pharmacological
effect of a drug may have been deleterious or of little
consequence to the organism, its effect could have been
beneficial nevertheless.

It can be said that the placebo effect is the
phenomenon that causes the patient to respond in a posi-
tive manner physiologically, if psychologically he expects
to respond in a positive manner.,

In the late 1920's psychologists became interested
in obtaining information about how they were affecting the

reSponses of their patients. They were also concerned with

the possibility that a different psychologist could obtain

 

lArthur Shapiro, "A Contribution to a History of
the Placebo Effect," Behavioral Science, 5 (1960), p. 109.

 

 

V.\.,N
a 9
La I

Bet

 

 

 

 

 

 

 

25

different results from the same patient doing the same
task.1 From the psychologists' interest, research projects
were designed that eXperimentally investigated the pheno-
mena of experimenter bias and exPerimenter effect. The
degree to which the tester is influenced by previous test
information has been labeled eXperimenter bias. The
degree to which the tester influences a person's test
results has been titled experimenter effect. In other
words, an experimenter bias occurs when experimenters
obtain scores from their subjects that they expect to
obtain. Experimenter effect occurs when different experi-
menters record different scores from the same subjects
performing essentially the same tasks.2
Since the first experiments eXploring eXperimenter
bias and effect, numerous investigations concerning these
phenomena have been reported in several areas. Robert
Rosenthal, an experimental psychologist at Harvard Univers-

ity, recently published a book that is a compilation of

the research exPloring these two phenomena.3

 

lRosenthal, "Experimenter Attributes as Determin-
ants of Subjects' Responses," p. 324.

2Freidman, Kurland, and Rosenthal, "Unintended
Determinant of Experimental Results," pp. 479.

3Robert Rosenthal, Experimenter Effects in Be-
havioral Research (New York:"Appleton-Century-Crofts,
1966). '

 

 

 

 

 

 

 

 

 

 

26

Even though experimenter influences were not
formally reported prior to 1929, people were aware that
many observations and decisions were directly influenced
by the observers' biases or expectancies or both. As
early as the eighteenth century it was determined that
humans observing the movement of the stars disagreed
violently on the velocity at which stars moved. It was
later discovered that these differences in observations
were not intentional errors but rather were beyond the
control of the observer and were classified as "personal
equations."1 In other words, if an astronomer had been
schooled to measure the velocity of the stars in a parti-
cular manner using a specific point of reference, in all
probability his data would differ significantly from
another astronomer who was measuring the speed of the
same star using a somewhat different technique. The
reason for these differences in data might be attributed
to differences in training and the use of different points
of reference. Although both astronomers were probably
unaware of it, their recorded data were being influenced
by their own biases and expectancies relating to the speed
at which a particular star should travel.

Experimenter biases and effects have been observed

as an everyday occurrence in society as well as in the

 

1Ibid., p. 3.

 

27

sciences. It has been illustrated that frequently man is
able to gain acceptance in social activities if the group
feels he can contribute to their ultimate goals; For
example, if the group expects that a certain individual
can perform well enough to enhance their status, this
expectancy may directly influence how the individual per-
forms. Whyte's study1 of social groups found that a man's
ability to bowl was continuously being evaluated; there-
fore, the man's ability to bowl was directly responsible
for the position he earned in the group. If the group
expected the man to bowl adequately or inadequately, they
partially influenced the man's actual performance.

Following are several studies that have been
reported illustrating the phenomena of experimenter bias
and experimenter effect. These examples were selected
from many sources in order to demonstrate the existence
of these phenomena in a variety of situations.

In 1936 it was found that personnel highly skilled
in agriculture were not capable of making selections of

plant life without exhibiting a definite bias.2

 

lWilliam Foote Whyte, Street Corner Society
(Chicago: University of Chicago Press, 1943), pp. 18-23.

 

2William G. Cochran and David J. Watson, "An
Experiment on Observer's Bias in the Selection of Shoot-
Heights," Empire Journal of Experimental Agriculture, 4
(1936), PP. 69-76.

 

 

28

It has been found that interviewers can alter
survey results by reflecting their expectancies during
the interview. The following factors have been discovered
to be the most influential in affecting survey results:
(1) interviewer resistance in stating a question, an
inclination to reword the question or assume the answer
or both; (2) relatively high ambiguity, subjectivity or
complexity in the concept or wording of the inquiry; and
(3) supplementing the questions with additional cues that
will elicit an expected response from the interviewee.1
Harvey2 reported that his experiment in interviewer biases
adequately illustrates that interviewers give sufficient
cues to evoke expected responses. Blakenship3 agreed
with Harvey; however, he attributed the cues to dishonesty

rather than a genuine unconscious bias.

 

1Robert H. Hanson and Eli S. Marks, "Influence of
the Interviewer on the Accuracy of Survey Results,"
Journal of the American Statistical Association, 53 (1958),
pp. 635-655.

2S. M. Harvey, "A Preliminary Investigation of the
Interview," British Journal of Psychology, 28 (1938),
pp. 263-287.

 

3Albert B. Blakenship, "The Effect of the Inter—
viewer Upon the Response in a Public Opinion Poll,"
Journal of Consulting Psychology, 4 (1960).

'I
' I
I

I 1
-

I
u,

‘

 

29

A detailed and controlled study was conducted by
Back1 on the importance of being accepted by the group.
The differences among ways in which cohesiveness was
achieved enabled the author to observe how communication
was employed to "sway" the group. It was found that if
cohesiveness of the group was determined by personal
attraction, the conversation would evolve around a lengthy
and pleasant tOpic. Each member of the group assumed the
role of making the conversation a personal effort. Con-
versation became succinct and quite efficient when
cohesiveness was believed to be based on task performance.
All conversation that was seemingly unimportant was term-
inated and the discussion centered around a theme that was
felt to achieve the task. When the group believed that
cohesiveness was being determined by prestige, the conver-
sation was altered to the extent that there was little or
no risk taken that could possibly hinder cohesiveness.
The group conversed cautiously and each member adjusted
his own actions to conform to the environment established
by the group. It is evident from this investigation that
the group's behavior was significantly governed by what

each member felt was expected of him.

 

1Kurt W. Back, "Influence Through Social Communi-

cation," Journal of Abnormal and Social Psychology, 46
(1951), PP. 9-23.

\m\

I I he“
L...» p m #8.». S t

 

 

 

 

 

 

 

 

30

To further illustrate the "real" existence of
experimenter bias and experimenter effect in research,
several investigations will be described in detail.

Rosenthal and Fodel conducted two replications of
an experiment concerning a person perception task and
found some interesting and revealing results. The authors
presented fifty—seven photographs that showed the faces
of fifty-seven different individuals to 104 students en-
rolled in an introductory psychology course. The students
were asked to rate the faces in the pictures as to how
successful they felt the person was. A rating scale
ranging from a -10 to +10 was used to judge the degree of
failure or success represented by each of the faces in
the pictures. From the original fifty-seven photos, the
authors selected twenty photos that had been rated as
zero by the 104 students. These twenty photos were used
in subsequent exPeriments. In the original experiment a
total of ten students who were psychology majors partici-
pated as experimenters. This group consisted of eight
males and two females. Two-hundred and six undergraduate
students enrolled in an introductory psychology course
served as subjects. All of the exPerimenters were given

the same instructions; however, half the eXperimenters

 

1Rosenthal, Experimenter Effects in Behavioral
Research, pp. 143-157:

 

 

I I \
~w': ‘4.
I 7' -
.t » \

were given ten of the twenty neutral photos and told that

 

31

their subjects should average a +5 rating of these photos.
The remaining experimenters were given the other half of
the twenty neutral photos and told that their subjects
should average a -5 rating of the photos. Prior to the
subjects' ratings of the photos, each experimenter rated
his ten photos using the same twenty point scale. The
experimenters were given specific instructions to be read
to their subjects. The results showed that difference
mean ratings obtained by the eXperimenter expecting suc-
cess (+5) scores and those biased toward failure (-5)
ratings were significant at the .007 level of confidence.
Without exception the eXperimenters expecting success
ratings obtained higher ratings than did their counterparts
expecting failure ratings. The mean scores obtained by
the female experimenters did not differ significantly from
the mean scores recorded by the male eXperimenters.
According to the authors, there was no relationship be-
tween mean photo ratings obtained from subjects or the
magnitude of the biasing phenomenon and the academic
success of the experimenters.

Because the results of the aforementioned experi-
ment were so "clean" and significant, Fode was compelled
to replicate it. He employed twelve male students who
were enrolled in an industrial psychology course as

eXperimenters. The subjects were eighty-six students

 

 

 

 

 

 

Stx
The
sel
Per.
tial

Sign

 

32

enrolled in an introductory course in psychology.
Basically Fode followed the same procedures as were pre-
sented in the original experiment; however, the experi-
menters did not manipulate the photos. The twenty neutral
photos, divided into two groups of ten, were presented on
a large poster and labeled so that the subjects could
express their ratings of each photo to the experimenter.
This was done in order to control for the possibility

that the experimenters were relating their biases to their
subjects by the way they presented the photos. Again,

the results were significant, this time at the 0.0003
level of confidence. As was reported in the original ex-
periment, all experimenters that expected successful
scores obtained higher ratings than did any of their
counterparts who were biased toward lower scores.

Fode followed his first replication with a second
one. In this experiment eight students from an advanced
industrial psychology class participated as experimenters.
The ninety subjects used in this investigation were also
students enrolled in an introductory psychology course.
The author used the Taylor Scale of Manifest Anxiety to
select his experimenters. It was determined that all ex~
perimenters were medium anxious. The results were essen—
tially the same as the two aforementioned studies with

significance at the 0.0005 level.

 

33

The authors concluded that the subjects observed
behavioral cues exhibited by the experimenters. These
cues were significant in influencing the response of each
subject even though the experimenters read identical in—
structions to each subject.

In a similar type of experiment, sound motion
pictures were filmed of the experimenters during the ex-
periment. This procedure was followed in an attempt to
determine how the experimenters revealed their biases to
their subjects. The film revealed that experimenters
exhibited different behavioral patterns when conducting
an experiment. These behavioral patterns were signifi-
cantly influenced by the experimenter's expectancies.
Those experimenters whose conduct revealed a more personal
attitude and warmth obtained ratings of the photos as more
successful; on the other hand, the experimenters who con—
ducted the experiment in a more professional manner and
exhibited more professional competence obtained ratings
that were judged to be failures.l

There is evidence that experimenter's behavior
change whenever they believe that they are to obtain

different results. Two experiments illustrate that

 

lFreidman, Kurland and Rosenthal, "Unintended
Determinant of Experimental Results," pp. 490.

 

34

experimenter biases can be altered within an experiment.
The results of these two experiments demonstrate that
experimenters obtain responses from their students that
they expect to obtain. The data also shows that experi—
menters alter their hypotheses in mid-experiment and then
record results that are in direct agreement with their
revised expectancies.l

There are several experiments reported that in—
vestigate experimenter bias and effect utilizing a verbal
conditioning task. Rosenthal and associates2 reported
that in their experiment, verbal conditioning occurred as
a function of experimenter expectancy. Again, it was re-
ported that experimenters exhibit specific behavioral
patterns that they believe will gain responses in accord
with their expectancies.

Sarason3 reported that the more hostility shown
by the experimenter, the more the subjects are likely to

use hostile verbs when developing sentences. It can

 

lRobert Rosenthal, Ray C. Murly, Gordon W.
Persinger and Linda Vikan—Kline, "Changes in Experimental
Hypotheses as Determinants of Experimental Results,"
Journal of Projective Techniques, 28 (1964), pp. 465-469.

 

2Robert Rosenthal, Paul Kohn, Patricia M.
Greenfield and Noel Carota, "Data Desirability, Experi-
menter Expectancy, and the Results of Psychological Re-
search," Journal of Personalityyand Social Psychology, 3
(1966), PP. 20—27. '-

 

3Irwin G. Sarason, "Individual Differences, Situ—
ational Variables, and Personality Research," Journal of
Abnormal and Social Psychology, 23 (1959) 336-341.

 

“A 'fiI4‘LHI'-l'

 

 

 

35

generally be stated that most research exploring
experimenter biases using verbal conditioning has dis-
covered that the experimenter's need for approval (anxi-
ety) does not validly predict experimenter's behavior
during the experiment.

In an investigation conducted by Masling,1 it was
reported that examiners tended to be more lenient in
scoring answers to various questions selected from the
Wechsler Bellevue II intelligence test when the subjects
were warm and friendly than when they were cold and in-
different. The examiners also gave the warm subjects
more verbal reinforcement and additional chances to re-
spond and document their answers. These examiner
characteristics vanished when administering the test to
"cold" subjects.

An experiment was conducted by Lord2 to determine
the effects of the experimenter's warm, cool, and natural
behavior when administering the Rorschach. She found that
subjects who were given the examination by a warm experi-
menter produced answers that were more detailed and

abstract than when they were given the test by a cold

 

lJoseph Masling, "The Effects of Warm and Cold
Interaction on the Administration and Scoring of an In-
telligence Test," Journal of Consulting Psychology, 23
(1959), 336-341.

2Edith Lord, "Experimentally Induced Variations
in Rorschach Performance," Psychological Monographs, 64
(1961), PP. 321—333.

 

 

 

36

examiner. Bellakl found essentially the same thing when
using a story—telling project as the stimulus. Luft2
also explored the influence of warm and cold behavioral
characteristics contact with these types of behavior. He
discovered that when the experimenter played the role of

a warm examiner, positive responses were obtained. Nega—
tive responses were obtained when the experimenter played
the role of a cold examiner.

All of the studies reported above employed human
subjects. However, investigations employing animal sub-
jects also have been conducted. Two classical experiments
are reported in this regard.

In 1963 Rosenthal and Fode3 reported significant
data that revealed experimenter bias and experimenter
effect in the conditioning of rats. Twelve students in
the field of experimental psychology were employed as
experimenters. The experimenters were given identical
instructions, except that six of them were made to believe

that their rats were "Maze-Bright;" the remaining

 

lLeopold Bellak, "The Concept of Projection: An
Experimental Investigation and Study of the Concept,"
Psychiatry, 7 (1944), pp. 353-370.

2Joseph Luft "Interaction and Projection," Journal
of Projective Techniques, 17 (1953), pp. 489—492.

 

3Robert Rosenthal and Kermit L. Fode, "The Effects
of Experimenter Bias on the Performance of the Albino Rat,"
Behavioral Science, 8 (1963), pp. 183-189.

 

 

 

 

EXpe

 

37

experimenters were told that their rats were "Maze-Dull."
In reality, all of the rats were essentially the same.
The experimenters assigned to the "Maze-Bright" rats were
instructed that they should definitely detect some type
of learning from their animals during the first day.
Also, their rats should exhibit a uniform learning rate
during the experiment. The experimenters observing the
"Maze—Dull" rats were told that these rats should not be
eXpected to show any learning during the first day and
very little, if any, during the entire experiment. The
experimenters utilized a T-maze, which had one white arm
and one gray arm. The maze was so constructed that the
arms could be interchanged.

All of the animals were naive to conditioning
tasks and were classified as Sprague-Dawley Albino rats.
The rats were arranged into thirteen groups and each
group was composed of two male and three female animals.
All rats had been deprived of food for twenty-four hours
prior to the experiment. Each experimenter was asked to
rate his feelings about working with rats prior to the
eXperimenter. This was done because none of the experi-
menters had previous experience working with animal
subjects. The authors assigned the experimenters to work
in pairs. Each pair consisted of a person who related
that he felt he would enjoy working with rats and a person

who felt he would dislike working with rats. One member

 

ex;
fou
ifie
flap:
lea I)
flap;

shew

 

38

of each pair was randomly assigned to a group of rats
designated as being "bright," and one member was to ob-
serve a group of "dull" rats.

Immediately before each experimenter was to test
his rats, he was asked to evaluate on a 20-point scale
how he felt his rats would perform. He was to assign a
+10 if he felt his rats would perform extremely well and
a -10 if he felt his rats would respond poorly.

The experimenters tested each of their rats ten
times a day for five days. Two observations were made
during each test; timing the animal to complete the run
and noting if the run had been made in a correct or in-
correct manner. The gray arm always contained the reward.
This arm was alternated between the right and left sides
equally; however, it was determined randomly when the
arms would be changed. Thus, the alterations varied from
day to day.

It was found that the rats who were labeled as
"Maze-Bright" and who were believed to be brighter by the
experimenters exhibited better performances on the first,
fourth, and fifth days than did the rats who were class-
ified as "dull." The "bright" rats showed systematic
improvement in maze running as would be expected if
learning was occurring. The "Maze-Dull" animals exhibited
improvement at running the maze up to the third day but

showed signs of regression on the fourth day and no change

 

 

 

39

on the fifth day. It was illustrated that the rats being
observed by experimenters who felt that they were "dull"
had much less chance to learn the conditioning task than
did the animals who were thought to be "bright." The
"dull" rats were recorded as being much slower in running
the maze correctly than were the "bright" rats. The
"Maze-Bright" animals exhibited improvement that was more
uniform than did the "Maze—Dull" rats.

At the termination of the experiment, the experi-
menters expressed their feelings about how they reacted
during their rats' performance. Three-fourths of the
experimenters reported that they felt good when their
rats performed correctly and were disappointed when their
animals' performance was inferior.

Another classical study conducted by Corado and
Isonl exploring experimenter bias and effect, employed
planaria (flat worms) as subjects. Seven experimenters
were assigned to record the head turns and body contrac-
tions of planaria. The experimenters were told that half
the planaria were hyperactive and half were hypoactive.
Actually, all of the worms were essentially the same.
This experiment revealed that experimenters reported

twice as many head turns and three times as many body

 

lLucion Corado and James R. Ison, "Observer Bias
in Conditioning of Planaria," Psychological Reports, 13
(1963). pp. 787-789.

 

\.
\I 9 .
\

 

40

contractions when observing worms that were labeled

"hyperactive,' as when recording the movements of the
"hypoactive"ones.

This investigation was replicated by the same
authors using ten different eXperimenters.l Half of the
experimenters recorded the head turns and body contrac-
tions of the "hypoactive" planaria and the other five
observers recorded the head turns and body contractions
of the "hyperactive" worms.

The results were complimentary to the findings of
the original experiment; however, they were more dramatic.
The group observing the "hyperactive" worms reported five
times as many head turns and twenty times as many body
contractions as did the group who were recording the move-
ments of the hypoactive planaria.

Rosenthal and Lawson2 conducted an investigation
that used rats as subjects to illustrate Operant learning.
Again, it was determined that experimenters' beliefs and
expectations about the performance of the animals were

responsible in part for the data obtained.

 

1Ibid.

2Robert Rosenthal and Reed Lawson, "A Longitudinal
Study of the Effects of EXperimenter Bias on the Operant
Learning of Laboratory Rats," Journal of Psychiatric
Research, 2 (1964), pp. 61-72.

 

 

 

mea

5861

are

ass1

 

41

Although this review of the literature is not
exhaustive, it does demonstrate that the phenomena of
eXperimenter bias and eXperimenter effect do occur in
behavioral research. These studies illustrate how an
eXperimenter can evoke or observe responses from his sub—
jects that are in accord with his expectancies. It has
been discovered through sophisticated and well-controlled
research projects that previous information about the
subjects governs significantly the behavior of the eXperi—
menters. Thus, it has been generally found that eXperi-
menters tend to obtain scores from their subjects that
are highly correlated with their expectancies.

Authorities in audiology and otology are aware of
outside influences that contribute to the measurement of
human auditory thresholds. Nevertheless, there is a
dearth of information concerning how the tester influences
the listener's thresholds. There is no information re-
ported in the literature expressing the significance of
testers having previous audiometric results of threshold
measurements. Authorities in audiology and otology have
seemingly assumed that tester bias and effect exist and
are significant in auditory threshold measurements. This

assumption has been made by High, Glorig, Nixon,l Sataloff,2

 

lHigh, Glorig, and Nixon, "Estimating Reliability
of Thresholds," pp. 247-262.

2Sataloff, "Pitfalls in Routine Hearing Tests,"
pp. 717-726.

 

42

and Ward;1 however, none of these authors have documented
their statements with research that supports this assump-
tion.

Chaiklin and Vrntry2 reported that they used a

control to eliminate the tester bias effect. They stated:
". . .the measurement process and the resulting thresholds
may be biased by the clinician's tendency to estimate
spondee threshold from pure-tone averages." These authors
believe that if the clinician adheres to stringent method-
ology, he will eliminate the bias effect. Again, the
authors did not cite evidence that reveals the existence
of the bias effect in auditory threshold measurements.
In a dissertation by Barrett,3 it was reported that the
author controlled for tester bias by employing a Random
Variable Attenuator; however, the writer did not report
that the bias effect had been shown to contribute signi—
ficantly to auditory measurements obtained.

If tester bias and tester effect are truly sig-

nificant variables in auditory threshold measurements,

they should be investigated under controlled conditions.

 

lWard, "Auditory Fatigue and Masking," pp. 242-243.

2Joseph B. Chaiklin and Ira M. Ventry, "Spondee
Threshold Measurement: A Comparison of 2 and 5-dB Methods,"
pp. 54-550

3Lyman S. Barrett, "Threshold Relationships in
Simulated Hearing Loss" (Unpublished Doctor's Thesis,
Stanford University, 1959).

 

 

cox
mad
are

bei

pub]
ment
attr
reSe
Will
hadg
and j
worki
in te

the p‘

as Det

 

43

According to Rosenthal,l most of the research
reported to date concerning experimenter bias and effect
has been conducted in the areas of survey research and
clinical psychology and very few investigations have been
conducted in a laboratory setting. It was the purpose of
the present study to investigate the phenomena of eXperi—
menter bias and experimenter effect upon audiometric test
results obtained by student testers under a variety of
conditions. Since auditory threshold measurements are
made using electronic equipment and quantified results
are recorded, these measurements can be considered as
being conducted in a laboratory setting.

The reasons given by Rosenthal for the lack of
published research to determine the phenomena of experi-
menter bias and effect in a laboratory setting may be
attributed to the following: (1) in survey or clinical
research it is more feasible to construct conditions that
will produce biases and effects; (2) investigators have
had greater Opportunity to conduct research in the field
and in the clinic; and (3) the behavior of the person
working in the laboratory can be more highly structured
in terms of recording responses than can the behavior of

the person performing the eXperiment in the field or in

 

lRobert Rosenthal, "Experimenter Modeling Effects
as Determinants of Subjects' Responses," Journal of Pro-
jective Techniques, 27 (1963), pp. 467-471.

 

 

 

 

 

 

44

in the clinic.1 Thus, it would appear that experimenter
biases and effects would be less likely to occur in a
laboratory setting. Nevertheless, Rosenthal feels that
there is no reason why these phenomena cannot occur in
some circumstances utilizing a laboratory setting.
However, he stated ". . .for a full understanding of how
these effects Operate, we must wait for the results of

research perhaps not yet begun."2

 

lIbid.

2Rosenthal, Experimenter Effects in Behavioral
Research, p. 119.

 

 

CHAPTER III

PROCEDURES

This investigation was composed of four individual
experiments that were designed to study the phenomena of
experimenter (tester) bias and effect as related to pure—
tone and speech reception threshold measurement and speech
discrimination. The variables were as follows: age of
subjects, audiometric tests employed, articulation of
subjects, status of subjects' hearing mechanism, sophis—
tication of testers, and types of audiometric information
(conditions).

This chapter is organized so that the experimental
conditions, testers, subjects and tests administered are
described for each experiment and the equipment and pro-
cedures employed are explained for the overall study.
However, it is felt that a brief overview of the entire

research project is appropriate at this time.

Overview of Experiments

 

There were two groups of testers with four testers
in each group. One group of testers was composed of pre—
doctoral students in audiology and the second group of

testers consisted of undergraduate audiology and speech

45

 

46

science majors who were enrolled in an introductory
audiology course. The subjects employed in this study
consisted of both adults and children. The adults were
selected on the basis of their hearing sensitivity and
audiometric configurations. The children were selected
on the basis of their hearing sensitivity and their
ability to articulate English phonemes.

The conditions, the testers, and the subjects are
described immediately below for the overall study.

The following paragraphs briefly describe the four

experiments in this study.

Experiment I

In this eXperiment four SOphisticated clinical
audiologists were employed as testers to study experi-
menter bias and effect upon pure-tone air-conduction
thresholds and speech reception thresholds of eight normal

hearing adults.

Experiment II

In this experiment the same four testers were
utilized to explore the bias and effect phenomena as
related to pure-tone air and bone-conduction thresholds
and speech audiometry scores, Speech reception thresholds
and Speech discrimination scores of eight adults with
sensorineural hearing disorders and eight adults with

conductive hearing impairments.

 

47

Experiment III

In this experiment the same four testers were
used to investigate the bias and effect phenomena as
related to pure-tone air-conduction thresholds and speech
audiometry scores (speech reception thresholds and Speech
discrimination scores) of sixteen normal hearing children.
Half of the children had defective articulation and the

remainder had normal articulation.

Experiment IV

In this experiment four unsophisticated testers
were used to study experimenter bias and effect as re-
lated to pure-tone air and bone-conduction thresholds of
sixteen hard-of—hearing adults. Half of these subjects
had a sensorineural hearing loss and the remaining eight
exhibited a conductive hearing disorder.

The variables listed above and in Chapter I were
investigated by studying the outcome of the four experi-

ments and the relations between these experiments.

Experimental Conditions

 

To determine the extent to which the testers in—
fluenced subjects' pure-tone air and bone-conduction
thresholds and speech audiometry scores, four types of
audiological conditions were constructed concerning the

type of information given to the testers at the beginning

 

 

 

48

of the test session as related to the subjects' hearing
sensitivity and speech discrimination. The conditions
were as follows: actual test results obtained previously,
better than actual test results, poorer than actual test
results, and no test results.

Actual test results were the subjects' pure-tone
thresholds and speech audiometric scores recorded from
the original audiograms obtained by the investigator of
the present study.

Better test results were the subjects' pure—tone
and speech audiometric scores recorded by improving the
thresholds by 10 dB with respect to their original audio-
grams. In other words, the pure-tone audiograms were
recorded as 10 dB better at each frequency tested by air
and bone-conduction. Also, speech audiometric scores
were recorded by improving the speech reception thresholds
by 10 dB and increasing the speech discrimination scores
by ten percent. However, when a subject had a Speech
discrimination score of 92% or better, the eXperimental
condition of better than actual test results could not be
fully employed, because it was not possible to make this
score better by 10 percent. Therefore, when a subject
had a speech discrimination score as just described, it
was made better by the maximum percentage. In other
words, speech discrimination scores ranging from 98% to
90% were changed to 100% for the condition of better than

actual test results.

 

49

Poorer test results were the subjects' pure-tone
and speech reception thresholds that were recorded poorer
by 10 dB with respect to their original audiograms. Also,
speech discrimination scores were recorded ten percent
poorer than those originally obtained.

No test results mean that the subjects' original
pure-tone and speech audiometric scores were not shown
to the testers. In other words, in this condition the
testers were not provided with any previous audiometric
data.

The following figure and table illustrate how
each of the conditions were recorded. All conditions
were recorded with respect to the subjtects' original
audiograms. These illustrations are representative of
the test results the testers received prior to testing
each subject.

Figure 1 shows the right ear air—conduction
thresholds for one subject according to the conditions of
actual, better, and poorer test results. The same pro-
cedure was used for recording bone—conduction thresholds.
The audiograms given to the testers at the beginning of
each test session showed right and left ear air and bone-
conduction thresholds plotted according to one of the
three conditions: actual, better, or poorer.

Table I is an example of the three possible condi-
tions that could be given to the tester at the beginning

of the test session.

 

 

   

50

FIGURE l.--Illustrative Audiogram containing the condi-
tions of (A) actual test results, (B) better
test results, and (P) poorer test results for
the right ear air-conduction thresholds for
one subject.

—1
m 0
'0

g: 10
-r-l

H 20
8 3o
3:;

m 40
§3 50
a a 60
2’.” 7o
.13 (D
931 80
8‘ 9o
'2
m 100
513’ 110

 

250 500 1000 2000 4000 8000

Frequency in Hertz

TABLE I.--Speech audiometric results with respect to the
conditions of actual, better, and poorer test
results.

 

Conditions Actual Better Poorer

 

Speech Reception
Threshold 30 dB 20 dB 40 dB

PB Score 80% 90% 70%

 

Testers

The testers were divided into two groups accord-

ing to their level of sophistication or amount of training

 

 

 

51

and eXperience in audiology. One of the groups consisted
Of four pre-doctoral students majoring in audiology at
Michigan State University. They were all males ranging
in age from twenty-three to thirty—five years with a mean
age Of twenty-eight years. Each of these students had
successfully completed the Master's Degree in the area of
audiology and was judged to be a SOphisticated clinical
audiologist by three professors Of audiology at Michigan
State University.

The second group was composed Of four undergraduate
female students who were majoring in audiology and speech
sciences at Michigan State University. Each of these
students was enrolled in an introductory audiology course.
They ranged in age from nineteen to twenty-one years with
a mean age of 20.2 years. Each of them had administered
only four pure-tone air and bone-conduction audiometric
tests prior to participating in this investigation. Each
was observed by the investigator of this study while ad-
ministering four practice audiometric tests prior to
participating in this study and were judged to be suffi-
ciently competent in basic skills necessary for partici-
pating in the eXperiment. Thus, these testers had minimal
training as Opposed to the four graduate students who

were experienced audiologists.

 

52

Subjects

All subjects were contacted by the investigator,
first by mail and then by telephone, prior to partici-
pating in the present investigation. A COpy Of the letter
sent to the subjects is in Appendix I. Each subject vol-
unteered to participate and received no monetary reward
for his participation.

Both adults and children were employed as subjects.
The adult subjects consisted of eight normal hearing indi-
viduals, sixteen individuals with sensorineural hearing
disorders, and sixteen persons with conductive hearing
disorders. The sample of children consisted of eight
normal hearing individuals with socially acceptable arti-
culation and eight normal hearing individuals with faulty
articulation. The subjects are described in greater de-

tail below under each of the four experiments.

Experimental Design

 

The design of each of the four experiments is
given below. The details of the procedures employed for
all four experiments, however, are reported later in the

chapter.

Experiment I

Conditions.——The four audiological conditions

were as follows: actual test results, better than actual

 

IE(

Ex;

inE

of-h
hEar;
and c
Subj e

W e:

 

 

 

53

test results, poorer than actual test results, and no
test results.

Testers.--Four sophisticated clinical audiologists
were employed as testers.

Subjects.--The subjects were eight normal hearing
adults ranging in age from twenty to twenty-five years
with a mean age of 21.4 years and a median age of 20.5
years. All but one of the subjects were female, and all
had socially acceptable articulation.

Tests Administered.——One ear Of each subject was

 

tested by each tester by pure-tone air—conduction audio-
metry at the following frequencies: 250, 500, 1000, 2000,
4000, and 8000 Hz. This was followed by Obtaining a speech

reception threshold on the same ear.

Experiment II

Conditions.--The same four conditions were employed
in this experiment as those in Experiment I.

Testers.—-The testers were the same as those used
in Experiment I.

Subjects.--The subjects consisted of sixteen hard-
Of-hearing adults. Eight subjects had a sensorineural
hearing loss that had been diagnosed both audiologically
and otologically. A criterion for selection of these
subjects was that their speech discrimination scores did

not exceed 88% at a 26 dB sensation level re their speech

 

 

 

ye.
32.
dis
per:
clos

word

eXper
the f
testee
aUdior
Crimin

tESted

3xPerin

”ere th.

 

54

reception threshold. Six of the subjects were male and
two were female ranging in age from forty to sixty—three
years with a mean age of 47.0 years and a median age of
43.5 years.

The remaining eight subjects, six females and two
males, exhibited a conductive hearing impairment that had
been diagnosed both audiologically and otologically.

These subjects ranged in age from eighteen to fifty-two
years with a mean age of 35.6 years and a median age of
32.0 years. No limitations were placed on these subjects'
discrimination scores for inclusion in the study, since
persons with conductive hearing losses generally score
close to 100% when phonetically balanced monosyllabic
words are presented at a 26 dB sensation level.

Tests Administered.-—Each of the subjects in this
experiment was tested bilaterally by air conduction at
the frequencies Of 500, 1000, and 2000 Hz and one ear was
tested by bone conduction at the same frequencies. Speech
audiometry (speech reception thresholds and speech dis-
crimination) was administered to the same ear as was

tested by bone conduction.

Experiment III

Conditions.--The conditions in this experiment
were the same four as those utilized in Experiments I and

II.

 

 

 

55

Testers.--The same four SOphisticated testers
participated in this experiment as in EXperiments I and II.

Subjects.--The subjects were sixteen normal hear-

 

ing children. Eight of these children, five females and
three males, ranged in age from five to seven years with
a mean age of 5.8 years and a median age Of 5.0 years.
These subjects were judged to have normal speech articu-
lation. The remaining eight, six males and two females,
ranged in age from five to nine years with a mean age of
6.3 years and a median age of 6.0 years. These children
had defective articulation characterized by substitutions,
omissions, and/or distortions. The diagnosis of defective
articulation was made by speech pathologists at the
Michigan State University Speech and Hearing Clinic.

Tests Administered.--Each of these subjects was

 

given a pure-tone air-conduction test at 500, 1000, and
2000 Hz for one ear and speech audiometry (speech recep-
tion threshold and speech discrimination) on the same ear

as the pure-tone thresholds.

Experiment IV

Conditions.--The conditions used in this eXperiment

 

were the same four as those employed in Experiments I, II,
and III.
Testers.--The testers were four undergraduate fe-

male students who were majoring in audiology and Speech

 

 

 

56

science at Michigan State University and who were taking
their first couse in audiology. Thus, they were unSOphis-
ticated testers in comparison to the pre-doctoral students.

Subjects.—-Sixteen hearing impaired adults served

 

as subjects. Although these subjects were different per-
sons from those who participated in EXperiment II, the
criteria for their selection were the same. However, no
limitations were placed on these subjects' speech dis-
crimination scores since this audiometric test was not
given to these individuals. Eight of the subjects, two
females and six males, were diagnosed as having a sensor-
ineural hearing loss. They ranged in age from thirty-
seven to sixty-five years with a mean age of 50.3 years
and a median age of 49.5 years. Eight Of the subjects
exhibited a conductive hearing impairment. All of these
subjects were females with the exception of one, and they
ranged in age from sixteen to thirty-five years with a
mean age Of 31.4 years and a median age of 30.0 years.

Tests Administered.—-Each subject was tested by

 

pure-tone air-conduction audiometry at 250, 500, 1000,
2000, and 4000 Hz bilaterally and by pure-tone bone-

conduction audiometry bilaterally at the same frequencies.

Equipment

 

The audiometric test equipment and calibration

equipment used in this study are listed below.

 

57

Test Equipment

Pure Tone Clinical Audiometer (Beltone, model
lS-C)

20 decibel Attenuation Pad (Constructed at Michigan
State University)

Earphones (Telephonics, model TDH-39)

Earphone Cushions (model MX 4l/AR)

Bone Vibrator (Radioear, model B70-A White Dot)

Narrow Band Masking Unit (Beltone, model NB-102)

Speech Audiometer (Grason-Stadler, model 162)

Commercial Test Room (Industrial Acoustic Company,
Inc. 1200 Series)

Tape Recorder (Ampex, model 601)

Calibration Equipment
Sound Level Meter (Bruel & Kjaer, type 4152)
Octave Band Filter Network (Bruel & Kjaer, type
1613)

Artificial Ear (Bruel & Kjaer, type 4152)

Condenser MicrOphone (Bruel & Kjaer, type 4132,

used in conjunction with the artificial ear)

Artificial Mastoid (Beltone, model M5A)

Voltmeter placed in the Audio Frequency Specto-

meter (Bruel & Kjaer, type 2112)

The equipment was located in the Michigan State
University Speech and Hearing Clinic. All audiometric
assessments were obtained with the subject seated in the
IAC sound-treated room. The ambient noise level in this
room was 42 dB SPL as measured on the C scale of the
sound level meter. An octave band analysis of the ambi-
ent noise in this room was conducted, and it was found
that the greatest amount Of ambient noise (40 dB average)
was in the octave below 125 Hz. The average ambient noise

for octave bands between 125 and 8000 was 14 dB SPL.1

 

1F. Harry Tokay, "Validity and Reliability of
Berkesy Audiometry with Preschool Age Children," (Unpub-
lished Doctoral Thesis, Michigan State University, 1967).

    

l . '
- n W'.2t‘ . ' I
.r ‘ - a. -

9

/‘ I /

_‘ I, . l
"I! \ - ‘ v j.“ I .. ” . F
i. 3? .- a P ' if
‘ . ‘ __r $1 * " 4,.

t" oar;— *-

 

58

Adjacent to the sound—treated IAC room was a
control room that contained the audiometric equipment
used in all testing. This room was a separate structure
and was not pre—fabricated. A window and a talk-back
system were located between the two rooms that enabled
visual and auditory communication between the tester and
the subject during test sessions. However, in the present
study the window was covered to eliminate the possibility
of the testers Obtaining visual clues from the subjects.
This procedure was followed to control for the possibility
that the tester may have had some acquaintance with the
subject. This acquaintance could influence significantly
the behavior of the subject or tester or both.1

On the following page is a block diagram Of the
test equipment used in this study.

A pure-tone audiometer that drove a matched pair
of TDH-39-10Z earphones housed in MX 4l/AR biscuit-type
cushions or a hearing aid—type bone vibrator was used for
all pure-tone testing. An auxillary attenuator was con-
structed, and it was found that employment Of this pad
reduced the minimum output (maximum attenuation) from 0
to —20 dB hearing level re ISO 1964. The response of this
pad was applicable for both air and bone-conducted stimuli

and was accurate within plus or minus one dB.

 

lRosenthal, EXperimenter Effects in Behavioral
Research, p. 87.

 

 

 

59

FIGURE 2.--Block diagram of the audiometric equipment used
to make audiometric assessments in this study.

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Earphone Bone Vibrator Earphones
Microphone 6féh/’
R L B R L
A
20 d3
Pad
LL
Beltone L
NB_ Beltone Grason-Stadler
Masking R 15c 162
Unit \
Ampex
601
Speaker used in
monitoring subject's
CONTROL ROOM responses

 

 

 

The air—conduction system of the pure—tone audio-
meter was checked with the artificial ear assembly employed
in conjunction with the sound level meter and the asso-
ciated octave band filter network. These physical measure—

ments were made according to the procedures outlined by

 

 

 

 

 

60

the American Standards Association.1 The linearity of
this system was determined by using the same calibrating
equipment.

The stability Of the bone—conduction system of
the pure-tone audiometer was periodically checked by
following the calibration procedures described in the
Operating Manual for the Beltone Artificial Mastoid.2
A voltmeter and the artificial mastoid assembly were em—
ployed when making the physical measurements. The output
voltage was converted to dB re 1 dyne and was compared to
the "HAIC Interim Bone Calibration."3

All Speech audiometry scores were Obtained by
using a speech audiometer that was capable Of amplifying
and attenuating the output of the tape recorder used to
present the tape recorded Speech stimuli. For this study,
the Speech audiometer drove a matched pair of TDH-39-1OZ
earphones housed in MX 4l/AR cushions. This audiometer
was calibrated so that audiometric zero was equivalent

to 20 dB re 0.0002 dyne/cmz.

 

l . "American Standard Specifications for
Audiometers for General Diagnostic Purposes," American
Standard Association, Incorporated (1960), p. 9.

2 . Operating Manual for the Beltone Arti-
ficial Mastoid, M5 Series, Chicago, Illinois, Beltone
Electronics Corporation.

3S. F. Lybarger, "Interim Bond Conduction Thresholds
for Audiometry," Journal of Speech and Hearing Research, 9
(1966). pp. 483-487.

 

 

 

   

61

Calibration of this system was conducted by using
"Speech Noise" instead of a 1000 Hz pure tone as described
by the American Standards Association. Tillman, Johnson,
and Olsen1 reported that the Spectral configuration Of
"Speech Noise" is closer to the spectrum Of speech pro-
duced by male speakers than is the 1000 Hz pure tone.
The calibration equipment used to check this system was
the same as that employed to calibrate the air-conduction
system of the pure-tone audiometer. The "Speech Noise"
was channeled into the sound level meter assembly at a
60 dB attenuator setting and with a deflection Of zero on
the Speech audiometer VU meter. The acoustic output of
the system was measured and this was considered to be'
equivalent to the intensity of the spondee words at the
same attenuator setting when the peaks Of the words pro-

duced a deflection of zero on the VU meter.‘

Recorded Speech Materials

 

All speech stimuli were recorded by an experienced
male speaker. The spondee words were taken from the CID
Auditory Test W—l word lists and were recorded on high

quality magnetic tape by a local male talker having

 

lTom W. Tillman, Robert M. Johnson, and Wayne 0.
Olsen, "Earphone versus Sound Field Threshold Sound-
Pressure Levels for Spondee Words," Journal of the
Acoustical Society of America, 39 (1966), pp. 125-133.

 

 

 

62

General American speech. The tape consisted of six
randomizations of thirty-six spondee words, namely, CID
Auditory Test W-l Lists A, B, C, D, E, and F.1 Each list
was produced by re-recording from a master list and
cutting and splicing the words into the appropriate order
to construct the six randomizations. Each word on the
tape was monitored by the speaker on a VU meter so that
each syllable of each word peaked at the same intensity
level as the 1000 Hz calibration tone that had been re-
corded at the beginning Of the tape. Both syllables of
each word were monitored by the speaker to a level Of plus
or minus one decible re the calibration tone.

The speech discrimination materials were taken
from two sources and recorded by the same male speaker
who recorded the spondee words.

Speech discrimination stimuli for the adult sub-

2 The

jects were taken from the NU Auditory Test NO. 6.
speech discrimination stimuli for the children were taken

from the Phonetically Balanced Kindergarten lists.3

 

lHirsh, gt $1., "Development of Materials for
Speech Audiometry," pp. 321-337.

2Tillman and Carhart, "An Expanded Test for Speech
Discrimination," pp. 1—12.

3Haskins, "Speech Discrimination for Children."

 

 

63

The NU Auditory Test NO. 6 consists of four lists
Of fifty monosyllabic words patterned after the CNC lists

1’ These

develOped and revised by Lehiste and Peterson.
lists were recorded on magnetic tape. The speaker used
the carrier phrase, "You will say. . ." before enunciating
one Of the monosyllabic words. The speaker monitored his
vocal production for these words by monitoring his vocal
output for the word "SAY" of the carrier phrase to zero
dB on the VU meter. Again, all monitoring was done by
means of Observing the deflection of the needle Of the VU
meter re 1000 Hz calibration tone that had been recorded
at the beginning Of the tape. Four additional lists,
necessary for the eXperimental conditions in the present
investigation, were made from scramblings of the original
four lists. The scramblings were achieved in the same
manner as eXplained for the randomizations of the spondee
words.

The four CNC lists were standardized earlier by

Rintelmann and Jetty3 on ten young adult subjects with

normal hearing following the procedures outlined by

 

lI Lehiste and G. E. Peterson, "Linguistic Consid—
erations in the Study of Speech Intelligibility," Journal
Of the Acoustical Society of America, 31 (1959), pp. 280-286.

2G. E. Peterson and I. Lehiste, "Revised CNC Lists
for Auditory Tests," Journal of Speech and Hearing Dis-
orders, 27 (1962), pp. 62-70.

3Rintelmann and Jetty, "Reliability Of Speech
Discrimination Testing."

 

 

 

v
\
l
1
1

PD:

DiSC

 

64

Tillman and Carhart.l The results were essentially the
same as those Obtained by Tillman and Carhart, and the
lists were equivalent plus or minus four percent at a 24
dB sensation level.

The speech discrimination materials for the
children consisted of four randomizations Of List 1 of
the PBK Word Lists. These words were recorded and random-
ized in the same manner as were the monosyllabic words

used for speech discrimination testing Of adults.

Procedures

Each subject was tested audiometrically five times.
The first examination was conducted by the investigator,
whereas the remaining sessions were conducted by four
student testers. The subject was seated in a comfortable
chair located in the IAC sound—treated room. The investi-
gator gave each subject the following instructions prior
tO testing him by pure—tone air-conduction audiometry:

You are now going to hear a tone. First, all Of
the tones will be in your right ear and then all of
the tones will be in your left ear. When you hear
the tone please push this button. Some of the tones
will be very faint, but push the button whenever you
think you hear the tone. Are there any questions?

Each Of the fifty-six subjects was first given a

pure-tone air-conduction test bilaterally by the

 

lTillman and Carhart, "An Expanded Test for Speech
Discrimination," pp. 1-12.

 

 

 

the

gen

 

65

investigator. This was accomplished by using the Hughson-
Westlake ascending technique.l Testing started at 1000
Hz and the subject's threshold was determined for that
frequency. Threshold was established when the subject
responded at the lowest hearing level three times. The
investigator followed the same procedures for 2000, 4000,
and 8000 Hz. The threshold at 1000 Hz was rechecked
before establishing thresholds at 250 and 500 Hz. Upon
completion of the air-conduction testing, bone-conduction
thresholds were Obtained at octave intervals between 250
and 4000 Hz. The investigator followed the same testing
techniques for measuring bone-conduction thresholds as he
did for establishing air-conduction thresholds; however,
masking was used routinely during bone-conduction testing.
The following instructions were given to each subject be-
fore the pure-tone bone-conduction test was administered.
You are now going to hear the same kind Of tone
through this apparatus; however, you will hear a noise
in this ear (pointing to the subject's ear that would
receive the masking agent). Try not to pay attention
to the noise, but rather listen very carefully for
the tone. When you hear the tone please push the
button just as you have been doing. Are there any
questions?
A narrow band masking unit was utilized to produce

the masking stimulus. Previous measurements of this noise

generator indicated that the band widths, determined at a

 

lCarhart and Jerger, "Preferred Method for Deter-
mination Of Thresholds," pp. 330—345.

 

66

level 3 dB down from the peak intensity, were greater than
the critical band widths defined by Fletcher.l’ 2 Effec—
tive masking for zero dB hearing level was determined at
each band by applying the critical band data Of Fletcher
in a manner outlined by Sanders and Rintelmann.3

The Hood Technique4 was employed to establish
masked thresholds. The procedures for this technique are
as follows:

Find the bone-conduction threshold with the bone—
conductor applied to the mastoid Of the tested ear
without masking Of the untest ear. Apply the masking
noise of the appropriate band to the untest ear by
means Of air conduction and find a bone-conduction
threshold reading. Increase the level Of masking
noise by 10 dB above threshold and retest the bone-
conduction threshold. If the bone conduction
threshold is raised by 10 dB increase the masking in—
tensity by another 10 dB and repeat. Continue this
procedure until the point is reached at which the
bone conduction remains constant with further addi-
tional incremental steps Of 10 dB Of masking noise.
This is the "change-over point" and gives the true
bone—conduction threshold of the tested ear.

 

lHarvey Fletcher and W. A. Munson, "Relation
Between Loudness and Masking," Journal of the Acoustical
Society of America, 9 (1937), pp. 1—10.

 

 

2Harvey Fletcher, "Auditory Patterns," Review of
Modern Physics, 12 (1940), pp. 47-65.

3Jay W. Sanders and William F. Rintelmann,
"Masking in Audiometry," Archives of Otolaryngology, 80
(1964), PP. 541—556.

 

4J. D. Hood, "Principles and Practice Of Bone
Conduction Audiometry," Laryngoscope, 70 (1960), p. 1215.

 

67

Immediately following the pure-tone air and
bone-conduction testing, forty Of the subjects (Experi-
ments I, II, and III) were given speech reception
threshold tests and thirty-two subjects (Experiments II
and III) were given speech discrimination tests. All
speech audiometric scores were Obtained monaurally, but
both ears were tested for all subjects. Each subject
was given the following instructions before the administra—
tion Of the spondee words used to establish the subjects'
speech reception thresholds.

You are now going to listen to a number Of two-
syllable words, for example, horseshoe, hothouse,
northwest, etc. You are to repeat the words that you
hear. You will hear thirty—six of these words at a
comfortable level and you are to repeat them. Follow-
ing this you will again listen to the same words;
however, they will be in a different order than the
first list you heard. Again you are to repeat these
words which will get fainter and fainter. If you are
not sure of the word please guess. You will hear the
words in your right ear first and then in your left
ear. Are there any questions?

The subject then heard and responded to an entire
list of spondee words presented at a suprathreshold level
that was comfortable for him. This procedure was based
on the findings of Tillman and Jerger.1 These authors

reported that the accuracy Of the speech reception

threshold is a function of word familiarity; therefore,

 

lTillman and Jerger, "Some Factors Affecting
Spondee Threshold," pp. 141-146.

 

 

 

68

if the listener has the opportunity Of hearing all the
words before his speech reception threshold is recorded,
he will tend to improve the preciseness Of his scores for
this test. Following this procedure, the subject was then
presented one Of the taped spondee lists. An ascending
technique was employed to establish threshold; however,
spondees were presented in 2-dB steps instead of 5-dB
steps as were the pure tones. A subject's speech recep-
tion threshold was established when he correctly responded
to four out Of eight spondee words at the lowest intensity
level.

Following Speech reception threshold measurements,
specific subjects (the thirty—two participating in Experi-
ments II and III) were given speech discrimination tests.
Prior to obtaining the speech discrimination scores the
following instructions were given to each subject.

You are not going to listen to some one—syllable
words. These words will be very short and you will
have to listen very carefully. Again, please guess
if you are not sure Of the word; however, only say
the word you hear or think you hear. If you talk or
explain that you cannot understand the words, it is
likely to cause you to miss the following word. Are
there any questions?

The adult subjects were administered fifty tape

recorded one-syllable phonetically balanced words from
the NU Auditory Test NO. 6. Children were given fifty
monosyllabic words taken from the PBK Word lists. Both

were presented at a 26 dB sensation level re the subjects'

speech reception thresholds.

 

69

The number of audiometric tests a subject received
was determined by the experiment in which he participated.
When the subject had been given the necessary number Of
audiometric tests by the investigator, he was instructed
when to return for additional testing. However, in each
experiment except Experiment III, all subjects were seen
by the testers immediately following the investigator's
audiometric evaluation. In experiment III where children
were used as subjects, the investigator conducted the
initial evaluation on a different day than did the testers.
However, each of these subjects was seen by all four
testers within forty-eight hours after the initial evalua-
tion.

Eight testers were selected to participate in
four separate experiments in this study. The testers
were divided into two groups. These categories were
specifically defined earlier in this chapter, but to
briefly reiterate, one group Of testers consisted of four
pre-doctoral students and the other group was composed of
four undergraduate students enrolled in an introductory
audiology course.

Prior to their examination Of the subjects, the
testers were given specific instructions. These instruc-
tions were read to the group of testers who were going to
participate in a particular experiment. This procedure

was followed in order that each tester head the same

 

70

instructions at the same time; therefore, each one heard
the same intonations, pauses, repetitions, and verbal
clues that were unintentionally read by the investigator.
Following are the instructions given to the testers.

These instructions are stated in their entirety. However,
depending on the experiment, some of the instructions were
deleted to meet the specific needs Of each experiment.

You are going to administer a pure-tone air or
bone-conduction test(s) or both to eight (or sixteen)
subjects. You will use the Carhart-Jerger modified
method of the original Hughson-Westlake ascending
technique to establish thresholds. You will test one
or both ears of each subject. This will be specified
to you just prior to testing the subject. In Obtain-
ing pure-tone thresholds, you will start at 1000 Hz
at suprathreshold and attenuate in 10 dB steps until
the subject fails to respond, at this point you will
increase the intensity in 5 dB steps until the subject
again responds. Following this procedure you will
record the subject's threshold as the lowest intensity
that he responds to three times. You will follow this
same procedure for the following frequencies: 250,
500, recheck 1000, 2000, 4000, and 8000 Hz.

You are to Obtain bone-conduction thresholds in
the same manner for the same frequencies as by air-
conduction except 8000 Hz. If masking is deemed
necessary according to Studebaker's criteria, you will
use narrow band white noise and the Hood Technique to
establish thresholds. You will increase the masking
stimulus 15 dB re effective masking beyond the point
where the subject's masked threshold fails to shift.

You will also administer speech audiometry to the
subjects. You will obtain a subject's speech recep-
tion threshold or speech discrimination score or both
on one ear. When administering the spondees you will
present an entire list Of CID W-l words at supre-
threshold (comfortable to the subject). You will
begin the test following the same procedures as you
did in establishing pure-tone thresholds except when
ascending you will ascend in 2-dB steps. Threshold
will be established when a subject correctly responds
to four out of eight words at the lowest intensity
level.

 

 

 

 

Vi!
thr
rec
mak
in 1
ity
tion

this

 

71

When determining a subject's speech discrimination
score, you will present a list of fifty monosyllabic
words at a 26 dB sensation level re the subject's
speech reception threshold.

As was previously stated, you will be testing
eight (or sixteen) subjects. The subjects may or may
not have an accompanying audiogram. When the subject
does have an audiogram you are to study it carefully.
You are to replicate the tests that are reported on
the audiogram. If you do not receive an audiogram
you will be told which tests to administer to which
ear.

You are not to discuss your results among your-
selves. I will be in the testing area to give the
subjects instructions, change transducers, and pre-
pare tapes. Are there any questions?

Each tester was given one Of the previously
described audiograms (actual, better, poorer, or no informa—
tion) just prior to testing each subject. Each tester had
ample time to review the information contained on the audio-
grams while the investigator seated the subject in the
soundtreated room IAC room. The investigator gave each
subject the same instructions he had given him during the
initial test session and also placed the earphones or bone
vibrator on the subject. This procedure was followed for
three basic reasons: (1) to make sure that each subject
received the same instructions and information, (2) to
make sure the transducers were correctly placed and always
in the same position, and (3) to control for the possibil—
ity of the subject's giving the tester additional informa-
tion causing him to alter his bias. Rosenthal classified
this type of design as a "blind experiment." He wrote:

. . .if the experimenter did not know the hypo-

thesis being tested but did know to which group each
subject belonged, the results of the study are more

 

 

 

 

 

 

0f

inf

sen

0f

fol.

bone

disc

6
R

 

72

likely to be biased in the direction Of the hypothesis
(or Opposite to it) rather than biased irrelevantly
with respect to the hypothesis.

The investigator remained in the control room
while the tester conducted the evaluation; however, his
only function was to determine when the transducers needed
changing, when different instructions were necessary for
the subject, and when the tapes containing the speech
stimuli needed to be changed.

By following these procedures, each subject in
all four experiments was tested five times. He was tested
by the investigator first and then by each tester under
one of four conditions. Therefore, in each experiment
(sub-groups Of eight subjects) each tester obtained audio-
metric scores on two subjects under each of the four
conditions. In each experiment a counter-balanced order
of condition presentation (actual, better, poorer, and no
information) was employed so that each condition was pre-
sented first, second, third, and fourth an equal number
of times. All audiometric tests were presented in the
following order: pure-tone air-conduction, pure—tone
bone-conduction, speech reception threshold, and speech

discrimination.

 

lRosenthal, Experimenter Effects in Behavioral
Research, p. 370.

 

 

 

 

 

be

dis

tes
fra
bone
metr

tion

 

73

Summary of Experiments

Experiment I

Testers.--Four SOphisticated pre-doctoral students
participated as testers.

Subjects.-—Eight normal hearing adults were em-

 

ployed as subjects.

Tests Administered.--Each subject was tested uni-

 

laterally by air conduction at the following frequencies:
250, 500, 1000, 2000, 4000, and 8000 Hz. This was followed

by Obtaining a speech reception threshold on the same ear.

Experiment II

Testers.--The testers were the same four as those
used in Experiment I.

Subjects.-—Sixteen hearing impaired adults served

 

as subjects. Eight of these subjects had a conductive
hearing impairment and eight had a sensorineural hearing
disorder.

Tests Administered.--Each Of these subjects was

 

tested monaurally in both ears by air conduction at the
frequencies of 500, 1000, and 2000 Hz and unilaterally by
bone conduction at the same frequencies. Speech audio-
metry (speech reception thresholds and speech discrimina-
tion) was administered to the same ear as the ear tested

by bone conduction.

/"

/

. " g 1‘7

./ - .
FP‘ ii. A
I II . y ,

 

74

EXperiment III

Testers.--The same four sophisticated testers
participated in this eXperiment as in Experiments I and
II.

Subjects.--Sixteen normal hearing children parti-
cipated as subjects. Eight of these children had socially
acceptable articulation and eight had defective articula-
tion.

Tests Administered.--Each subject was given a

 

pure-tone air-conduction test at 500, 1000, and 2000 Hz
on a single ear and speech audiometry, (speech reception
threshold and speech discrimination) was administered on

the same ear as the pure-tone thresholds.

Experiment IV

Testers.--Four undergraduate students majoring in
speech and hearing therapy and enrolled in an introductory
audiology course were employed as testers.

Subjects.--Sixteen hard-Of-hearing adults served
as subjects. Eight of these individuals had a conductive
hearing loss and eight had a sensorineural hearing disorder.

Tests Administered.--Each subject was tested by

 

pure-tone air-conduction audiometry at 250, 500, 1000,
2000, and 4000 Hz monaurally in both ears by pure-tone

air and bone-conduction.

 

CHAPTER IV

RESULTS AND DISCUSSION

This chapter is divided into two basic portions.
The first part presents the results of the clinical Sig-
nificance between pure-tone and speech audiometry scores.
The second part gives the statistical significance between
mean pure-tone and speech audiometry scores. Each Of
these major divisions is sub-divided into four sections-—
the four experiments described in Chapter III of this
investigation. The primary results Obtained for each
experiment are reported followed by a discussion of these
results.

One method Of showing the relationship between
testers, conditions, and stimuli employed in this study
is by examining the means, standard deviations, standard
errors Of measurement, and ranges Of the audiometric
scores Obtained in each eXperiment.

Another method Of Observing this relationship is
by constructing composite audiograms showing the mean
pure-tone air and bone-conduction thresholds.

The tables in this chapter present the means,

standard deviations, standard errors of measurement and

75

   

76

ranges of the stimuli used in this investigation, while
the figures show composite audiograms of the mean pure-
tone air and bone-conduction thresholds obtained in the

present study.

Experiment I

 

In this eXperiment four pre-doctoral students in
audiology participated as testers. Each of them tested
eight normal hearing adults under one Of four conditions.
The conditions were as follows: actual test results
(actual), better than actual test results (better), poorer
than actual test results (poorer), and no test results
(no info.). Each tester obtained pure-tone air-conduction
thresholds on a single ear for the following frequencies:
250, 500, 1000, 2000, 4000, and 8000 Hz. They also
measured Speech reception thresholds on the same ear that

pure-tone thresholds were Obtained.

Clinical Significance

Pure-Tone Air-Conduction Thresholds.--Table II

 

shows the means, standard deviations, standard errors of
measurement, and ranges Of the pure-tone air-conduction
thresholds recorded for eight normal hearing adults by
four pre-doctoral students in audiology under four condi-
tions. This table also shows the results (original)
Obtained for these subjects by the investigator of this

study.

 

 

77

 

TABLE II.--Means (M), Standard Deviations (SD), Standard
Errors Of Measurement (SEm), and Ranges of Pure-
Tone Air—Conduction Thresholds* Obtained for
Eight Normal Hearing Adults by the Investigator
and Four Pre-Doctoral Students Under Four Condi-

 

 

 

tions.
Frequency in Hertz
Condition
250 500 1000 2000 4000 8000

Original

M 3.75 5.0 -3.75 -2.50 1.87 8.12

SD 4.84 5.0 3.3 3.5 3.47 8.99

Range 15.00 15.00 10.00 10.00 5.00 25.00
Actual

M 2.50 1.87 -5.0 -5.0 3.75 2.50

SD 4.33 4.28 3.53 6.12 4.14 7.50

SEm 2.50 3.95 3.53 4.33 3.95 9.18

Range 15.00 15.00 10.00 15.00 15.00 25.00
Better

M 3.75 1.87 -5.62 -5.00 2.50 4.37

SD 4.14 4.96 4.63 6.12 5.00 6.34

SEm 2.50 3.95 3.95 5.00 4.63 7.07

Range 10.00 15.00 15.00 20.00 15.00 20.00
Poorer

M 3.75 3.12 -3.75 -3.12 4.37 4.37

SD 4.14 4.96 4.14 5.55 3.90 10.39

SEm 3.53 3.95 3.53 3.06 4.33 6.12

Range 10.00 15.00 15.00 15.00 10.00 30.00
NO Info.

M 0.62 1.87 -5.62 -6.25 3.75 5.00

SD 4.63 4.28 3.90 4.14 4.84 8.66

SEm 3.95 3.95 3.95 5.00 3.95 7.28

Range 15.00 15.00 10.00 10.00 15.00 25.00

 

*In dB (re ISO 1964 standards)

Examination Of Table II reveals that there is very
little difference between means for the pure-tone air-

conduction thresholds obtained under the four conditions

 

 

78

at any frequency tested. Further, for this study the
standard errors of measurement (SEm) provides the most
meaningful information. The SEm is an estimate of the
standard deviation that would be Obtained for a series of
measurements on the same individual.1

Table II shows that the SEm for all frequencies
under all conditions is in good agreement with the data
that has been published on the reliability of pure-tone.
air-conduction audiometry.2' 3 Test re-test thresholds
that are within plus or minus 5 dB are considered to show
accaptable clinical reliability. Thus, first, in compar-
ing conditions (actual, better, poorer, and no info.) the
reliability of all testers was good from condition to con-
dition at all frequencies tested. Second, the small SEm
at each condition for all frequencies except 8000 Hz shows
that the reliability between testers was good for each
condition. Therefore, the high repeatability, or lack Of
variability shown in Table II at all frequencies except
8000 Hz indicates that previous audiometric information did
did not significantly cause SOphisticated audiologists to

influence normal hearing adults' pure-tone air-conduction

 

1Jerger, "Evaluation of Some Auditory Measures,"

2Ibid.

 

3Witting and Hughson, "Inherent Accuracy of
Audiograms," pp. 259-269.

 

79

thresholds. In other words, according to these results,
it seems that previous audiometric information did not
cause SOphisticated audiologists to exhibit information
did not cause SOphisticated audiologists to exhibit a
bias that clinically affected the reliability or repeat-
ability of pure-tone air-conduction audiometry.

Figure 3 shows a composite audiogram of the mean
air-conduction thresholds of the aforementioned eight
normal hearing adults Obtained by the four testers under
the four conditions and the original mean thresholds Ob-

tained by the investigator.

FIGURE 3.--Mean Air-Conduction Thresholds for Eight Normal
Hearing Adults Obtained by the Investigator and
Four Pre-Doctoral Students Under Four Conditions.

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

-10
2;
\0
O'\
H
o -5
m
1'1
m
O
a 0
-a
r4
0
>
O
A 5
m
a
-H
H
m
m
m

10

250 500 1000 2000 4000 8000
' Frequency in Hertz

Original . . Actual x---x 'Better A—--~-A Poorer CPH-D

NO Info. V-H-V

 

 

 

'5?"

 

”I

ll;

 

 

80

The above figure illustrates the high degree Of
repeatability Obtained for the mean air-conduction
thresholds for six frequencies under four test conditions.
Again, according to reports in the literature on clinical
reliability, none of the above means exceed the acceptable
amount of error, 5 dB, for pure-tone audiometry.

Speech Reception Thresholds.--The means, standard
deviations, standard errors of measurement, and ranges of
the speech reception thresholds Obtained for eight normal
hearing adults by four SOphisticated testers under four
conditions are presented in Table III. The original SRTs

obtained for these subjects by the investigator are also

reported in this table.

TABLE III.--Means (M), Standard Deviations (SD), Standard
Errors of Measurement (SEm), and Ranges Of
Speech Reception Thresholds* Obtained for
Eight Normal Hearing Adults by the Investiga-
tor and Four Pre-Doctoral Students Under Four

 

 

 

Conditions.
SRTs Condition
Original Actual Better Poorer No Info.
M -2.75 -3.25 -4.00 -3.50 -4.25
SD 1.71 1.71 2.44 2.22 1.19
SEm 1.73 2.12 2.00 2.00
Range 4.00 4.00 8.00 2.00 4.00

 

*In dB re speech audiometric zero

12.—-

 

 

 

 

81

The data contained in Table III show that the
variability between mean speech reception thresholds under
all conditions is minimal. The SEm between the conditions
indicates that the previous audiometric information given
to each of the testers prior to their testing each subject
did not cause them to influence the subjects' speech recep-
tion thresholds significantly. The standard errors Of
measurement reveal impressively consistent results across
all conditions. The average SEm for all four conditions
was approximately 2 dB, which means that even when the
audiologist is provided either true or erroneous test re-
sults, repeated speech reception thresholds would be no
greater than plus or minus 2 dB sixty-eight percent of the
time. From the data in Table III, it would be reasonable
to eXpect that a SOphisticated audiologist would record
speech reception thresholds as clinically accurate no
matter what type of previous audiological information he
had about the subject. According to the literature on
speech reception theshold reliability, the acceptable
53m for this threshold is 6 dB.1

Therefore, the results obtained by the four pre-

doctoral students on these subjects under these conditions

 

1Joseph B. Chaiklin and Ira M. Ventry, ". . .A
Comparison Of 2- and 5- dB Methods," pp. 47-59.

 

82

£311c>vv an acceptable amount of error for speech reception
i:111:eesshold testing. As was evident in pure-tone air-

<:c>r1€111ction audiometry, the sophisticated testers were not

biased by the previous audiometric information, and they

did not influence the subjects' speech reception thresholds
significantly.

In summary, the descriptive results of this experi-
Inemrt show that all of the differences Obtained between
means were within the limits Of clinical reliability. The
average standard error of measurement for pure-tone air-
coruiuction thresholds was 4.46 dB, which is basically the
sauna as the 5 dB test—retest reliability acceptable for

pure-tone air—conduction audiometry. The average standard
error of measurement for speech reception thresholds was

1.96 dB, which is 4 dB better than the test-retest reli-
ability acceptable for spondee thresholds.

These results indicate that skilled audiologists
are not biased by previous audiometric information and,
therefore, do not clinically influence air-conduction

thresholds or speech reception thresholds Obtained from
normal hearing adults.

Statistical Significance

A number of null hypotheses were postulated to
test the differences among mean pure-tone and speech

audiometry scores Obtained in this investigation. The

 

 

83

exact number of null hypotheses tested varied from one
experiment to another according to the specific auditory
stimuli that were presented. Therefore, the null hypo-
theses postulated for each experiment are enumerated in-
dividually. The null hypotheses were tested statistically.
A presentation of the statistical technique used and the
results are reported followed by a discussion Of the sig-
nificance Of the results.

Pure-Tone Air-Conduction Thresholds.--In order to
test differences among mean pure-tone air-conduction
thresholds obtained in this eXperiment the following null
hypothesis was postulated:

There are no significant differences among mean pure-
tone air-conduction thresholds obtained on normal
hearing adults as a function Of the following vari-
ables: testers, conditions, and frequencies.

A three-factor analysis of variance with repeated
measures on one factor was employed to determine the sig-
nificance of differences among the variables in the
principle comparisons for pure-tone air-conduction audio-
metry.1 A three-dimensional table was constructed to
organize the data. In this experiment there were four

rows representing testers, six columns representing fre-

quency, and four slices representing conditions. The

 

1B. J. Winer, Statistical Principles in EXperi-
mental Desi n (New York: McGraw-Hill Book CO., 1962),

PP- - .

 

 

i l mi.l

 

84

Peated measures were on the frequencies. Thus, an A x

X C three-entry table was used to record the data.

.e analysis of variance was computed with a Control Data

btpor ation 3600 Digital Computer.1 The pure-tone air-

:onduction thresholds were the criterion measures entered

in each cell. The significance of the variations was

determined by the F-ratio.

The means used in comparing differences between

testers, conditions, and frequencies are shown in Table

II, page 77, and a summary of the analysis is given in

Appendix A.

The only variable that was statistically signifi—

cant was frequency. Levels Of confidence within testers,

within conditions, and interactions were not statistically

significant. Thus, the aforementioned null hypothesis

was rejected. However, it was expected that frequency

would show significance since research has demonstrated
that the inherent stability of human auditory thresholds

varies as a function Of frequency. In other words, a

different amount of variability occurs on repeated pure-
tone air-conduction thresholds, depending upon which fre-

quency is being presented to the listener. Thus, the

 

Michigan State University, Agricultural Experi-
mmm Station, Analysis of Variance with Equal Frequencies
inEach Cell," STAT Series Description #14 (March, 1966).

 

 

   

[-
J

85

statistical significance found for frequency can be
attributed to variations as a function of frequency for

the subjects' pure-tone air-conduction thresholds rather

than the bias Of the testers.

Speech Reception Thresholds.--In order to test
the differences among speech reception thresholds Obtained

in.tflris experiment the following null hypothesis was postu-
lated:

'There are no significant differences among the mean
speech reception thresholds Obtained on normal hear-

ing adults by four pre-doctoral students under four
test conditions.

The significance of differences among the vari-

ables in this comparison was determined by a two-factor

I l l I
analySis of variance. The two factors were arranged in

a two-dimensional table of a 4 x 4 design. The four

columns represented the test conditions, while the testers

were represented by four rows. The speech reception

thresholds obtained were the criterion measures entered

in each cell. The F-ratio was used in testing the statist-

ical significance of the variance attributable to the two
Kain effects and the two-way interaction.

The mean speech reception thresholds used in the
comparison are Shown in Table III, page 80, and a
summary of the analysis is given in Appendix A.

Winer, Statistical Principles in Experimental
Design, pp. 298-318. ,

 

 

 

 

86

The F-statistic was not significant for the two
main factors or the interaction between the two factors.

TINJS, 'the above null hypothesis could not be rejected.

The three-way analysis of variance on the pure—
tone air-conduction thresholds and the two—way analysis
of VJariance on the speech reception thresholds computed
for: this experiment indicate that SOphisticated audio-

logigsts were not influenced significantly by previous

audiometric results. These analyses illustrate that well—

trained testers obtained mean air—conduction thresholds
and speech reception thresholds that were quite comparable
across conditions with very little variance.

The two-way analysis Of variance, which did not

employ frequency as one Of the variables, showed no sig-

nificance between testers and test conditions or the in-

teraction between these. These findings give further

support to the rationale that in the three-way analysis
of variance, where frequency was the only significant
factor, the significant finding was not due to tester
bias but rather to the inherent variations between auditory
thresholds as a function Of frequency.

Based on the statistical results of this experi—
ment, it can be seen that experimenter (tester) bias was
not exhibited significantly by sophisticated audiologists
when Obtaining pure—tone air—conduction and speech recep-

tion thresholds on normal hearing adults.

 

 

 

87

Experiment II

In this eXperiment the same four pre-doctoral

students were employed as testers. Each of them tested

sixteen hard-of—hearing adults. The subjects were divided

intx> two pathological groups. Eight exhibited a sensori-

neural hearing loss and eight had a conductive hearing

disorder. Each tester Obtained pure-tone air-conduction

thresholds bilaterally and bone-conduction thresholds on

23 single ear under one of four conditions for the follow-

ing frequencies; 500, 1000, and 2000 Hz. The conditions

were the same as those described in EXperiment I. The

testers also measured speech reception thresholds and
Speech discrimination scores on the same ear that pure-

tone bone-conduction thresholds were Obtained.

Clinical Significance

Pure-Tone Air-Conduction Thresholds.--Table IV

gives the means, standard deviations, standard errors Of

measurement, and ranges of the air-conduction thresholds
Obtained by four SOphisticated audiologsits under four
test conditions for eight adults with a sensorineural
hearing loss and eight adults with a conductive hearing

impairment. The table also shows the results (original)

Obtained by the investigator for these subjects.

 

 

 

 

 

88

 

TABLE IV.--Means (M), Standard Deviations (SD), Standard

Errors of Measurement (SEm), and Ranges of
Pure-Tone Air-Conduction Thresholds* Obtained
for Eight Sensorineural (SN) Adult Subjects
and Eight Conductive (Cond.) Adult Subjects by
the Investigator and Four Pre—Doctoral Students
Under Four Conditions.

 

Frequency in Hertz

 

500 1000 2000
Condition SN Cond . SN Cond . SN Cond .
Original
M. 36.25 36.87 45.00 31.25 56.87 30.62
SD 18.15 11.43 11.98 11.65 12.73 9.16
Range 45.00 35.00 40.00 45.00 45.00 35.00
Actual
M 35.62 35.00 45.62 31.87 56.87 29.37
SD 19.11 11.18 9.49 15.19 12.97 9.82
SEm 3.06 4.67 6.08 5.86 5.59 3.53
Range 50.00 35.00 35.00 55.00 45.00 35.00
Better
M 36.25 34.37 46.87 32.50 52.50 27.50
SD 20.11 11.16 11.97 12.23 12.50 8.29
SEm 4.33 6.08 3.95 3.95 7.28 5.86
Range 50.00 40.00 40.00 45.00 40.00 30.00
Poorer
M 36.25 34.37 46.87 32.50 55.00 27.50
SD 19.84 7.68 11.97 13.56 12.50 9.68
SEm 3.53 5.30 3.06 3.95 6.87 5.30
Range 55.00 25.00 40.00 50.00 50.00 35.00
NO Info.
M 35.00 36.25 46.25 31.87 56.25 29.37
SD 17.85 7.68 11.38 13.56 14.94 9.68
SEm 3.06 5.30 3.53 4.67 3.95 2.50
Range 50.00 40.00 40.00 55.00 55.00 35.00

 

*In dB re ISO 1964

Table IV shows that the mean air-conduction

thresholds are essentially the same under all conditions

for the sensorineural subjects.

Examination of this

 

 

 

 

 

 

89

table also reveals that all SEm for all frequencies under
all conditions are within the limits of reliability for
air-conduction audiometry. The minimal variability for
all frequencies under each condition indicates clinically
that previous audiometric test result information did not
significantly influence SOphisticated audiologists when
Obtaining air-conduction thresholds for subjects with
sensorineural hearing disorders. However, it should be
noted that the least amount of variability occurred when
the testers were provided no previous audiometric informa-
tion.

The data in Table IV reveal that the mean air-
conduction thresholds and the SEm for the adults with
conductive hearing impairments are approximately the same
for all frequencies under all conditions. The repeat-
ability for the conductive subjects is essentially the
same as it is for the sensorineural subjects. Therefore,
according to these results, previous audiometric informa-
tion did not cause skilled testers to influence signifi-
cantly hard-of—hearing adults' air-conduction thresholds
regardless of the type of hearing disorder. In other
words, the data indicate that SOphisticated audiologists
Obtained air-conduction thresholds that were clinically
accurate and they were not influenced by past test results.

Figures 4 and 5 Show composite audiograms of the

mean air-conduction thresholds of the sensorineural and

 

 

 

 

 

 

90

conductive adult subjects Obtained respectively by the
four testers under four conditions and the original mean
thresholds measured by the investigator. Both figures
show'that the degree Of repeatability between the mean

air-conduction thresholds was within 5 dB or less for all

conditions.

FIGURE 4.--Mean Air-Conduction Thresholds for Eight Adults
with a Sensorineural Hearing Disorder Obtained
by the Investigator and Four Pre—Doctoral
Students Under Four Conditions.

 

 

 

 

 

 

 

 

 

35 I
,‘ Original
.w 40 ° ‘
: Actual x---—x
H
0 Better A--—~-—A
E
"’ 45 Poorer Dunn
m
'° NO Info. V—---V
:
-H
'3 50
>
m
.4
vi
a.
'3 55
m
m
:2

60

 

 

 

 

 

 

 

500 1000 2000

Frequency in Hertz

15.-s"

 

 

 

 

 

91

 

FIGURE 5.--Mean Air-Conduction Thresholds for Eight Adults
with a Conductive.Hearing Impairment Obtained
by the InvestigatOr and Four Pre-Doctoral Stu-
dents Under Four Conditions.

 

 

 

 

 

 

 

 

 

 

 

 

E 25-
ON
H
C)
3 Original. .
m 30 ' Actual x-——x I
'o
.g Better A-o-o-A
H 000 .-
g, 35 Poorer a- c
3 NO Info. V-u—V
ow
c
-H
3 40
m
m
500 1000 200)

 

 

 

 

Frequency in Hertz

Pure—Tone Bone-Conduction Thresholds.--The means,

 

standard deviations, standard errors of measurement, and
ranges Of the pure-tone bone-conduction thresholds Ob-
tained for eight adults with a sensorineural hearing loss
and eight adults with a conductive hearing impairment by
four skilled testers under four test conditions are
listed in Table V. The table also shows the results
(original) Obtained for these subjects by the investigator

of this study.

 

92

 

TABLE V.--Means (M), Standard Deviations (SD), Standard

Errors of Measurement,
Tone Bone-Conduction Thresholds* Obtained for
Eight Sensorineural (SN) Adult Subjects and
Eight Conductive (Cond.) Adult Subjects by the
Investigator and Four Pre-Doctoral Students Under
Four Conditions.

(SEm), and Ranges Of Pure-

 

 

Frequency in Hertz

 

500 1000 2000

Condi tion SN Cond . SN Cond . SN Cond .
Original

M. 38.12 15.62 44.37 10.62 54.37 8.12

SD 12.48 5.82 11.57 6.34 9.82 8.63

Range 35.00 15.00 40.00 20.00 30.00 20.00
Actual

M. 36.87 16.87 45.62 11.25 53.75 8.12

SD 13.44 6.58 10.73 5.99 11.38 6.58

SEm, 4.33 5.00 5.59 3.95 4.67 5.00

Range 35.00 20.00 35.00 20.00 30.00 20.00
Better

.M 36.87 15.62 45.62 13.12 51.25 8.12

SD 10.86 3.90 10.43 4.96 13.16 8.99

SEm_ 3.53 5.59 3.53 3.06 5.86 2.50

Range 30.00 10.00 35.00 15.00 40.00 25.00
Poorer

M 36.87 17.50 45.00 12.50 51.25 8.75

SD 13.44 6.61 11.45 6.12 13.86 8.19

SEm. 5.00 3.95 1.76 3.53 5.86 3.06

Range 40.00 20.00 40.00 20.00 45.00 20.00
No Info.

M 36.87 16.87 45.00 15.62 51.87 10.62

SD 13.44 5.55 11.16 8.45 14.34 8.07

SEm 5.30 5.59 4.33 7.07 5.59 5.59

Range 35.00 15.00 40.00 30.00 45.00 25.00

 

* In dB re ISO 1964

Examination of Table V shows that the mean bone-

conduction thresholds are in close agreement under all

test conditions for the adults with a sensorineural

 

 

 

 

 

93

hearing disorder. The SE:m for all frequencies under all
conditions do not deviate significantly from the amount
Of error acceptable in bone-conduction audiometry. The
largest SEIn occurred when the testers were given no pre-
vious test result information. On the other hand, for
air-conduction thresholds (see Table IV), the least
variability occurred when the testers were not provided
with previous test results while testing sensorineural
adults. From these results it can be seen that previous
audiometric information did not cause SOphisticated audio-
10gists to exhibit a bias that affected the reliability
Of bone-conduction audiometry when Obtaining thresholds
on subjects with a sensorineural hearing loss.

Table V also shows that the skilled testers were
equally as accurate in obtaining thresholds on conductive
adult subjects. The mean bone-conduction thresholds were
approximately the same for all frequencies under all con-
ditions. The SEm reveal a high degree Of repeatability,
indicating that the SOphisticated testers were not in-
fluenced significantly by the previous test results.
Therefore, according to these results, it seems that
skilled testers Obtain bone-conduction thresholds based
on the subjects' hearing sensitivity and that they are

not influenced by previous audiometric information.

 

 

 

 

94

Figure 6 shows composite audiograms of the mean

bone-conduction thresholds of the eight subjects with a

sensorineural hearing disorder Obtained by the four

testers under four conditions and the original mean

thresholds measured by the investigator. Figure 7 Shows

these same data for the eight conductive hearing impaired

subjects.

FIGURE 6.--Mean Bone-Conduction Thresholds for Eight

Hearing Level in dB (ISO 1964)

35

40

45

50

55

Adults with a Sensorineural Hearing Disorder
Obtained by the Investigator and Four Pre-
Doctoral Students Under Four Conditions.

 

 

Original.

Actual x---x

 

Better A- -- v-A

Poorer 13"3'0

 

No Info. V-u-V

 

 

 

 

 

 

 

 

 

500 1000 2000

Frequency in Hertz

 

 

95

FIGURE 7.—-Mean Bone-Conduction Thresholds for Eight
Adults with a Conductive Hearing Impairment
Obtained by the Investigator and Four Pre-
Doctoral Students Under Four Conditions.

 

 

 

 

 

 

 

 

 

 

 

 

 

 

500 1000 2000

Frequency in Hertz

These figures show that although the variability
between conditions is less than 5 dB at all frequencies
for both groups of subjects, the variability was somewhat
greater for the bone-conduction thresholds of the conduc-
tive loss group than for the sensorineural loss group.

Speech Reception Thresholds.--Tab1e VI reveals
the means, standard deviations, standard errors Of
measurement, and ranges of the speech reception thresholds
recorded for eight sensorineural adult subjects and eight
conductive adult subjects. The table also shows the
original thresholds measured by the investigator of this

study.

3 5

KO

ox

H

a 1 Original. .

" 10 Actual x---x

93 ..
Better A-o-o-A

c

-a

.—| Poorer Duo-U

a) 15 - __-.

5 NO Info. v-u-v

.4

o»

c

-a

g 20

m

m I- l

 

 

I
u 0' -_' ~
Emissau .- _._n

I t
1 0L1.

 

 

96

TABLE VI.--Means (M), Standard Deviations (SD), Standard
Errors of Measurement (SEm), and Ranges of
Speech Reception Thresholds* Obtained for
Eight Sensorineural (SN) Adult Subjects and
Eight Conductive (Cond.) Adult Subjects by the
Investigator and Four Pre-Doctoral Students
Under Four Conditions.

 

 

 

Condition
Subjects
Original Actual Better Poorer NO Info. ‘5
SN
M 38.00 34.75 36.50 35.25 35.50
SD 19.72 19.18 19.08 20.46 17.74
Range 50.00 50.00 48.00 56.00 48.00
Cond.
M 27.25 28.25 27.50 28.25 26.50
SD 9.99 10.79 12.03 11.19 10.80
SEm 2.23 2.91 2.23 1.87
Range 38.00 40.00 46.00 42.00 40.00

 

*In dB re Audiometric:§ero

Tha table also shows that the means are approxi-
mately equivalent for each pathological group under all
conditions. The SEm are in extremely close agreement and
are all below the amount of error acceptable for spondee
threshold clinical reliability. Although the SEm are
essentially the same under all conditions for the sensori-
neural subjects, they are approximately 2 dB larger than
the SEm for all conditions for the adults with a conductive
hearing loss.

These results illustrate that even though the

variability was slightly greater for the sensorineural

{ill-m ‘ ‘..‘,T.".'1EI;I,' "iz in: I"

‘3'

.W
-1 up...

r' - -rv —~ - I
, I‘ VI..:I“:."\*1K:R. ".

 

 

 

 

97

 

group, trained audiologists were not significantly biased
by previous audiometric information when measuring speech
reception thresholds Of hard-of—hearing subjects regardless
Of the type Of loss.

Speech Discrimination Scores.--The means, standard
deviations, standard errors Of measurement, and ranges of
the speech discrimination scores Obtained for eight sen-
sordJueural adult subjects and eight conductive adult sub-
jects hw'four pre-doctoral students under four test
conditions are contained in Table VII. The original
speech discrimination scores measured by the investigator

of this study are also listed in the table.

TABLE VII.-- Means (M), Standard Deviations (SD), Standard
Errors of Measurement (SEm), and Ranges of
Speech Discrimination Scores* Obtained for
Eight Sensorineural (SN) Adult Subjects and
Eight Conductive (Cond.) Adult Subjects by
the Investigator and Four Pre-Doctoral Stu-
dents Under Four Conditions.

 

 

 

Condition
Subjects
Original Actual Better Poorer No Info.
SN
M 72.50 73.75 74.25 73.00 79.75
SD 14.20 16.10 20.89 22.95 15.69
SEm 5.95 8.51 9.89 9.19
Range 38.00 48.00 62.00 68.00 50.00
Cond.
M 94.75 92.75 93.50 95.25 93.75
SD 2.22 4.46 4.87 3.59 4.52
SEm 4.12 3.80 2.64 3.31
Range 6.00 8.00 10.00 10.00 14.00

 

*In percent correct

 

 

98

 

Table VII shows that the mean speech discrimination
scores for the sensorineural subjects are approximately the ,-
samua under all conditions except for the condition Of "no
information." However, all Of the mean discrimination
scores are higher than the mean score obtained by the in- f
vestigator. Therefore, the data do not reveal that the
testers were biased in the direction Of the previous test
results. The SEm between conditions indicate close agree—
ment except for the condition of actual test results,

‘whirfli show the least amount Of variability. However, the
SEm for all conditions exceed the acceptable amount Of
errory fOur percent, for speech discrimination. Clin-
icallyy it is known that subjects with a sensorineural
Ihearing loss usually have relatively poor speech discrim-
.Lnation, and further there is considerable variability in
speech discrimination within a population Of sensorineural
hearing loss subjects. This finding is evident in the
present study as shown by the large standard deviations
and.large ranges for all test conditions. This is
dramatically evident when compared to these same measures
(If variability among the conductive loss group. Never-
'the1ess, based on the data in Table VII, it seems that
the variability of the speech discrimination scores was

a function Of the subjects' responses, rather than a sig-
nificant influence of the testers' biases. This seems

evident because all mean discrimination-scores were

 

 

 

'L' .

 

_.r-,-.‘

o F hi.” . fl“,
ill.

 

 

99

 

recorded in one direction, better than the scores
measured by the investigator.

The mean speech discrimination scores for the
conductive subjects in Table VII are essentially the same
under all conditions. Further, the variability between
the scores under all conditions is minimal. The SEm for
all conditions are quite small and well within the limits
of speech discrimination reliability. These results in-
dicate that skilled testers were not influenced by pre-
vious audiometric information when Obtaining speech
discrimination scores on adults with a conductive hearing
impairment. This finding shows considerable contrast
with the sensorineural group.

Table VII shows that sophisticated audiologists
Obtained speech audiometric scores that were in close
agreement with each other when testing the speech discrim-
ination Of conductive hearing impaired persons. However,
the degree of repeatability measured was poorer when
testing speech discrimination of subjects with a sensori-
neural hearing loss.

In summary the descriptive results of this
eXperiment show that the differences Obtained between
means for pure-tone air and bone-conduction thresholds,
speech reception thresholds, and Speech discrimination
scores were within the limits of clinical reliability

for both pathological groups except for speech discrimination

 

 

100

of the sensorineural loss group. The average standard
error Of measurement for pure-tone air-conduction
thresholds was 4.5 dB for the sensorineural group and 4.7
dB for the conductive group thus closely approaching the
5 dB test-retest reliability acceptable for pure-tone
air-conduction audiometry.

For bone-conduction audiometry, the SEm were 4.6
dB for the sensorineural adults and 4.5 dB for the conduc-
tive adults. Again, these results show accurate test-
retest reliability for bone-conduction thresholds.

The average standard errors of measurement for
speech reception thresholds for the sensorineural subjects
and conductive subjects were 4.3 dB and 2.3 dB respectively.
Both of these SEm are less than the amount of error per-
missible for clinically acceptable speech reception
thresholds.

The average SEm for speech discrimination scores
were 8.3% for the subjects with a sensorineural hearing
loss and 3.5% for the subjects with a conductive hearing
disorder. The SEm for the sensorineural subjects is not
within the four percent limit of test-retest reliability
for speech discrimination testing. However, since the
SEm under all conditions were approximately the same,
these errors were attributed to the subjects' inabilities
to discriminate consistently on repeated measures rather

than being influenced significantly by testers'

 

rat-I

 

 

 

 

 

a.
,,
I.
.
.1
E
i. .
.1 n)
h.
1

 

 

 

 

 

101

expectancies. On the other hand, the SEm for the
conductive subjects are within the amount of error accept-
able for speech discrimination testing. These results
indicate that when subjects responded consistently to
monosyllable words, there was very little difference in
the scores Obtained on repeated measures.

According to these results, it is evident that
SOphisticated audiologists are not influenced significantly
by previous audiometric results when Obtaining pure-tone
air and bone-conduction thresholds, speech reception
thresholds, and speech discrimination scores from hard-Of-

hearing adults regardless of the type of hearing loss.

Statistical Significance

 

Pure-Tone Air-Conduction Thresholds.--In order to

 

test the differences among pure-tone air-conduction
thresholds obtained in this eXperiment the following null
hypothesis was postulated:
There are no significant differences among pure-tone
air-conduction thresholds Obtained for hard-of—hearing
adults as a function Of the following variables:
testers, conditions, pathologies, and frequencies.
A four-factor analysis of variance with repeated
measures on one factor was employed to determine the sig-

nificance of differences among the variables in the

principle comparisons for pure-tone air—conduction

 

"WWW

pl
1

 

 

 

 

 

'

x

7 Or

102

l A four-dimensional table was used to record

audiometry.
the data. In this experiment there were four rows repre-
senting testers, three columns representing frequency,
four slices representing conditions, and two slices
representing pathology. Therefore, an A x B x C x D four-
entry table was employed to arrange the data. The pure—
tone air-conduction thresholds were the criterion measures
entered in each cell.

The means used in comparing differences between
testers, conditions, frequencies, and pathologies are
shown in Table IV on page 88 Of this chapter. A summary
of the analysis is shown in Table XVII of Appendix B.

The factors Of pathology and frequency and the
interactions between these were statistically significant.
Levels within testers and within conditions were not sig—
nificant. This indicates that although the testers did
Obtain thresholds from the sensorineural subjects that
were significantly different from the thresholds obtained
from the conductive subjects, they did not Obtain diff-
erent thresholds as a function of the conditions. Thus,
according to these results, the above null hypothesis was

rejected. However, this does not mean that the rejection

of the null hypothesis indicates that the testers were

 

lWiner, Statistical Principles in Experimental

 

 
   

   

 

 

 

 

 

103

influenced significantly by the previous test results
because, as previously mentioned, the means obtained by
the four testers under the four conditions were not
statistically significant within each pathology group.

Pure-Tone Bone-Conduction Thresholds.--In order

 

to test the differences among pure-tone bone-conduction
thresholds Obtained in this eXperiment the following null
hypothesis was postulated:
There are no significant differences among pure-tone
bone-conduction thresholds Obtained on hard-of—hearing
adults as a function of the following variables:
testers, conditions, pathologies, and frequencies.

A four-factor analysis Of variance that was the
same design as the one described for the air-conduction
thresholds in this experiment was used to test the signi-
ficance of differences among the variables in the principle
comparisons for pure-tone bone-conduction audiometry. The
pure-tone bone-conduction thresholds were the criterion
measures entered in each cell.

The means used in comparing differences between
testers, conditions, frequencies, and pathologies are
contained in Table V on page 92. A summary of the analysis
is shown in Table XVIII Of Appendix B.

The factors Of pathology and frequency and the
interactions between these were statistically significant.
Levels within testers and within conditions, however, were

not Significant. A statistically significant interaction

occurred between testers, conditions, and frequencies.

 

 

 

 

 

 

 

104

Therefore, its implications need not be explored further.
The important finding here was that the interaction within
testers and within conditions was not significant.

Based on these findings the above null hypothesis
was rejected. However, as stated above, this rejection
does not indicate that the testers influenced signifi-
cantly the subjects' bone-conduction thresholds because,
as can be seen, the means Obtained by the four testers
under the four conditions were not statistically signi-
ficant within each pathological group.

Speech Reception Thresholds.--In order to test

 

the differences among speech reception thresholds Obtained
in this eXperiment the following null hypothesis was postu-
lated:
There are no significant differences among mean
speech reception thresholds Obtained on hard-Of-
hearing adults as a function Of the following vari-
ables: testers, conditions, and pathologies.

A three-factor analysis was used to determine the
significance of differences among the variables in the
principle comparisons for speech reception thresholds.

A three-entry table was designed to arrange the data. In
this eXperiment there were four rows representing testers,
four columns representing conditions, and two slices
representing pathologies. Thus, an A x B x C table was
employed to record the data. The speech reception

thresholds were the criterion measures entered in each

cell.

 

 

 

105

The means used in comparing differences between
testers, conditions, and pathologies are Shown in Table
VI, page 96, and a summary of the analysis is contained
in Table XIX, Appendix B.

The F-statistic was not significant for the three
main factors or the interaction among the three factors.
Therefore, the above null hypothesis could not be rejected.
Thus, the testers did not significantly influence the
speech reception thresholds Obtained under any of the four
conditions for either of the two pathological groups.

Speech Discrimination.--In order to test the

 

differences among speech discrimination scores in this

experiment the following null hypothesis was postulated:
There are no significant differences among the mean
speech discrimination scores Obtained for hard-Of-
hearing adults as a function of the following vari-
ables: testers, conditions, and pathologies.

The significance Of differences among the vari-
ables were tested in the same manner as the variables for
Speech reception threshold in this experiment. The
analysis Of variance for the mean speech discrimination
scores was the same design as for the speech reception
thresholds. The obtained speech discrimination scores
(means) were the criterion measures entered in each cell,
and the F-ratio was used in testing the statistical sig-

nificance of the variance attributable to the three main

effects and the three-way interaction.

 

 

 

 

 

 

106

The mean speech discrimination scores used in the
comparison are shown in Table VII, page 97, and a
summary of the analysis is given in Table XX of Appendix B.

The only factor that was statistically significant
was pathology. Levels within testers, within conditions,
and interactions were not statistically significant.

Thus, the aforementioned null hypothesis was rejected.

It was not unexPected that pathology would show signifi-
cance since clinically it is known that hearing acuity is
a function of pathology. In other words, subjects with a
sensorineural hearing loss have discrimination scores
that are poorer than the subjects with a conductive hear-
ing impairment. Therefore, the statistical significance
found for pathology can be attributed to variations as a
function of the subjects' ability to discriminate mono-
syllabic words rather than the bias Of the testers
supported by the fact that the standard deviations,
standard errors of measurement, and ranges for these sub-
jects were relatively large. However, each tester under
each condition Obtained essentially the same scores.

The four-way analyses of variance on the pure-
tone air and bone-conduction thresholds and the three-way
analyses of variance on the speech audiometry scores con-
ducted for this eXperiment indicate that SOphisticated
audiologists were not influenced significantly by previous

audiometric results. According to the statistical results

 

 

 

107

of this experiment, it is evident that skilled testers
Obtained pure-tone air and bone-conduction thresholds,
Speech reception thresholds, and Speech discrimination
scores that are clinically and statistically accurate

for hard-of—hearing adults regardless Of the type Of
hearing loss, and they did not influence subjects' audio-
metric scores based on previous information. The only
exception to this finding was with the Speech discrimina-
tion scores of the sensorineural loss subjects. Although _
statistically they were not significantly different, the
scores for three of the four conditions showed slightly
greater variability according to stringent measures of

clinical acceptability.

Experiment III

In this experiment the same four testers were em-
ployed. They each tested sixteen normal hearing children.
Eight of the children had normal articulation and eight
had defective articulation. Each tester Obtained pure-
tone air-conduction thresholds, speech reception thresholds,
and Speech discrimination scores on a single ear under one
of four test conditions. The conditions were the same as

those described in Experiment I.

Clinical Significance
Pure-Tone Air-Conduction Thresholds.--Table VIII

contains the means, standard deviations, standard errors

 

 

108

of measurement, and ranges Of the pure-tone air-conduction
thresholds recorded for eight normal hearing children with
normal articulation and eight normal hearing children with
defective articulation by four pre-doctoral students under
four conditions. The frequencies tested were: 500, 1000,
and 2000 Hz. The column titled "original" presents the
results Obtained for the subjects by the investigator of
this study.

Examination of Table VIII shows that there is very
little difference between the mean pure-tone air-conduction
thresholds Obtained under the four conditions. This table
reveals that the SEm for all frequencies under all condi—
tions are in good agreement with pure-tone air-conduction
test-retest reliability. However, there is one exception.
This exception occurs at 500 Hz under the condition of
better test results for the defective articulating child-
ren. The reason for this exception cannot be fully ex-
plained or understood; however, at 500 Hz it appears that
previous information did have an effect on one condition
(better).

Although the standard errors Of measurement for
'these children were somewhat larger than the standard
errors of measurement for the normal hearing adults in
IExperiment I, they were approximately the same under all
conditions and were basically within the limits of repeat-

ability for pure-tone audiometry. Clinically this high

 

 

 

 

 

109

 

TABLE VIII.--Means (M), Standard Deviations (SD), Standard
Errors of Measurement (SEm), and Ranges Of
Pure-Tone Air-Conduction Thresholds* Obtained
for Sixteen Normal Hearing Children--Eight
with Normal Articulation (N.A.) and Eight

with Defective Articulation (D.A.)--by the
Investigator and Four Pre-Doctoral Students

Under Four Conditions.

 

 

Frequency in Hertz

 

500 1000 2000

Condition N.A. D.A. N.A. D.A. N.A. D.A.
Original

M 16.25 10.62 6.87 5.00 6.25 4.37

SD 4.84 3.70 4.96 2.82 4.84 3.12

Range 15.00 10.00 15.00 10.00 15.00 10.00
Actual

M 10.62 9.37 1.25 3.12 1.87 3.12

SD 5.82 3.70 2.16 3.42 3.47 2.78

SEm 7.70 5.00 6.84 4.67 5.86 5.00

Range 15.00 15.00 5.00 10.00 10.00 10.00
Better

M 11.87 5.00 3.12 3.12 3.75 2.50

SD 6.09 3.46 3.47 3.42 4.84 2.82

SEm 6.84 8.83 5.86 4.67 4.33 4.67

Range 20.00 10.00 10.00 10.00 15.00 10.00
Poorer

M 13.75 10.62 3.12 5.62 3.12 5.00

SD 5.99 3.70 3.47 3.12 4.96 4.47

SEm 4.33 5.00 5.59 4.67 5.30 6.37

Range 20.00 10.00 10.00 10.00 15.00 10.00
No Info.

M 11.25 11.25 3.75 4.37 1.87 2.50

SD 2.16 2.60 4.14 3.70 3.95 2.82

SEm 6.61 3.95 6.12 3.06 5.00 3.06

Range 5.00 10.00 10.00 10.00 10.00 10.00

 

TrIn dB re ISO 1964 standard

degree Of repeatability indicates that previous audiometric

information did not cause SOphisticated audiologists to

influence significantly normal hearing children's pure-tone

 

110

air-conduction thresholds. According to these results,

it seems that previous audiometric information is only
slightly more influential in creating a bias among well-
trained testers when they are measuring pure-tone air-
conduction thresholds on children than when they are making

these same measurements on adults.

FIGURE 8.--Mean Air-Conduction Thresholds for Eight Normal
Hearing Children with Normal Articulation Ob-
tained by the Investigator and Four Pre-Doctoral
Students Under Four Conditions.

 

 

 

 

 

 

 

 

 

0
Ci
3 5 Original . .
8 Actual x---x
H
" Better A—o-o—A
93
c. 10 Poorer Duo-D
-H
H No Info. V-u-V
m
5
.4 15
O
a
-a
u
m
0
‘3 20

 

 

 

 

 

 

 

500 1000 2000

Frequency in Hertz '

 

111

FIGURE 9.--Mean Air-Conduction Thresholds for Eight Normal
Hearing Children with Defective Articulation
Obtained by the Investigator and Four Pre-
Doctoral Students Under Four Conditions.

 

 

 

 

 

 

 

 

g ()

H

0 Original. .
m

H

~v 5 Actual x---x
m

U Better A-o-o-A
c

"" Poorer am- D

H

$.10 NO Info. V-u-V
a

Ox

2

'I.‘

m 15

m

m

 

 

 

 

 

 

500 1000 2000

Frequency in Hertz

Figure 8 shows a composite audiogram of the mean
air-conduction thresholds of the children with normal
articulation, while Figure 9 illustrates the same informa-
tion for the children with defective articulation.

The above figures also show the relatively high
ciegree of repeatability of the mean air-conduction
‘thresholds for three frequencies under four test condi-
1:ions with the exception at 500 Hz for the actual condi-
txion among the children with defective articulation. It
should also be noted that the mean original audiometric
tflnresholds for the children with normal articulation are
approximately 3 to 4 dB poorer than the thresholds Ob—

tained by the testers under the four conditions. The

 

 

 

 

 

 

(I
/

~ A

« ‘.

 
    

' _ I j
. .‘ .‘ /
u," ‘. A
a: a. a. -. _- .‘
i; ‘_ _~ .. Q t . . I -_ . ; .
‘2‘ * M IX” F: . E 1.“) ' , .' o- , l ‘ . g ‘

- ud- ._..

  

1..
O

 

112

for this difference is not entirely clear; however, this
group Of subjects may have shown some improvement between
the first and second test sessions because of a learning
or practice effect.

Speech Reception Thresholds.--The means, standard
deviations, standard errors Of measurement and ranges Of
the speech reception thresholds Obtained for eight normal
hearing children with normal articulation and for eight
normal hearing children with defective articulation by
four eXperienced testers under four conditions are pre-
sented in Table IX. The original SRTs Obtained for these
subjects by the investigator are also reported in the

table.

TABLE IX.--Means (M), Standard Deviations (SD), Standard
Errors of Measurement (SEm), and Ranges Of
Speech Reception Thresholds (SRT)* for Sixteen
Normal Hearing Children--Eight with Normal
Articulation (N.A.) and Eight with Defective
Articulation (D.A.)--by the Investigator and
Four Pre-Doctoral Students Under Four Conditions.

 

 

 

CONDITION
Subjects Original Actual Better Poorer NO Info .
N . A.
M 7.75 7.50 7.00 8.00 6.25
SD 1.56 4.09 3.74 2.82 2.33
SEm 4.41 3.67 2.34 2.44
Range 4.00 16.00 12.00 8.00 8.00
D.A.
.M 4.50 6.50 6.00 6.50 7.50
SD , 1.32 2.49 2.99 3.78 2.77
SEm 3.46 3.00 4.00 3.80
Range 4.00 10.00 10.00 14.00 12.00

 

*InHB re audiometric zero (20 dB SPL)

 

 

 

 

 

113

The data in the previous table reveal that the
variability between the mean speech reception thresholds
under all conditions are within the limits of clinical
test-retest reliability for spondee words. The SEm between
the conditions indicate that previous audiometric informa—
tion given to sophisticated testers did not cause them to
influence significantly normal hearing children's speech
reception thresholds. The only indication observed that
suggests some greater variability among the four testers -
when compared to the investigator is that, in general,
for all subjects in all four experimental conditions the
ranges for the speech reception thresholds were quite
large. NO other differences were found. Therefore,
according to these results, it seems that previous audio-
metric information did not produce a clinically signifi-
cant bias effect when sophisticated testers Obtained
speech reception thresholds for children even when the
children had defective articulation.

Speech Discrimination.-—Table X presents the means,

 

standard deviations, standard errors of measurement, and
ranges of speech discrimination scores recorded for six-
teen normal hearing children. Eight of these children had
normal articulation, whereas eight had defective articula-
tion. The original speech discrimination scores recorded
for these subjects by the investigator are also reported

in this table.

 

 

 

 

 

114

TABLE X.--Means (M), Standard Deviations (SD), Standard
Errors Of Measurement (SEm), and Ranges of Speech
Discrimination Scores* for Sixteen Normal Hearing
Children—-Eight with Normal Articulation (N.A.)
and Eight with Defective Articulation (D.A.)--
by the Investigator and Four Pre-Doctoral Stu—
dents Under Four Conditions.

 

 

CONDITION
Subjects Original Actual Better Poorer NO Info.

 

N.A.
M 98.50 95.25 98.00 99.00 97.75
SD 1.32 6.92 2.23 1.41 3.38
SEm 8.09 2.23 2.44 3.39
Range 4.00 20.00 6.00 4.00 10.00
D.A.
M 92.75 88.75 93.50 93.50 92.50
SD 2.88 9.63 5.41 5.35 5.91
SEm 11.70 6.00 4.53 7.93
Range 12.00 36.00 20.00 22.00 24.00

 

3FIn percent correct.

Table X reveals that, in general, the mean scores
for children with normal articulation under all conditions
were very comparable. The SEm between the conditions show
essentially the same amount of repeatability which is
within the limits of test-retest reliability for speech
discrimination scores except for the condition of actual
test results. However, the mean score and the SEm for
this condition do not indicate that the testers were
biased in the direction Of the previous information since
11nder this condition the poorest scores were reported as
‘well as the largest standard errors of measurement. The

reason for this exception may, in part, be found by

 

 

 

115

examining the scores of the individual subjects.
Examination of the raw scores show that one tester recorded
a speech discrimination score that was twenty percent
poorer than recorded by any of the other testers for the
same subject. This accounts for the poorer mean score

and the larger SEm but does not explain why this score

was Obtained.

Table X also shows that the mean scores for child-
ren with defective articulation are in very close agreement
under all conditions except for the condition of actual
test results. This same relationship is revealed also
for the SEm. Again, it is difficult to explain why the
greatest amount Of variability occurred under the actual
condition. It is not felt that it occurred because Of the
bias phenomenon since the condition of "no information"
shows a poorer mean score and a greater amount of vari-
ability than either the better or poorer test result
conditions. Therefore, the testers did not record speech
discrimination scores in the direction of the previous
information nor did they record more accurate scores with
no information than they did with better or poorer informa-
tion.

The SEm, except for the actual condition noted
above, vary essentially from 2 to 8%. However, it should
be noted that for the group with normal articulation, the

SEm are smaller than 4%, whereas for the group with

 

 

 

 

116

defective articulation, the SEm range from 4.5 to nearly
8%. This greater variability in the speech discrimination
scores among the subjects with defective articulation is
also reflected in the standard deviations and ranges.

In summary, the descriptive results Of this experi-
ment show that the differences obtained between means for
pure-tone air-conduction thresholds and speech reception
thresholds were within the limits Of clinical reliability
for both groups of children. The average standard error
Of measurement for pure-tone air-conduction thresholds
for the children with normal articulation was 6.7 dB which
is close to the 5 dB test—retest reliability acceptable
for pure-tone air-conduction audiometry. The children
with defective articulation had an average SEm of 4.9 dB
for pure-tone air-conduction thresholds.

The average standard errors of measurement for
speech reception thresholds were 3.21 dB and 3.56 for the
children with normal articulation and those with defective
articulation respectively. Both of these SEm are approx—
imately 2 dB better than the test—retest reliability that
is clinically acceptable for spondee thresholds.

' The average standard errors of measurement for
speech discrimination scores for the children with normal
articulation was 4.03% which is in very close agreement
to the 4% amount Of error acceptable in speech discrimina-

tion testing. The children with defective articulation

 

 

 

." l I
8.39312"! . ~ I 0 R‘ 7
1 *"I e "1&- 5!“ " '
.- I “ , ~e °~*' "

 

117

had an average SEm Of 7.3% which is not in good agreement
with the test-retest reliability of speech discrimination
scores.

From these results it can be seen that skilled
audiologists are not biased by previous audiometric re-
sults, and they do not clinically influence air-conduction
thresholds, speech reception thresholds, or speech dis-
crimination scores Obtained from normal hearing children
even children who have defective articulation. However, "
it is evident from the data in Table X that SOphisticated
testers recorded scores that had more variability when
obtaining Speech discrimination scores for children with
defective articulation than they did when administering
this audiometric test to children with normal articula-
tion. This variability seems to be a function Of defec-
tive articulation rather than a function Of the bias
phenomenon. In other words, the speech discrimination
scores Obtained for the children with defective articula-
tion were seemingly based on the testers' personal inter—
pretation of the Spoken word rather than on previous test
results. Therefore, the sophisticated audiologists were
not biased by the previous audiometric information, and
they did not influence these subjects' speech discrimina-

tion scores.

£50m: ems-19.24335. “In“: - ’

  

‘0

Ill«.‘fi‘l‘“.'."4'
.qt::"" '

 

 

 

 

 

 

 

 

 

118

Statistical Significance

Pure-Tone Air-Conduction Thresholds.--In order to

 

test the differences among mean pure-tone air-conduction
thresholds obtained in this eXperiment the following null
hypothesis was postulated:
There are no significant differences among mean pure-
tone air-conduction thresholds Obtained on normal
hearing children as a function Of the following vari—
ables: testers, conditions, and articulation.

A four-factor analysis of variance with repeated
measures on one factor was employed to determine the sig—
nificance of differences among the variables in the
principle comparisons for pure-tone air-conduction audio-
metry.

The analysis of variance for the mean pure-tone
air-conduction thresholds was the same design as for the
pure—tone thresholds described in Experiment II. Thus,
an A x B x C x D four-entry table was used to record the
data. The pure-tone air-conduction thresholds were the
criterion measures entered in each cell.

The means used in comparing differences between
testers, conditions, articulation (normal articulation
and defective articulation), and frequencies are shown
in Table VIII, page 109, and a summary of the analysis is
given in Table XXI Of Appendix C.

The only factor that was statistically significant

was the frequency factor. Levels Of confidence within

 

 

 

 

     
 

/ . _
s .2- .--.__..
‘ f? s thwfi a)" - ,
s ._ .. i ”3

 

119

 

testers, within conditions, within articulation and
interactions were not statistically significant. There—
fore, the above null hypothesis was rejected. The reason
for the frequency factor being significant is probably

the same in this case as it was in EXperiment I. Thus,

the testers did not significantly influence the pure-tone
air-conduction thresholds Obtained under any of the four
conditions for either the children with normal or defective
articulation.

Speech Reception Thresholds.--In order to test

 

differences among speech reception thresholds Obtained in
this eXperiment the following null hypothesis was postu-
lated:
There are no significant differences among the mean
speech reception thresholds obtained on normal hearing
children as a function Of the following variables:
testers, conditions, and articulation.

The analysis of variance for the mean speech
reception thresholds was the same design as for the speech
reception thresholds described in EXperiment II. An
A x B x C three—entry table contained the data recorded
for this experiment.

The mean speech reception thresholds used in the
comparison are shown in Table IX, page 112. A summary of
the analysis is shown in Appendix C, Table XXII.

The F—statistic was not significant for the three

principle variables or the interactions among the variables.

 

 

 

120

Thus, the above null hypothesis could not be rejected.

In other words, the testers did not significantly influ—
ence the speech reception thresholds measured under any

of the four conditions for either the children with normal
or defective articulation.

Speech Discrimination Scores.--In order to test the

differences among speech discrimination scores Obtained in .5

this eXperiment, the following null hypothesis was postu—
lated: '
There are no significant differences among the mean
speech discrimination scores obtained on normal hear-
ing children as a function of the following variables:
testers, conditions and articulation.

The three-way analysis for the mean speech dis-
crimination scores was the same design as for the speech
reception thresholds described in Experiment II. The
mean speech discrimination scores used in the comparison
are shown in Table X, page 114. A summary of the analysis
is given in Table XXIII of Appendix C.

Levels of confidence within testers, within condi-
tions and interactions were not statistically significant.
However, the factor of articulation was statistically
significant. This means that even though testers did not
Obtain significantly different scores under the four
conditions, they did Obtain Speech discrimination scores
from the children with normal articulation that were sig-

nificantly different from the scores obtained for children

with defective articulation. Therefore, based on these

' ' ’ .UM'J’.‘ '

v..‘
‘A-Ll

2':-
~ 0 n»

 

 

 

 

.‘l f
’5 : ‘y‘ ‘ p”, . I... M"
I '. I 1‘ 2? n . *I y a c
-- .3 v F 4!

 

121

findings, the above null hypothesis was rejected. However,
this is not to assume that the rejection Of the null hypo-
thesis indicates that the testers exhibited a bias because,
as shown, the means obtained by the four testers under the
four conditions were not statistically significant.

The four-way analysis Of variance on the pure-tone
air-conduction thresholds and the three-way analyses Of
variance on the speech reception thresholds and speech
discrimination scores computed for this eXperiment indi-
cate that SOphisticated audiologists were not influenced
significantly by previous audiometric results. These
analyses illustrate that skilled testers obtained mean
air-conduction thresholds, speech reception thresholds,
and speech discrimination scores that were in close agree—
ment under all conditions.

Based on the statistical results of this eXperi-
ment, it can be seen that experimenter (tester) bias was
not exhibited significantly by SOphisticated audiologists
when Obtaining pure—tone air-conduction thresholds, speech
reception thresholds, and Speech discrimination scores on
normal hearing children with either normal articulation
or defective articulation. However, it should be recalled
that there was considerably more variability in the speech
discrimination scores Of the children with defective arti-

culation than among the children with normal articulation.

 

 

I..s

‘5. i

I .'
1 ~‘.. ,. o -. q
‘ ‘:. .i
' a "
a I
‘ 7 7 .

 

122

Experiment IV

In this experiment four speech and hearing therapy
undergraduate students were used as testers. Each of them
tested sixteen hard-Of-hearing adults. Eight of the
adults had a sensorineural hearing loss while eight had a
conductive hearing impairment. Each tester Obtained pure-
tone air and bone-conduction thresholds bilaterally under
one of four test conditions. Although each subject was
tested bilaterally by air and bone-conduction, analysis
was made on a single ear. The conditions were the same
as those described in Experiment I. The frequencies

tested were 250, 500, 1000, 2000, and 4000 Hz.

Clinical Significance

Pure-Tone Air-Conduction ThresholdS.--Table XI
shows the means, standard deviations, standard errors Of
measurement, and ranges Of the pure-tone air-conduction
thresholds recorded for eight adults with a sensorineural
hearing loss. 'Table XII contains the same information
for the adult subjects with conductive hearing impairments.

Table XI shows relatively close agreement between
the mean pure-tone air-conduction thresholds obtained by
the testers under all conditions. The table also reveals
that the SEm are in close agreement with each other under
all conditions; however, the SEm are quite large at 250

Hz under all conditions except for the condition of no

 

 

 

 

TABLE

 

 

123

TABLE XI.--Means (M), Standard Deviations (SD), Standard
Errors of Measurement (SEm), and Ranges of
Pure-Tone Air-Conduction Thresholds* Obtained
for Eight Sensorineural Adults by the Investi-
gator and Four Undergraduate Students Under
Four Conditions.

 

_:
’——‘

Frequency in Hertz

 

Condition 250 500 1000 2000 4000
Original
M 23.13 25.00 20.63 35.63 63.75
SD 13.67 14.14 12.10 7.26 13.16
Range 50.00 50.00 35.00 20.00 40.00
Actual
M 20.00 21.15 20.00 33.75 64.38
SD 11.18 10.82 13.69 8.92 14.67
SEm 8.47 7.07 4.67 4.67 5.86
Range 40.00 40.00 40.00 30.00 40.00
Better
M 22.50 23.13 19.38 33.75 62.50
SD 13.91 11.43 14.01 8.65 12.99
SEm 10.75 5.30 4.33 4.67 3.53
Range 45.00 40.00 40.00 25.00 40.00
Poorer
M 17.50 23.13 20.00 33.13 62.50
SD 11.45 10.58 13.46 9.66 15.81
SEm 11.85 6.84 3.06 4.33 5.00
Range 35.00 35.00 40.00 30.00 35.00
No Info.
M 21.88 21.25 20.00 35.00 64.37
SD 8.99 10.82 14.14 12.74 16.28
SEm 6.08 6.61 3.06 6.84 4.67
Range 35.00 35.00 40.00 40.00 50.00

 

*In dB re 1964 ISO standard.

information.

Although the SEm were large at 250 Hz, the

thresholds recorded at this frequency were not in the

direction of the previous test results given to the

testers (eXperimental conditions).

Based on these results

 

‘ao‘

 

 

 

 

 

124

it is doubtful that the testers were truly biased by the
previous test results.

The mean pure-tone air-conduction thresholds in
Table XII are similar under all conditions. The table
shows that they are approximately equivalent except for
the condition of no information. It was for this condi-
tion at all frequencies that the greatest amount of vari-
ability was found (note the large SEm). Again, the data
in Table XII illustrate that the mean air-conduction
thresholds obtained by the unskilled testers were not in
the direction of previous audiometric results given to
the testers (experimental conditions). Therefore, accord-
ing to these results, it was not possible to determine
whether the thresholds obtained were affected by the
testers' biases. However, the data do indicate that un-
skilled testers obtain air-conduction thresholds that
deviate more from the listeners' true thresholds when
the tester has no previous information than when he has
true or erroneous test results. This may mean that testers
who are not familiar with audiometric testing techniques
feel more secure in recording thresholds when they have
previous test results to review. Except for the condition
of no information, however, the SEm were essentially

within acceptable limits for pure—tone reliability.

 

125

TABLE XII.--Means (M), Standard Deviations (SD), Standard
Errors of Measurement (SEm), and Ranges of
Pure-Tone Air-Conduction Thresholds* Obtained
for Eight Conductive Adult Subjects by the
Investigator and Four Undergraduate Students
Under Four Conditions.

 

 

Frequency in Hertz

 

 

Condition 250 500 1000 2000 4000
Original
M 46.85 45.00 33.75 28.75 33.75
SD 15.79 16.95 16.15 14.52 16.15
Range 45.00 55.00 50.00 40.00 50.00
Actual
M 43.75 41.87 33.75 26.25 31.25
SD 15.56 18.86 20.11 15.36 19.48
SEm 5.86 5.30 4.33 5.00 5.00
Range 50.00 65.00 60.00 50.00 65.00
Better
M 41.87 43.80 35.62 25.62 31.87
SD 14.98 18.83 19.11 12.85 18.36
SEm 6.12 5.00 5.30 7.28 5.30
Range 45.00 55.00 60.00 45.00 55.00
Poorer
M 43.75 43.75 31.87 24.37 34.37
SD 16.72 16.72 15.59 12.60 15.09
SEm 5.30 3.53 4.67 6.37 3.06
Range 45.00 55.00 50.00 45.00 45.00
No Info.
M 39.37 40.00 34.37 23.75 32.50
SD 16.47 18.70 21.99 16.66 18.70
SEm 9.01 7.50 6.08 9.35 5.59
Range 50.00 65.00 70.00 50.00 50.00

 

 

ITn dBire ISO 1964

Figures 10 and 11 show composite audiograms of the
mean air-conduction thresholds of the adults with sensori-
neural hearing disorders and conductive hearing impairments

respectively. They illustrate the degree of repeatability

 

 

 

1' u|l|ll

Julie: .31» 1.! |.I

 

 

126

FIGURE 10.—-Mean Air-Conduction Thresholds for Eight Adults
with a Sensorineural Hearing Disorder Obtained
by the Investigator and Four Undergraduate
Students Under Four Conditions.

 

 

126a

 

15

 

20

 

 

 

 

25

 

 

r Eight 5:: 30

 

nbstfimxz
argraduate

35

 

 

40

 

45

 

Hearing Level in dB (ISO 1964)

50

 

55

 

 

60 .

 

65

 

 

 

 

 

 

 

 

 

 

250 500
Frequency in Hertz

Original.

No Info.

v---—v

Actual x---x

1000 2000

Better A-'-o-A

4000

Poorer Elm-ll

 

of the mean air-conduction thresholds for five frequencies

under four test conditions.

127

The mean thresholds labeled

"original" are those obtained by the investigator.

FIGURE ll.--Mean Air-Conduction Thresholds for Eight Adults
with a Conductive Hearing Disorder Obtained by

the Investigator and Four Undergraduate Students
Under Four Conditions.

20

25

30

35

40

Hearing Level in dB (ISO 1964)

45

50

Original
No Info.

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

v-«-v

250

Actual x---x

500

1000

2000

Frequency in Hertz

Better

A-----A

4000

Poorer 0"" D

 

 

 

128

These figures show that for the subjects with a
sensorineural loss the greatest amount of variability
between means was found in the low frequency region.
Further, in general, there was more variability between
the mean air-conduction thresholds among the conductive
hearing impaired subjects than among the sensorineurals.

Pure-Tone Bone-Conduction Thresholds.--The means,
standard deviations, standard errors of measurement, and
ranges of the pure-tone bone-conduction thresholds for
eight adults with sensorineural hearing disorders are
contained in Table XIII. The same information for adults
with conductive hearing impairments is listed in Table
XIV.

The mean bone-conduction thresholds in Table
XIII show very good agreement under all conditions. The
SEm for all frequencies under all conditions are within
the limits of test-retest reliability for bone-conduction
audiometry. Clinically, this high degree of repeatability
indicates that previous audiometric test results do not
significantly cause unskilled testers to influence sensori-
neural adults' pure-tone bone-conduction thresholds.
According to these results, it is evident that the previous
audiometric information reviewed by the testers did not
elicit a bias that affected clinically the repeatability

of bone-conduction audiometry.

 

 

129

TABLE XIII.--Means (M), Standard Deviations (SD), Standard
Errors of Measurement (SEm), and Ranges of
Pure-Tone Bone-Conduction Thresholds* Obtained
for Eight Sensorineural Adult Subjects by the
Investigator and Four Undergraduate Students
Under Four Conditions.

 

 

Frequency in Hertz

 

Condition 250 500 1000 2000 4000
Original
M 18.13 26.25 20.63 31.88 58.75
SD 11.16 10.82 11.02 7.04 7.39
Range 30.00 35.00 35.00 15.00 20.00
Actual
M 16.25 26.25 18.75 31.88 58.75
SD 8.56 9.27 10.53 8.63 8.56
SEm 5.30 4.33 3.06 4.33 2.50
Range 30.00 30.00 35.00 20.00 20.00
Better
M 16.25 26.25 21.25 31.88 58.75
SD 7.80 9.92 12.18 9.66 7.39
SEm 5.68 2.50 3.95 5.00 0.00
Range 25.00 35.00 35.00 25.00 20.00
Poorer
M 15.00 29.38 17.50 32.50 60.63
SD 7.50 9.16 10.00 10.89 6.34
SEm 5.68 4.67 5.30 6.37 3.06
Range 25.00 30.00 30.00 30.00 15.00
No Info.
M 17.50 29.38 18.75 33.13 57.50
SD 6.61 9.82 8.56 7.88 9.35
SEm 5.30 4.67 3.95 4.33 3.53
Range 25.00 30.00 25.00 20.00 25.00

 

*In dB re ISO

1964

 

 

130

TABLE XIV.--Means (M), Standard Deviations (SD), Standard
Errors of Measurement (SEm), and Ranges of
Pure-Tone Bone-Conduction Thresholds* Obtained
for Eight Conductive Adult Subjects by the
Investigator and Four Undergraduate Students
Under Four Conditions.

 

 

Frequency in Hertz

 

Condition 250 500 1000 2000 4000
Original
M 5.62 17.50 12.50 6.25 15.62
SD 5.26 9.66 9.01 9.60 15.29
Range 15.00 25.00 30.00 30.00 45.00
Actual
M 3.12 14.37 11.25 4.27 13.75
SD 3.47 9.16 7.80 9.16 13.16
SEm 4.33 4.67 4.33 4.67 6.37
Range 10.00 30.00 25.00 30.00 40.00
Better
M 6.87 18.12 15.00 5.00 15.00
SD 5.55 8.26 9.35 10.00 12.50
SEm 4.33 3.95 5.00 3.53 4.67
Range 15.00 25.00 35.00 35.00 40.00
Poorer
M 5.00 18.75 16.25 6.87 17.50
SD 5.00 8.92 7.90 12.97 12.24
SEm 4.33 4.33 6.61 4.67 7.70
Range 15.00 30.00 30.00 35.00 40.00
No Info.
M 3.75 15.00 13.75 8.12 16.87
SD 5.44 9.35 9.60 11.97 12.97
SEm 4.67 5.59 5.59 6.37 7.07
Range 15.00 25.00 30.00 35.00 45.00

 

*In dB re ISO 1964

 

 

 

 

 

131

The data in Table XIV are comparable to the
thresholds shown in Table XIII. The SEm are within the
limits of bone-conduction reliability and are approximately
the same under all conditions. Based on these results, it
can be seen that ineXperienced testers were not influenced
significantly by the previous audiometric test results.

It can be assumed from the data in Table XIV that previous
audiometric information did not produce a bias that was
clinically observable.

The following figures are composite audiograms of
mean bone-conduction thresholds. Figure 12 gives the mean
thresholds of the sensorineural subjects under all condi—
tions obtained by four undergraduate students and the
investigator of this study. Figure 13 shows the same
data for the conductive subjects.

These figures show that for the conductive hearing
impaired subjects the greatest amount of variability be-
tween means was found for the frequencies of 500, 1000,
and 2000 Hz. Also, there was more variability between
the mean bone-conduction thresholds among the conductive

hearing impaired subjects than among the sensorineurals.

 

 

 

 

 

 

132

FIGURE 12.--Mean Bone-Conduction Thresholds for Eight
Adults with a Sensorineural Hearing Disorder
Obtained by the Investigator and Four Under-
graduate Students Under Four Conditions.

 

 

. A ,la—ww—fi-flm

 

s for Eight

aaring Disczia:
and You: Under-

Conditions.

 

132a

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

No Info.

WV

 

15
20 ii
1'29 f.
’ V
l‘ -
o O . \
I \..‘
25 I . '\
'."/ ‘3
3o ‘
3
\D
:3 35
o
m
12'
93 40
c
«4
H
3
3 45
o~
c
oa
H
m
g 50
55
60
65
250 500 1000 2000 4000
Frequency in Hertz
Original. Actual x---x- Better A—~-A

Poorer a....D

Hearing Level in dB (ISO 1964)

133

FIGURE 13.--Mean Bone-Conduction Thresholds for Eight
Adults with a Conductive Hearing Disorder
Obtained by the Investigator and Four Under-
graduate Students Under Four Conditions.

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

0

5
10
15
20

250 500 1000 2000 4000
Frequency in Hertz

Original. . Actual x—--x Better A-----A Poorerlmqu

No Info. V-u-V

The descriptive results of this experiment show
that the standard errors of measurement for pure-tone
air and bone-conduction thresholds obtained by four un-
SOphisticated testers under four conditions are within
the limits of clinical reliability for both pathological
groups. The average standard errors of measurement of
air—conduction thresholds for the eight sensorineural

subjects and for the eight conductive subjects were 5.9 dB

 

EV 8.3%-Bari

 

I I \\
\.
. “.0; I .
71??!” I’flm- ,9, l -

 

134

and 5.7 dB, respectively. Both of the SEm are
approximately the same as the 5 dB acceptable amount of
error allowed in air-conduction audiometry. The average
SEm of bone-conduction thresholds under all conditions
was 4.2 dB for the subjects with sensorineural hearing
disorders and 5.1 dB for the subjects with conductive
hearing impairments. Both of these SEm are within the
limits of bone-conduction test-retest reliability.
Although it cannot be thoroughly eXplained, the bone—
conduction thresholds for both pathological groups showed
less variability than the air-conduction thresholds for
the same groups. However, Carhart and Hayes reported
that bone-conduction audiometry is as reliable or more
so than air-conduction audiometry.l

According to these results, it seems that unskilled
testers were no more influenced by previous audiometric
results when testing pathological ears than were SOphist-
icated audiologists. This is not to say that unskilled
testers are as adept in obtaining pure-tone thresholds
that are as clinically accurate as those obtained by
SOphisticated audiologists; however, within the limita-

tions of this study, the unskilled testers demonstrated

 

1Carhart and Hayes, "Clinical Reliability of Bone
Conduction," pp. 1084-1101.

 

 

 

 

 

thl

ES

 

135

 

that they were capable of obtaining thresholds that were
essentially as free from errors as the thresholds obtained

by the skilled audiologists.

Statistical Significance

Pure-Tone Air-Conduction Thresholds.--In order to
test the differences among pure-tone air-conduction
thresholds obtained in this experiment the following null
hypothesis was postulated:

There are no significant differences among pure-tone
air-conduction thresholds obtained for hard-of—hearing
adults as a function of the following variables:
testers, conditions, pathologies, and frequencies.

A four-factor analysis of variance that was the
same design as the one described for the air-conduction
thresholds in EXperiment II was used to test the signifi-
cance of differences among the variables in the principle
comparison for pure-tone air-conduction audiometry. The
air-conduction thresholds were the criterion measures
entered in each cell.

The means used in comparing differences between
testers, conditions, frequencies, and pathologies are
contained in Tables XI and XII on pages 123 and 125
respectively. A summary of the analysis is contained in

Table XXIV of Appendix D.

The factors of pathology and frequency and the

 

interaction between these were statistically significant.

Levels within testers and within conditions were not

 

 

 

136

significant. This indicates that even though the testers
did not obtain different thresholds as a function of the
conditions, they did obtain thresholds from the sensori-
neural subjects that were significantly different from
the thresholds obtained from the conductive subjects at
the frequencies tested. Thus, according to these results,
the above null hypothesis was rejected. However, this
does not mean that the rejection of the null hypothesis
indicates that the testers were influenced significantly
by the previous test results. As previously mentioned,
the means obtained by the four testers under the four con—
ditions were not statistically significant within each
pathology under each condition. Therefore, these data
indicate that the testers did not obtain different air-
conduction thresholds as a function of conditions.

Pure-Tone Bone—Conduction Thresholds.--In order
to test the differences among pure-tone bone-conduction
thresholds obtained in the eXperiment the following null
hypothesis was postulated:

There are no significant differences among pure-tone
bone-conduction thresholds obtained for hard-of—hearing
adults as a function of the following variables:
testers, conditions, pathologies, and frequencies.

A four-factor analysis of variance that was the
same design as the one described for the air-conduction
thresholds in Experiment II was used to test the signifi-
cance of differences among the variables. The bone-conduc-

tion thresholds were the criterion measures entered in

each cell.

 

137

The means used in comparing differences between
testers, conditions, frequencies, and pathologies are
contained in Tables XIII and XIV on pages 129 and 130
respectively. A summary of the analysis is listed in
Table XXV of Appendix D.

The factors of pathology and frequency and the
interactions between these were statistically significant.
Levels within testers, within conditions, and interactions
between them were not statistically significant. These
results indicate that testers did not obtain thresholds
that were significantly different as a function of the
conditions; however, the testers did obtain thresholds
from the sensorineural subjects that were significantly
different from the thresholds obtained from the conductive
subjects at the frequencies tested. Therefore, based on
these results the above null hypothesis was rejected;
however, the variables that were significant do not indi—
cate that the testers were influenced significantly by
the previous test results. As mentioned earlier, the
means obtained by the four testers were not statistically
significant within each pathology for the four conditions.
Therefore, these data indicate that the testers did not
obtain different bone-conduction thresholds as a function
of conditions. According to the statistical results of
this eXperiment, it is evident that tester bias was not

exhibited significantly by unskilled testers when

 

""7

 

 

 

 

 

 

obta

haru
ind
by

auc

St

 

 

138

obtaining pure-tone air and bone-conduction thresholds on
hard-of—hearing subjects. Statistically these results
indicate that unskilled testers were no more influenced
by previous audiometric information than were skilled

audiologists.

General Discussion

It is felt that it is appropriate at this time to
subjectively analyze the reasons for not obtaining clin-
ical or statistical significance in this investigation.
This seems warranted especially since the phenomena of
eXperimenter bias and effect have been observed experi-
mentally in a variety of areas.

1. This was an extremely structured study that
limited the testers in making subjective judgments. The
testers were restricted to obtaining audiometric scores
and did not have the Opportunity of giving the subjects
instructions or placing the transducers on the subjects.
The testers were given specific instructions on how to
obtain pure-tone air and bone-conduction thresholds,
Speech reception thresholds, and speech discrimination
scores. Therefore, if they were biased by the previous
audiometric results, they were limited in their opportun-
ity to deviate from the required techniques in establish-
ing audiometric scores that were in accord with the

previous test results.

 

 

 

139

The only time that the testers had an opportunity
to make a subjective judgment was when obtaining speech
audiometry scores. However, the only testers that ob-
tained Speech audiometry scores were the skilled audiolog-
ists who seemingly were more concerned in obtaining good
agreement between pure—tone audiometric results and
Speech audiometric results than in obtaining scores that
were in accord with the previous test results.

2. Each tester knew that he was participating in
a thesis project and, in all probability, was extremely
careful in obtaining audiometric scores that were as
accurate as he could obtain.

3. Even though the testers were instructed to
study the previous test results and were given ample time
to review them, there was no measurable means of determin-
ing whether the testers reviewed the scores on the audio-
grams or merely scanned the audiograms to determine which
audiometric tests had been given or which audiometric test
they were to replicate.

4. It seemed that the testers were more concerned
about adhering to the procedures they were to follow than
with the previous audiometric information. This was part-
icularly evident when maSking was employed. Each tester
adhered strictly to the procedures given to him when ob-
taining masked thresholds and recorded the subjects'
thresholds only after following the masking procedure

given to him.

 

 

 

 

 

140

 

It should be pointed out however, that if a
tester does not employ an audiometric procedure that is
highly structures in defining a threshold or discrimina-
tion score, it is conceivable that the tester may be
biased by previous test result information.

In summary it is felt that the non-significances

 

found in this investigation were meaningful based on the

findings of other studies. The results illustrated that

 

the bias phenomenon is not observable when making audio-
metric assessments while following stringent clinical
procedures. Therefore, based on these results, the audio-
logist should review any pertinent audiometric test results
that the patient may have had and not be unduly concerned
that these previous test results may influence his audio-

metric findings.

 

CHAPTER V

SUMMARY, CONCLUSIONS, AND RECOMMENDATIONS

The purpose of this investigation was to determine
whether certain types of previous audiometric information
caused the tester to influence pure-tone and speech audio-
metric results. Specifically, this study was concerned
with the possibility of testers' expectancy affecting
audiometric responses obtained from normal hearing adults,
hard-of—hearing adults having either conductive or sensori-
neural hearing disorders, and normal hearing children

having either normal or defective articulation.

Summary

This study was composed of four individual experi-
ments. In each eXperiment four testers were employed to
administer pure-tone or pure-tone and speech audiometric
tests to adults and children under four test conditions.
The adult subjects were selected on the basis of their
hearing sensitivity, whereas the children were selected
(x1 tile basis of their ability to articulate English
Phonemes. The four experimental conditions were as

follcavvs: actual test results, better than actual test

141

   

 

 

 

142

results, poorer than actual test results, and no test
results.

The testers were divided into two groups. One
group was composed of four pre-doctoral students majoring
in audiology. The second group consisted of four under-
graduate students majoring in Speech and hearing therapy.
These testers were classified as being sophisticated E [

(skilled) audiologists and unSOphisticated (unskilled)

 

testers respectively. @-

A summary of each eXperiment is given below.

Experiment I

Conditions.--The four experimental conditions

 

were: actual test results, better than actual test results,
poorer than actual test results, and no test results.
Testers.--Four pre-doctoral students in audiology
were employed as testers.
Subjects.--The subjects were eight normal hearing
adults ranging in age from twenty to twenty five years
witfll a mean age of 21.4 years and a median age of 20.5
years. All of the subjects were female with the exception
of one.

Tests Administered.--One ear of each subject was

 

teSted by each tester by pure-tone air-conduction audio-
metry at the following frequencies: 250, 500, 1000, 2000,
4000' and 8000 Hz. This was followed by obtaining a

Speec:}1 reception threshold on the same ear.

 

 

 

 

143

Experiment II

Conditions.--The same four conditions were employed

 

in this eXperiment as those in EXperiment I.
Testers.--The testers were the same as those used
in Experiment I.

SubjectS.--The subjects consisted of sixteen hard-

of-hearing adults. Eight subjects, six males and two
females, had a sensorineural hearing loss. These subjects
ranged in age from forty to sixty-three years with a mean
age of 47.0 years and a median age of 43.5 years. The
remaining eight subjects, six females and two males ex-
hibited a conductive hearing impairment. These subjects
ranged in age from eighteen to fifty-two years with a
mean age of 35.6 years and a median age of 32.0 years.

Tests Administered.--Each of the subjects in this

 

experiment was tested bilaterally by air-conduction at

the frequencies of 500, 1000, and 2000 Hz and one ear was
tested by bone-conduction at the same frequencies. Speech
audiometry (Speech reception thresholds and Speech dis-
crimination scores) was administered to the same ear that

unis ‘tested by bone-conduction.

EXperiment III

Conditions.--The conditions in this eXperiment

 

were: 'the same four as those utilized in Experiments I and

II.

 

 

144

Testers.--The same four sophisticated testers
participated in this experiment as in Experiments I and
II.

SubjectS.--The subjects were sixteen normal hear-
ing children. Eight of these children, five females and
three males, ranged in age from five to seven years with
a mean age of 5.8 years and a median age of 5.0 years.
These subjects were judged to have normal speech articula-
tion. The remaining eight children, six males and two
females, ranged in age from five to nine years with a
mean age of 6.3 years and a median age of 6.0 years.

These children had defective articulation characterized
by substitutions, omissions, and/or distortions.

Tests Administered.--Each of these subjects was
given a pure-tone air-conduction test at 500, 1000, and
2000 Hz for one ear and speech audiometry (speech recep-
tion threshold and speech discrimination) was administered

to the same ear that was tested by air-conduction.
Experiment IV

Conditions.--The conditions used in this eXperi-
merrt:*were the same four as those employed in Experiments
I, II, and III.

Testers.--The testers were four undergraduate
Studeerits who were majoring in speech and hearing therapy.

Each of these testers was enrolled in an introductory

 

‘ ,' 3' ‘0 {4'4 0:.!”£;‘i€mfii.&ifi~‘v.ilm-g" 1’ {I ':~

 

 

 

 

145

audiology course at the time of this study. Thus, they
were unsophisticated testers in comparison to the pre-
doctoral students.

Subjects.--Sixteen hearing impaired adults served
as subjects. Eight of the subjects, two females and six
males, were diagnosed as having a sensorineural hearing
loss. They ranged in age from thirty-seven to Sixty-five
years with a mean age of 50.3 years and a median age of
49.5 years. Eight of the subjects exhibited a conductive
hearing impairment. All of these subjects were females
with the exception of one and they ranged in age from
sixteen to thirty-five years with a mean age of 31.4 years
and a median age of 30.0 years.

Tests Administered.--Each subject was tested by

 

pure-tone air and bone-conduction audiometry at the

following frequencies: 250, 500, 1000, 2000, and 4000 Hz.

Conclusions

 

Within the limits imposed by the design of this
stnndy, the following conclusions appear warranted.

1. There are no significant differences between
the pure-tone air-conduction thresholds obtained by
Sophisticated and unSOphisticated testers as a function
0f the type of previous audiometric information (true,
erroneous, of no information) reviewed by these testers.

Howexrear, there are significant differences in the pure-tone

 

 

 

 

 

146

air-conduction thresholds obtained by these testers as a
function of the subjects' age and hearing sensitivity
(adults with normal hearing, adults with a sensorineural
hearing loss, adults with a conductive hearing loss and
children with normal hearing). SOphisticated testers ob—
tain pure-tone air-conduction thresholds for adults,
regardless of the subjects' hearing sensitivity, with
essentially the same degree of repeatability. However,
these same testers obtain pure-tone air-conduction thres-
holds from children with normal hearing that are slightly
less reliable than the same thresholds obtained for normal
hearing adults. Further, although the unskilled testers
obtain pure-tone air-conduction thresholds that reveal a
relatively high degree of repeatability, regardless of the
subjects' hearing sensitivity, the degree of repeatability
is Slightly less than it is for the SOphisticated testers.
2. There are no Significant differences between
pure-tone bone-conduction thresholds obtained by SOphist-
icated and unSOphisticated testers as a function of the
type of previous audiometric information (true, erroneous,
or no information) reviewed by these testers. However,
there are Significant differences in the pure-tone bone-
Conduction thresholds obtained by these testers as a
furuztzion of the subject's hearing sensitivity (adults
Witll a sensorineural hearing disorder and adults with a

conductive hearing impairment). SOphisticated testers

 

 

 

 

 

147

obtain pure-tone bone-conduction thresholds for adults,
regardless of the subject's hearing sensitivity, with
essentially the same degree of repeatability. The unSOphr
isticated testers obtain pure-tone bone-conduction
thresholds that reveal good test-retest repeatability,

regardless of the subject's hearing sensitivity; however,

 

this degree of reliability is slightly less than it is
for the sophisticated testers.

3. There are no significant differences between
the speech reception thresholds obtained by SOphisticated
testers as a function of the type of previous audiometric
information (true, erroneous, or no information) reviewed
by these testers. However, there are significant differ-
ences in Speech reception thresholds obtained as a function
of the subject's hearing sensitivity (normal hearing adults,
adults with a sensorineural hearing loss, and adults with
a conductive hearing loss). SOphisticated testers tend
to obtain Speech reception thresholds that Show a higher
degree of repeatability for adults with normal hearing or

a (nonductive hearing loss than they do for subjects with
a sensorineural hearing loss. Also, there tends to be a
slightly higher degree of repeatability for normal hearing
adults than for normal hearing children. Further, there
are significant differences in speech reception thresholds
Obtained as a function of the subject's ability to arti-

culaizee English phonemes (normal hearing children having

 

 

 

.' \‘
- \
E q I . ’ ' \\'
I _, . a. m : if}; I I” _ .

 

148

normal articulation and children with normal hearing
having defective articulation). These same testers obtain
speech reception thresholds that reveal slightly less
variability for children with normal articulation than

for children with defective articulation.

4. There are no significant differences between
the Speech discrimination scores obtained by sophisticated
testers as a function of the type of previous audiometric
information (true, erroneous, or no information) reviewed
by these testers. However, there are significant differ-
ences in speech discrimination scores obtained as a
function of the subject's age and hearing sensitivity and
ability to articulate English Phonemes (adults with a
sensorineural hearing loss, adults with a conductive
hearing loss, children with normal hearing having normal
articulation, and children with normal hearing having
defective articulation). Sophisticated audiologists ob-
tain discrimination scores that reveal greater variability
for adults with a sensorineural hearing disorder than they

dc>.for adults with a conductive hearing loss or children
vwitfil normal Speech articulation. These same testers ob-
tain speech discrimination scores that show slightly
greater variability for children with defective articula-
ticuu than for children with normal articulation.

5. Essentially all of the audiometric results

Obtacilied in this study by the four SOphisticated

 

 

 

r“ m 10°

 

149

audiologists and by the four unskilled testers under four
experimental conditions (actual test results, better than
actual test results, poorer than actual test results, and
no test results) show a high degree of test-retest reli-
ability. In other words, the testers employed in this
study obtained audiometric scores that were clincally

and statistically reliable, regardless of the type of pre—
vious information concerning the subjects. Therefore,
having previous test result information failed to elicit
an observable bias among the testers regardless of their

academic or clinical eXperience.

Recommendations for Further Research

 

The present study should be repeated in a less
structured situation. The testers should have the oppor—
tunity of giving the subjects instructions, placing the
transducers on the subjects, and obtaining audiometric
scores in any manner they deem acceptable. This may
cause the testers to give the subjects instructions that
unmild cause the subjects to respond in a manner that

wmnild.be more in accord with the testers' expectancies.

 

 

 

 

 

 

 

 

BIBLIOGRAPHY

Books

Blalock, Hubert M. Social Statistics. New York: McGraw-
Hill Company, Inc., 1963.

Chaiklin, Joseph B. and Ventry, Ira M. "Functional Hearing 1
Loss," Modern Developments in Audiology, James
Jerger, editor. New York: Academic Press, Inc.,
1963.

 

O'Neill, John J. and Oyer, Herbert J. Applied Audiometry.
New York: Dodd, Mead & Company, Inc., 1966.

Rosenthal, Robert. Experimenter Effects in Behavioral
Research. New York: Appleton-Century-Crofts,

Ward, Dixon. "Auditory Fatigue and Masking," Modern
Developments in Audiology. James Jerger, editor.
New York: Academic Press, Inc., 1963.

Whyte, William Foote. Street Corner Society. Chicago:
University of Chicago Press, 1943.

Winer, B. J. Statistical Principles in Experimental De-
sign. New York: McGraw-Hill Book Company, Inc.,
1962.

Reports

Harris, J. Donald, and Myers, Cecil. "Experiments on
Fluctuations of Auditory Acuity," Medical Research
Laboratory U.S. Naval Submarine Base, New London,
Connecticut. Volume XI No. 13, Report No. 196
(22 June 1952) Project NM 003 041.21.08.

150

 

151

Tillman, Tom W., and Carhart, Raymond. "An Expanded Test
for Speech Discrimination Utilizing CNC Monosyll—
abic Words: Northwestern University Auditory Test
No. 6 USAF School of Areospace Medicine, Areospace
Medical Division (AFSC) Brooks Air Force Base,
Texas (1966), 1-12.

Periodicals

Back, Kurt W. "Influence Through Social Communication,"
Journal of Abnormal and Social Psychology, 46
(1951), 9—23.

Bellak, Leopold. "The Concept of Projection: An Experi-
mental Investigation and Study of the Concept,"
Psychiatry, 7 (1944), 353-370.

Blakenship, Albert B. "The Effect of the Interviewer
Upon the Response In a Public Opinion Poll,"
Journal of Consulting Psychology, 4 (1940), 134-
136.

Carhart, Raymond. "Individual Differences in Hearing for
Speech," Annals of Otolggy, Rhinology: and Laryng-
oloqy, 55 (1946), 223-267.

Carhart, Raymond. "Speech Reception in Relation to Pattern
of Pure Tone Loss," Journal of Speech Disorders, 11
(1946), 97—108.

Carhart, Raymond. "Problems in the Measurement of Speech
Discrimination," Archives of Otolaryngology, 82
(1965), 253-260.

Carhart, Raymond and Hayes, Claude. "Clinical Reliability

of Bone Conduction Audiometry," Laryngoscope, 50
(1940), 1084—1101.

Carhart, Raymond and Jerger, James. "Preferred Method of
Determination of Thresholds," Journal of Speech
and Hearing Disorders, 24 (1959), 330-345.

 

Chaiklin, Joseph B. and Ventry, Ira M. "Spondee Threshold
Measurement: A Comparison of 2- and 5- dB Methods,"
Journal of Speech and Hearing Disorders, 29 (1964),
47—59.

Cochran, William G. and David J. Watson. “An Experiment
on Observer's Bias in the Selection of Shoot-
Heights," Empire Journal of Experimental Agri—
culture, 4 (1936), 69-76.

 

 

 

 

 

 

152

Corado, Lucion and Ison, James R. "Observer Bias in
Conditioning of Planarian," Psychological Reports,
13 (1963), 787-789.

Fletcher, Harvey. "Auditory Patterns," Review of Modern
Physics, 12 (1940) 47-65.

 

Fletcher, Harvey. "A Method of Calculating Hearing Loss
for Speech from an Audiogram," Journal of the
Acoustical Society of America, 22 (1950), 1-5.

 

Fletcher, Harvey and Munson, W. A. "Relation Between
Loudness and Masking," Journal of the Acoustical
Society of America, 9 (1937), 1-10.

Freidman, Neil, Kurland, Daniel, and Rosenthal. "Experi-
menter Behavior as an Unintended Determinant of
Experimental Results," Journal of Projective
Techniques, 29 (1965), 479-490.

 

Hanson, Robert H., and Marks, Eli S. "Influence of the
Interviewer on the Accuracy of Survey Results,"
Journal of the Amgrican Statistical Association,
53 (1958), 635-655.

Harvey, S. M. "A Preliminary Investigation of the Inter-

view," British Journal of Psychology, 28 (1938),
263-287.

High, Wallace 5., Glorig, Aram, and Nixon, James. "Esti-
mating the Reliability of Auditory Threshold
Measurements," Journal of Auditory Research, 1
(1961), 247-262.

Hirsh, Ira, Davis, Hallowell, Silverman, S. Richard,
Reynolds, Elizabeth G., Eldert, Elizabeth, Benson,
Robert. "Development of Materials for Speech
Audiometry," Journal of Speech and Hearing Dis-
orders, 17 (1952), 321-337.

Hood, J. D.) "Principles and Practice of Bone Conduction
Audiometry," Laryngoscope, 70 (1960), 1211-1228.

Hudgins, C. V., Hawkins, J. F., Karlin, J. E., and
Stevens, S. S. "The Development of Recorded
Auditory Tests for Measuring Hearing Loss for
Speech," Laryngoscope, 57 (1947), 57-89.

 

Jerger, James F. "Comparative Evaluation of Some Auditory
Measures," Journal of Speech and Hearing Research,
5 (1962), 3-17.

 

4..)

 

 

 

153

Jerger, James F., and Carhart, Raymond, Tillman, Ton W.
"Some Relations Between Normal Hearing for Pure
Tones and for Speech," Journal of Speech and
Hearing Research, 2 (1959), 126—140.

 

Lehiste, I. and Peterson, G. E. "Linguistic Considera-
tions in the Study of Speech Intelligiliby,"
Journal of the Acoustical Society of America,
31 (1959), 280-286.

Lord, Edith. "Experimentally Induced Variations in
Rorschach Performance," Psychological Monographs,
64 (1961), 321-333.

 

Luft, Joseph. "Interaction and Projection," Journal of
Projective Techniques, 17 (1953), 489-492.

 

 

Lybarger, S. F. "Interim Bone Conduction Thresholds for
Audiometry," Journal of Speech and Hearing Research,
9 (1966), 483—487.

 

Masling, Joseph. "The Effects of Warm and Cold Inter-
action on the Administration and Scoring of an
Intelligence Test," Journal of Consulting Psychol-
pgy, 23 (1959), 336-341.

 

Menzel, Otto J. "Error in Audiometry," Eye Ear Nose and
Throat Monthly, 42 (1963), 74.

 

 

Peterson, G. E. and Lehiste, I. "Revised CNC Lists for
Auditory Tests," Journal of Speech and Hearing
Disorders, 27 (1962), 62-70.

 

 

Rhum, Howard B. and Carhart, Raymond. "Objective Speech
Audiometry," Journal of Speech and Hearing Re-
search, 1 (1958), 169-178.

 

Rintelmann, William F. amd Harford, Earl R. "Type V
Bekesy Pattern: Interpretation and Clinical
Utility," Journal of Speech and Hearing Research,
10 (1967), 733-744.

 

Rosenthal, Robert. "Experimenter Attributes as Deter-
minants of Subjects' Responses," Journal of Projec-
tive Techniques, 27 (1963), 324-331.

 

 

Rosenthal, Robert. "EXperimenter Modeling Effects as
Determinants of Subjects' Responses," Journal of
Projective Techniques, 27 (1963), 467-471.

 

 

 

 

 

 

154

Rosenthal Robert and Fode, Kermit L. "The Effects of

Experimenter Bias on the Performance of the Albino
Rat," Behavioral Science, 8 (1963), 183—189.

Rosenthal, Robert, Kohn, Paul, Greenfield, Patricia M.,
and Carota, Noel. "Data Desirability, EXperimenter
Expectancy, and the Results of Psychological Re-
search," Journal of Personality and Social Psygho—
logy, 3 (1966), 20-27.

Rosenthal, Robert and Lawson, Reed. "A Longitudinal Study
of the Effects of EXperimenter Bias on the Operant
Learning of Laboratory Rats," Journal of Psychiatric
Research, 2 (1964), 61-72.

Rosenthal, Robert, Murly, Ray C., Persinger, Gordon W.,
and Vikan-Kline, Linda. "Changes in Experimental
Hypotheses as Determinants of Experimental Results,"
Journal of Projective Techniqpes, 28 (1964), 465-
469.

Sanders, Jay W. and Rintelmann, William F. "Masking in
Audiometry," Archives of Otolaryngology, 80 (1964),
541-556.

Sarason, Irwin G. "Individual Differences, Situational
Variables, and Personality Research," Journal of

Abnormal and Social Psychology, 23 (1959), 336-
341.

 

Sataloff, Joseph. "Pitfalls in Routine Hearing Tests,"
Archives of Otolaryngology, 73 (1961), 717-726.

Shapiro, Arthur. "A Contribution to a History of the
Placebo Effect," Behavioral Science, 5 (1960),
109-135.

 

Stream, Richard W. and McConnell, Freemann. "A Comparison
of Two Methods of Administration in Bekesy-Type
Audiometry," Journal of Auditory Research, 4 (1961),
263-271.

Studebaker, Gerald A. "Clinical Masking of Air- and Bone-
Conducted Stimuli," Journal of Speech and Hearing
Disorders, 29 (1964), 23-35.

 

Tillman Tom W. and Jerger, James F. "Some Factors Affect-
ing the Spondee Threshold in Normal-Hearing Sub-
jects," Journal of Speech and Hearing Research, 2
(1959), 141-146.

 

 

 

 

'Y V /‘/ I / x -.

" , / /

 

155

Tillman, Tom W., Johnson, Robert M. and Olsen, Wayne 0.
"Earphone versus Sound Field Threshold Sound
Pressure Levels for Spondee Words," Journal of
the Acoustical Society of America, 39 (I966), 125-
133.

Witting , E. G. and Hughson, Walter. "Inherent Accuracy
of a Series of Repeated Clinical Audiograms,"
LaryngoscoPe, 50 (1940), 259-269.

 

Unpublished Material

Haskins, Harriet. "A Phonetically Balanced Test of Speech
Discrimination for Children" (Unpublished Master's
Thesis, Northwestern University, 1949).

Rintelmann, William F. and Jetty, Albert J. "Reliability
of Speech Discrimination Testing Using CNC Mono-
syllabic Words" (Unpublished Study, Michigan State
University, 1968).

Tokay, F. Harry. "Validity and Reliability of Bekesy

Audiometry with Preschool Age Children" (Unpublished
Doctor's Thesis, Michigan State University, 1967).

Other Sources

 

. "American Standard Specification for Audio-
meters for General Diagnostic Purposes," American
Standards Association, Incorporated, No. S3-3-l960.

"Operating Manual for the Beltone Artificial
Mastoid, M5 Series," Beltone Electronics Corpora-
tion, Chicago, Illinois, 1968.

 

APPENDICES "

 

156

 

 

 

 

TP.

,
l

 

157

APPENDIX A

RESULTS OF PURE-TONE AIR-CONDUCTION THRESHOLDS

TABLE XV.--Summary of Analysis of Variance Comparing Differ-
ences of Testers (A), Conditions (B), and Fre-
quency (C) for Eight Normal Hearing Adults by
Pure-Tone Air-Conduction Audiometry.

 

 

 

 

Sum of Mean F-
Source of Variance Squares df Square Statistic
A 32.682 3 10.894 0.171
B 70.182 3 23.394 0.368
AB 807.421 9 89.713 1.414
Z=D+AD+BD+ABD 1014.583 16 63.411
C 2833.463 5 566.692 16.059*
AC 143.098 15 9.539 0.270
BC 93.098 15 6.206 0.175
ABC 869.921 45 19.331 0.547

Z=CD+ACD+BCD+ABCD 2822.916 80 35.286

TOTAL 8687.369 191

 

*Significant beyond the 0.01 level.

 

 

158

APPENDIX A (Cont.)

RESULTS OF SPEECH RECEPTION THRESHOLDS

TABLE XVI.--Summary of Analysis of Variance Comparing
Differences of Testers (A), and Conditions
(B) for Eight Normal Hearing Adults for Speech
Reception Thresholds.

 

 

 

Sum of Mean F-
Source of Variance Squares df Square Statistic
A 2.000 3 0.666 0.333
B 5.000 3 1.666 0.833
AB 63.000 9 7.000 3.500
Z=C+AC+BC+ABC 32.000 16 2.000
TOTAL 102.000 31

 

 

159

APPENDIX B

RESULTS OF PURE-TONE AIR-CONDUCTION THRESHOLDS

TABLE XVII.--Summary of Analysis of Variance Comparing
Differences of Testers (A), Conditions (B),
Pathologies (C), and Frequency (D) for Six-
teen Hard-of-Hearing Adults by Pure-Tone Air-
Conduction Audiometry.

 

 

 

Sum of Mean F-

Source of Variance Squares df Square Statistic
A 116.015 3 38.671 0.079
B 22.265 3 7.421 0.015
C 9422.005 1 9422.005 19.250*
AB 1655.338 9 183.926 0.375
AC 37.890 3 12.630 0.025
BC 25.390 3 8.463 0.017
ABC 4779.296 9 531.032 1.084
Z=E+AE+BE+CE+ABE+

ACE+BCE+ABCE 15662.499 32 489.453
D 1254.166 627.083 5.112*
AD 56.250 9.375 0.076
BD 87.500 6 14.583 0.118
CD 5529.166 2 2764.583 22.539*
ABD 3052.083 18 169.560 1.382
ACD 37.500 6.250 0.050
BCD 25.000 4.166 0.033
ABCD 1775.000 18 98.611 0.803
Z=DE+ADE+BDE+CDE+

ABDE+ACDE+BCDE+

ABCDE 7849.999 64 122.656
TOTAL 51387.369 191

 

*Significant beyond the 0.01 level.

 

160

APPENDIX B (Cont.)
RESULTS OF PURE-TONE BONE-CONDUCTION THRESHOLDS

TABLE XVIII.--Summary of Analysis of Variance Comparing
Differences of Testers (A), Conditions (B),
Pathologies (C), and Frequency (D) for Six-
teen Hard-of—Hearing Adults by Pure—Tone
Bone-Conduction Audiometry.

 

 

 

 

 

 

Sum of Mean F-

Source of Variance Squares df Square Statistic
A 15.104 3 5.034 0.019
B 52.604 3 17.534 0.066
C 49088.020 1 49088.020 186.631*
AB 1040.104 9 115.567 0.439
AC 77.604 3 25.868 0.098
BC 42.187 3 14.062 0.053
ABC 2004.687 9 222.743 0.846
Z=E+AE+BE+CE+ABE+

ACE+BCE+ABCE 8416.666 32 263.020
D 460.156 2 230.078 5.106*
AD 16.927 6 2.821 0.062
BD 57.552 6 9.592 0.212
CD 4231.510 2 2115.755 46.962*
ABD 2865.364 18 159.186 3.533*
ACD 93.489 6 15.581 0.345
BCD 25.781 6 4.296 0.095
ABCD 1799.218 18 99.956 2.218
Z=DE+ADE+BDE+CDE+

ABDE+ACDE+BCDE+

ABCDE 2833.339 64 45.052
TOTAL 73170.312 191

 

*Significant beyond the 0.01 level.

 

161

APPENDIX (Cont.)

RESULTS OF SPEECH RECEPTION THRESHOLDS

TABLE XIX.--Summary of Analysis of Variance Comparing
Differences of Testers (A), Conditions (B),
and Pathologies (C) for Sixteen Hard-of-

Hearing Adults for Speech Reception Thresholds.

 

 

 

Sum of Mean F-

Source of Variance Squares df Square Statistic
A 3.187 3 1.062 0.002
B 6.687 3 2.229 0.006
C 976.562 1 976.562 2.702
AB 2902.562 9 311.395 0.861
AC 12.187 3 4.062 0.011
BC 18.687 3 6.229 0.017
ABC 1355.562 9 150.618 0.416
Z=D+AD+BD+CD+ABD+

ACD+BCD+ABCD 11562.000 32 361.312
TOTAL 16737.437 63

 

 

162

APPENDIX B (Cont.)

RESULTS OF SPEECH DISCRIMINATION SCORES

TABLE XX.--Summary of Analysis of Variance Comparing
Differences of Testers (A), Conditions (B),
and Pathologies (C) for Sixteen Hard-of—Hearing
Adults for Speech Discrimination Scores.

 

 

 

Sum of Mean F-

Source of Variance Squares df Square Statistic
A 373.500 3 124.500 0.675
B 97.500 3 32.500 0.176
C 5402.250 1 5402.250 29.320*
AB 3498.000 9 388.666 2.109
AC 76.250 3 25.416 0.137
BC 140.250 3 46.750 0.253
ABC 2683.250 9 298.138 1.618
Z=D+AD+BD+CD+ABD+

ACD+BCD+ABCD 5896.000 32 184.250
TOTAL 18167.000 63

 

*Significant beyond the 0.01 level.

 

163

APPENDIX C

RESULTS OF PURE-TONE AIR-CONDUCTION THRESHOLDS

TABLE XXI.--Summary of Analysis of Variance Comparing
Differences of Testers (A), Conditions (B),
Articulation (C) and Frequency (D) for Six-

teen Normal Hearing Children by Pure-Tone Air-

Conduction Audiometry.

 

 

 

Sum of Mean F-

Source of Variance Squares df Square Statistic
A 109.375 3 36.458 1.538
B 128.125 3 42.708 1.802
C 4.687 1 4.687 0.197
AB 370.833 9 41.203 1.738
AC 25.520 3 8.506 0.358
BC 92.187 3 30.729 1.296
ABC 419.270 9 46.585 1.965
Z=E+AE+BE+CE+ABE+

ACE+BCE+ABCE 758.333 32 23.697
D 2259.375 2 1129.687 58.228*
AD 18.750 6 3.125 0.161
BD 78.125 6 13.020 0.671
CD 153.125 2 76.562 3.946
ABD 197.916 18 10.995 0.566
ACD 47.916 6 7.986 0.411
BCD 53.125 6 8.854 0.456
ABCD 366.666 18 20.370 1.049
Z=DE+ADE+BDE+CDE+ ’

ABDE+ACDE+BCDE+

ABCDE 1241.666 64 19.401
TOTAL' 6325.000 191

 

*Significant beyond

the 0.01 level.

i) ‘5

‘1
.* 1.:

   

164

APPENDIX C (Cont.)

RESULTS OF SPEECH RECEPTION THRESHOLDS

TABLE XXII.--Summary of Analysis of Variance Comparing
Differences of Testers (A), Conditions (B),
and Articulation (C) for Sixteen Normal
Hearing Children for Speech Reception

 

 

 

Thresholds.
Sum of Mean F-
Source of Variance Squares df Square Statistic
A 83.687 3 27.895 2.672
B 5.687 3 1.895 0.181
C 7.562 1 7.562 0.724
AB 214.062 9 23.784 2.278
AC 19.687 3 6.562 0.628
BC 11.687 3 3.895 0.373
ABC 130.062 9 14.451 1.384

Z=D+AD+BD+CD+ABD+
ACD+BCD+ABCD 334.000 32 10.437

TOTAL 806.437 63

 

 

165

APPENDIX C (Cont.)

RESULTS OF SPEECH DISCRIMINATION SCORES

TABLE XXIII.-—Summary of Analysis of Variance Comparing
Differences of Testers (A), Conditions (B),
and Articulation (C) for Sixteen Normal
Hearing Children for Speech Discrimination

 

 

 

Scores.
Sum of Mean F-
Source of Variance Squares df Square Statistic
A 143.687 3 47.895 0.814
B 175.187 3 58.395 0.992
C 473.062 1 473.062 8.043*
AB 457.062 9 50.784 0.863
AC 109.687 3 36.562 0.621
BC 8.187 3 2.729 0.046
ABC 260.062 9 28.895 0.491
Z=D+AD+BD+CD+ABD+
ACD+BCD+ABCD 1882.000 32 58.812
TOTAL 3508.937 63

 

*Significant beyond the 0.01 level.

 

 

 

 

166

APPENDIX

RESULTS OF PURE-TONE AIR-CONDUCTION THRESHOLDS

TABLE XXIV.--Summary of Analysis of Variance Comparing
Differences of Testers (A), Conditions (B),
Frequency (C), and Pathologies (D) for Six-
teen Hard-of—Hearing Adults by Pure-Tone
Air-Conduction Audiometry.

 

 

 

Sum of Mean F-

Source of Variance Squares df Square Statistic
A 49.609 3 16.536 0.012
B 24.609 3 8.203 0.006
D 828.828 1 828.828 0.638
AB 1262.578 9 140.286 0.108
AD 212.734 3 70.911 0.054
BD 88.984 3 29.661 0.022
ABD 8993.203 9 999.244 0.769
Z=E+AE+BE+DE+ABE+

ADE+BDE+ABDE 41547.500 32 1298.359
C 17652.656 4 4413.1640 37.12685*
AC 155.468 12 12.955 0.108
BC 155.468 12 12.955 0.108
CD 32966.093 4 8241.523 69.333*
ABC 2258.906 36 62.747 0.527
ACD 197.031 12 16.419 0.138
BCD 239.531 12 19.960 0.167
ABCD 2179.843 36 60.551 0.509
Z=CE+ACE+BCE+CDE+

ABCE+ACDE+BCDE+

ABCDE 15215.000 128 118.867
TOTAL 124028.046 319

 

*Significant beyond the 0.01 level.

 

 

 

 

APPENDIX D (Cont.)

167

RESULTS OF PURE-TONE BONE-CONDUCTION THRESHOLDS

TABLE XXV.--Summary of Analysis of Variance Comparing
Differences of Testers (A), Conditions (B),
Frequency (C), and Pathologies (D) for Six—
teen Hard-of—Hearing Adults by Pure-Tone Bone-
Conduction Audiometry.

 

 

 

 

 

Sum of Mean F-

Source of Variance Squares df Square Statistic
A 43.437 3 14.479 0.046
B 190.312 3 63.437 0.205
D 30225.312 1 30225.312 97.846*
AB 813.437 9 90.381 0.292
AD 8.437 3 2.812 0.009
BD 91.562 3 30.520 0.098
ABD 2214.687 9 246.076 0.796
Z=E+AE+BE+DE+ABE+

ADE+BDE+ABDE 9885.000 32 308.906
C 25739.375 4 6434.8437 67.707*
AC 167.500 12 13.958 0.146
BC 170.625 12 14.218 0.149
CD 15120.000 4 3780.000 39.773*
ABC 1422.500 36 39.513 0.415
ACD 186.875 12 15.572 0.163
BCD 191.250 12 15.937 0.167
ABCD 1786.875 36 49.635 0.522
Z=CE+ACE+BCE+CDE+

ABCE+ACDE+BCDE+

ABCDE 12165.000 128 95.039
TOTAL 100422.187 319

 

*Significant beyond the 0.01 level.

1 If“ . '1 -. r' I

   

168

APPENDIX E

PURE-TONE AIR-CONDUCTION THRESHOLDS*
OBTAINED IN EXPERIMENT I

Frequency in Hertz
250 500 1000 2000 4000 8000

Subjects ORIGINAL
l 5 5 O 0 0 20
2 5 0 -10 -5 5 0
3 0 5 0 0 5 0
4 10 15 -5 -5 5 20
5 5 5 -5 -10 5 15
6 10 10 -5 0 5 10
7 0 0 -5 0 -5 5
8 -5 0 0 0 0 -5

ACTUAL
l 0 5 0 -5 10 0
2 5 0 -10 -10 5 5
3 0 0 -5 0 5 -10
4 10 10 -5 .-10 5 15
5 5 0 -10 -15 5 10
6 5 5 0 5 5 0
7 0 0 -5 -5 —5 5
8 -5 -5 -5 0 0 -5

BETTER
l 5 0 -5 -5 10 5
2 5 0 -15 -15 0 5
3 0 0 0 0 5 -5

/

I
\
-' , 7 1'3- fl": '_

          

I I
!

f

' u
_/I ,
I’ 4 .' ' 3
|_ ‘1 ‘ - _ Y .
"J .- “ : . _ .
I .; f: 1'k‘ ‘ 1 . h“ . ¢.- ._ I f .
‘ '. “I ‘ I .' ° ‘ s ‘

: Han

        

.1.-I-.‘.'.'-_'.."'.'

APPENDIX E (Cont.)
Frequency in Hertz
250 500 1000 2000 4000 8000

 

Subjects BETTER (Cont.)
4 10 10 -5 -10 10 15
5 0 0 -10 -10 0 5
6 10 10 0 5 0 10
7 0 0 -5 0 —5 5
8 0 -5 -5 -5 0 -5
POORER _
1 0 0 5 0 10 10
2 5 0 -10 -S 0 0
3 0 0 -5 0 10 -10
4 10 10 -5 -5 5 15
5 0 0 -5 -15 5 20
6 10 15 0 5 5 5
7 5 0 -5 0 0 5
8 0 0 -5 -5 0 -10

NO INFORMATION

1 0 0 -5 -10 5 0
2 0 0 -10 -10 5 0
3 0 0 0 0 10 -5
4 5 10 0 -10 10 20
5 0 5 -10 -10 5 15
6 10 5 -5 0 0 10
7 -5 0 -10 -5 -5 5
8 -5 -5 -5 -5 O -5

 

 

IIn dBIre ISO 1964

 

170

APPENDIX E (Cont.)

SPEECH RECEPTION THRESHOLDS*
OBTAINED IN EXPERIMENT I

 

Original Actual Better Poorer Igformation
Subjects
1 -4 -4 —6 -4 -6
2 -4 -6 -8 —4 -4
3 -4 -2 —2 —4 -4
4 0 -2 0 0 -2
5 -4 -6 -6 -4 —6
6 0 -2 -4 -4 -4
7 -4 —2 —2 -4 -4
8 -2 —2 -4 2 -4

 

*In dB re Audiometric zero.

 

PURE-TONE AIR AND BONE-CONDUCTION
THRESHOLDS* OBTAINED IN EXPERIMENT II

171

APPENDIX F

 

Frequency in Hertz 500 1000 2000
Air Bone Air Bone Air Bone
Sensorineural
Subjects ORIGINAL
l 45 55 60 50 60 55
2 20 30 20 20 35 35
3 50 45 40 40 45 45
4 55 50 45 60 60 60
5 50 40 45 40 50 50
6 10 20 40 40 80 65
7 50 45 60 55 65 65
8 10 20 50 50 60 60
ACTUAL
l 50 55 60 65 70 65
2 20 25 25 20 35 35
3 45 35 50 45 45 40
4 55 55 45 50 55 55
5 50 40 50 40 55 45
6 5 20 40 50 80 65
7 50 45 50 55 55 65
8 10 20 45 50 60 60
BETTER
l 50 50 60 50 60 60
2 25 35 20 20 25 25

 

 

APPENDIX F (Cont.)

172

 

Frequency in Hertz 500 1000 2000
Air Bone Air Bone Air Bone

Sensorineural

Subjects BETTER (Cont.)
3 45 40 50 45 45 40
4 55 50 45 55 50 50
5 50 40 50 45 50 45
6 5 20 40 45 65 65
7 55 40 60 55 65 65
8 5 20 50 50 60 60

POORER
l 45 55 60 50 60 55
2 25 35 20 20 25 20
3 45 40 45 40 45 40
4 60 55 50 60 55 60
5 50 40 50 40 55 50
6 5 15 40 45 75 65
7 50 35 60 55 65 60
8 10 20 50 50 60 60
NO INFORMATION

1 45 55 60 50 60 55
2 25 25 20 20 25 20
3 45 35 45 45 45 40
4 55 55 50 60 60 60
5 45 40 50 45 55 50
6 5 20 40 50 80 65

 

173

APPENDIX F (Cont.)

 

Frequency in Hertz 500 1000 2000
Air Bone Air Bone Air Bone

Sensorineural

Subjects NO INFORMATION (Cont.)
7 50 45 55 55 65 65
8 10 20 50 50 60 60

Conductive

Subjects ORIGINAL
1 50 20 55 20 50 20
2 50 15 30 15 35 10
3 35 10 30 5 30 0
4 40 25 30 10 30 0
5 15 20 10 0 15 0
6 35 5 35 5 30 0
7 45 15 35 15 30 20
8 25 15 25 15 25 15

ACTUAL

l 50 30 65 20 50 20
2 40 15 35 10 30 5
3 40 15 20 15 25 10
4 35 25 30 10 30 0
5 15 15 10 0 15 5
6 35 10 30 5 35 0
7 45 15 40 15 30 15

8 20 10 25 15 20 10

 

 

 

 

 

 

i—A _d__

 

174

APPENDIX F (Cont . )

 

Frequency in Hertz 500 1000 2000
Air Bone Air Bone Air Bone

Conductive

Subjects BETTER
1 50 15 55 20 40 25
2 40 20 30 15 25 10
3 45 20 40 15 30 0
4 35 20 25 15 25 0
5 10 15 10 5 10 0
6 35 10 35 5 35 0
7 50 15 35 15 30 15
8 20 10 25 15 25 15

POORER

1 45 30 60 25 45 20
2 40 15 30 15 25 15
3 35 10 20 10 30 0
4 35 25 30 10 25 0
5 20 20 10 5 10 0
6 35 10 35 5 35 5
7 40 15 35 15 30 20
8 25 15 25 15 20 10

NO INFORMATION

1 50 25 65 35 45 25
2 50 10 35 15 35 10
3 40 20 30 15 30 15

4 30 20 25 10 30 0

 

APPENDIX F (Cont. )

175

 

Frequency in Hertz 1000 2000
Air Bone Air Bone Air Bone

Conductive

Subjects NO INFORMATION (Cont.)
5 10 20 10 5 10 0
6 40 10 30 10 30 5
7 50 20 35 20 30 15
8 20 10 25 15 25 15

 

*In dB re ISO 1964

 

176

APPENDIX F (Cont.)

SPEECH RECEPTION THRESHOLDS* AND
SPEECH DISCRIMINATION SCORES#
OBTAINED IN EXPERIMENT II

 

 

Condi- _
tion Original Actual Better Poorer No Info.
SRT PB SRT PB SRT PB SRT PB SRT PB

Sensori-

neural

Subjects
1 56 86 56 84 56 88 58 88 50 90
2 20 82 20 80 18 96 18 88 20 94
3 48 88 40 84 42 96 42 96 44 96
4 58 76 56 80 56 80 54 88 56 86
5 46 8O 46 92 48 78 46 84 46 84
6 8 50 6 50 8 34 2 28 8 66
7 56 68 46 76 52 74 52 68 48 76
8 12 50 8 44 12 48 10 44 12 46

Conduc-

tive

Subjects
1 44 94 46 84 50 92 48 100 44 94
2 30 96 30 92 32 98 32 94 32 92
3 28 96 32 94 28 90 28 98 24 92
4 26 96 24 98 26 98 24 98 26 100
5 6 92 6 90 4 88 6 90 4 90
6 30 96 34 98 32 96 34 94 30 100
7 32 98 32 96 28 100 32 98 32 96
8 22 90 22 90 20 86 22 90 20 86

 

*In dB re Audiometric zero.
#In percent correct.

 

177

APPENDIX G

PURE-TONE AIR-CONDUCTION THRESHOLDS*
OBTAINED IN EXPERIMENT III

Frequency in Hertz

 

 

500 1000 2000
Normal Artic.
Subjects ORIGINAL
1 15 0 5
2 15 15 0
3 20 10 5
4 20 5 10
5 15 10 15
6 20 5 10
7 20 10 0
8 5 0 5
ACTUAL
l 10 0 0
2 10 5 0
3 20 0 0
4 5 0 0
5 5 0 10
6 20 0 0
7 10 5 0
8 5 0 5
BETTER

l 0 0 5
2 15 5 0
3 15 0 0

.1:- '-r. “a." Wa‘t5‘3‘ - r

 

178

APPENDIX G (Cont.)

Frequency in Hertz

 

500 1000 2000
Normal Artic.
Subjects BETTER (Cont.)
4 15 0 5
5 15 5 15
6 20 10 0
7 10 5 0
8 5 0 5
POORER
1 10 0 0
2 20 5 5
3 20 0 0
4 15 0 0
5 15 10 15
6 15 5 5
7 15 5 0
8 0 0 0
NO INFORMATION

1 10 0 0
2 10 5 0
3 15 0 0
4 10 0 5
5 10 0 10

6 15 10 0

 

 

 

 

APPENDIX G (Cont.)

179

Frequency in Hertz

 

500 1000 2000
Normal Artic.
Subjects NO INFORMATION (Cont.)
7 10 10 0
8 10 5 0
Defective Artic.
Subjects ORIGINAL
l 10 0 5
2 5 5 5
3 10 5 0
4 15 10 10
5 15 5 5
6 10 10 10
7 15 0 0
8 5 5 0
ACTUAL
l 5 0 5
2 10 5 10
3 15 10 5
4 10 0 5
5 10 0 0
6 15 10 0
7 10 0 0
8 0 0 0

 

 

 

 

180

APPENDIX G (Cont.)

Frequency in Hertz

 

500 1000 2000
Defective Artic.
Subjects BETTER
1 5 0 0
2 10 5 5
3 10 10 5
4 10 10 10
5 0 0 0
6 0 0 0
7 0 0 0
8 5 0 0
POORER

l 15 10 15
2 10 0 0
3 15 10 10
4 5 5 0
5 15 5 5
6 5 10 10
7 15 0 0
8 5 5 0

NO INFORMATION

1 15 0 5
2 10 0 0
3 10 10 0

 

APPENDIX G (Cont.)

Frequency in Hertz

 

500 1000 2000
Defective Artic.
Subjects
5 15 0 0
6 5 10 10
7 15 0 0
8 10 5 0

 

*In dB re ISO 1964

 

 

 

 

 

 

APPENDIX G (Cont.)

182

SPEECH RECEPTION THRESHOLDS* AND

SPEECH DISCRIMINATION SCORES#
OBTAINED IN EXPERIMENT III

 

 

Condi-
tion Original Actual Better Poorer No Info.
SRT PB SRT PB SRT PB SRT PB SRT PB

Normal

Artic.

Subjects
1 8 98 6 100 4 96 4 100 2 90
2 6 98 8 92 6 98 8 100 6 96
3 10 98 16 90 16 94 12 100 10 100
4 6 100 8 100 6 100 6 96 4 100
5 10 100 0 80 4 100 10 100 8 100
6 6 98 8 100 8 96 6 98 6 96
7 8 96 6 100 8 100 6 100 8 100
8 8 100 8 100 4 100 12 98 6 100

Defective

Artic.

Subjects
1 6 96 4 80 6 86 4 96 12 76
2 4 96 8 96 4 100 4 100 10 100
3 6 88 12 60 12 80 16 90 8 96
4 6 92 10 94 6 92 6 92 6 94
5 6 90 4 92 10 100 10 78 8 88
6 4 100 6 96 8 98 6 100 8 100
7 2 94 6 98 0 94 4 96 4 90
8 2 96 2 94 2 98 2 96 0 96

 

*In dB re Audiometric zero.

#In percent correct.

 

 

183

APPENDIX H

PURE-TONE AIR-CONDUCTION THRESHOLDS*

OBTAINED IN EXPERIMENT IV

Frequency in Hertz

 

250 500 1000 2000 4000
Sensorineural

Subjects ORIGINAL
1 25 15 5 45 70
2 25 30 30 30 40
3 55 55 35 50 75
4 15 15 15 30 55
5 15 30 30 35 75
6 20 20 20 30 65
7 25 30 30 30 80
8 5 5 0 35 50

ACTUAL
: 1 20 15 o 40 65
2 10 15 35 30 45
3 40 45 40 50 85
4 20 20 20 30 55
5 20 25 25 40 75
6 20 20 15 20 55
7 30 25 25 25 85
8 0 5 0 35 50

BETTER
1 15 15 0 45 65
2 10 20 35 25 40
3 45 45 40 45 75

 

 

 

 

184

APPENDIX H (Cont.)(Air-conduction thresholds)

Frequency in Hertz

 

250 500 1000 2000 4000
Sensorineural
Subjects BETTER (Cont.)
4 15 15 10 30 50
5 30 30 25 40 75
6 35 25 20 20 60
7 30 30 25 30 80
8 0 5 0 35 55
POORER
1 5 15 5 45 65
2 10 20 30 25 35
3 35 40 40 50 80
4 20 20 10 25 55
5 15 30 30 35 75
6 25 20 15 20 55
7 30 35 30 30 85
8 0 5 0 35 50
NO INFORMATION

1 25 10 0 45 65
2 20 20 30 25 35
3 40 40 40 60 85
4 20 15 10 20 55
5 20 30 30 40 75

6 20 20 20 20 65

 

185

APPENDIX H (Cont.) (Air-conduction thresholds)

Frequency in Hertz

 

250 500 1000 2000 4000
Sensorineural
Subjects NO INFORMATION (Cont.)
7 25 30 30 35 85
8 5 5 0 35 50
Conductive
Subjects ORIGINAL
1 65 70 45 40 60
2 55 55 30 25 20
3 55 50 55 50 35
4 25 15 5 5 10
5 20 25 20 15 45
6 65 60 55 45 40
7 45 40 30 20 45
8 45 45 30 30 15
ACTUAL
l 60 70 50 40 65
2 45 45 30 30 20
3 50 50 60 45 30
4 15 5 0 -5 0
5 25 25 15 15 40
6 55 60 60 40 40
7 45 40 30 20 45

 

.L
g-

 

     
  

   
     

/

/ h /,
/

‘.
. \o
". ‘ .; In .
«15...: ‘ _¢l_ ,

         

 

1. .-

I
.' -
.
I
I ‘ g '.
I V,
x ‘ ‘4 .. . , . _
’3‘. "'1 ;_ I H" '
en "‘ . ;' .fl . _ ‘ .1 . z

OOOOO

186

APPENDIX H (Cont.) (Air-conduction thresholds)

Frequency in Hertz

 

250 500 1000 2000 4000
Conductive
Subjects BETTER
1 55 65 50 45 60
2 50 55 35 25 15
3 45 60 65 4o 45 ‘
4 20 10 5 o 5 i
5 15 20 15 20 45 '
6 60 60 55 3o 35
7 45 4o 35 20 4o
8 45 4o 25 25 10
POORER
1 65 65 45 35 6O
2 55 55 3o 25 2o
3 45 so 45 45 4o
4 20 10 5 o 15
5 20 3o 15 15 45
6 65 60 55 30 4o
7 35 4o 35 20 4o
8 45 4o 20 25 15
NO INFORMATION

1 60 7o 55 50 55
2 50 45 25 20 15
3 45 50 70 4o 40

4 10 5 0 0 5

 

 

 

 

187

APPENDIX H (Cont.)(Air-Conduction Thresholds)

Frequency in Hertz

 

250 500 1000 2000 4000
Conductive
Subjects NO INFORMATION (Cont.)
5 15 20 5 10 45
6 50 50 55 25 35
7 45 45 30 20 55 l‘
8 40 35 25 25

10 1

 

*In dB re ISO 1964

 

 

 

APPENDIX H (Cont.)

188

PURE-TONE BONE-CONDUCTION THRESHOLDS*

OBTAINED IN EXPERIMENT IV

Frequency in Hertz

 

250 500 1000 2000 4000
Sensorineural
Subjects ORIGINAL
1 20 15 10 40 65
2 20 35 30 25 45
3 40 50 35 45 65
4 25 15 20 25 55
5 10 25 30 30 65
6 20 20 15 30 60
7 0 25 25 25 65
8 10 25 0 35 50
ACTUAL

15 20 10 40 65
2 20 35 30 20 45
3 35 45 35 45 65
4 20 20 15 25 55
5 15 30 25 40 60
6 10 15 15 25 65
7 5 25 20 25 65
8 10 20 0 35 45

BETTER

l 15 15 5 45 65
2 20 30 35 20 45
3 30 50 35 45 65

 

 

 

 

 

 

 

 

 

 

 

APPENDIX H(Cont.)

189

 

(Bone-conduction thresholds)

Frequency in Hertz

 

250 500 1000 2000 4000
Sensorineural
Subjects BETTER (Cont.)
4 25 20 25 25 55
5 15 25 30 30 65
6 10 20 20 20 60
7 5 25 20 30 65
8 10 25 0 40 50
POORER
1 15 20 10 40 65
2 20 35 30 20 50
3 30 50 30 45 65
4 20 25 10 15 . 60
5 10 30 20 40 65
6 10 20 15 30 65
7 5 30 25 25 65
8 10 25 0 45 50
NO INFORMATION

1 15 20 10 40' 65
2 20 35 25 25 45
3 30 50 30 45 65
4 20 20 15 20 55
5 15 35 30 40 6S
6 20 20 15 30 60

 

 

I '. a a
I 7 ' 3L1 ' .

 

190

APPENDIX H (Cont.) (Bone-conduction thresholds)

Frequency in Hertz

 

 

 

250 500 1000 2000 4000
Sensorinaural
Subjects NO INFORMATION (Cont.)
7 5 25 20 30 65
8 15 30 5 35 40
Conductive .‘
Subjects ORIGINAL
1 0 15 25 20 45
2 10 25 10 0 0
3 0 30 25 20 20
4 5 5 -5 -10 0
5 10 20 10' 5 30
6 0 5 10 10 10
7 5 15 15 5 20
8 15 25 10 0 0
ACTUAL
l 0 15 20 15 40
2 5 25 15 0 0
3 0 30 25 20 20
4 5 5 0 -10 0
5 0 15 5 0 20
6 0 0 5 0 0
7 5 10 10 10 20

 

 

 

 

 

 

191

APPENDIX H (Cont.) (bone-conduction thresholds)

Frequency in Hertz

 

250 500 1000 2000 4000
Conductive
Subjects BETTER
l 0 20 20 15 40
2 15 30 15 0 0
3 5 25 30 25 15
4 0 5 -5 -10 0
5 5 20 15 0 25
6 5 5 20 5 10
7 10 20 15 5 20
8 15 20 10 0 10
POORER
1 0 20 40 30 40
2 0 30 15 0 0
3 0 25 25 25 25
4 5 0 -5 -10 0
5 15 20 10 0 20
6 5 10 20 10 20
7 5 20 15 0 20
8 10 25 10 0 15
NO INFORMATION

1 0 25 25 15 45
2 0 20 15 0 0
3 0 25 30 25 15

 

 

 

 

 

 

 

APPENDIX H (Cont.)

192

(bone-conduction thresholds)

Frequency in Hertz
250 500 1000 2000 4000

 

Conductive
Subjects

5
6

NO INFORMATION (Cont.)

5 15 5 0 20
O 0 15 25 10
10 15 S 10 25

15 20 15 0 15

 

*In dB re ISO 1964

 

 

 

 

 

 

193

APPENDIX I

A C0py of the Letter Sent to Each Adult
Subject Who Participated in This Study

We here at the Michigan State Universit Speech and
Hearing Clinic are conducting a researc project in the

area of hearing disorders.

Our records indicate that you have previously been given

a hearing evaluation and that the evaluation indicated
that you have a hearing loss. The project that we are
conducting involves patients who have hearing losses; and
the ultimate goal of our research is to determine a better
way of serving the hard-of—hearing pOpulation.

We would sincerely appreciate it if you would volunteer
as a subject.

If you do volunteer for this research it will mean: (1)
that you will have a free series of hearing tests, (2)

that you will have to come to the Michigan State University
Speech and Hearing Clinic for a two hour test session, and
(3) it is expected that the knowledge gained from this re-
search project will be of benefit to individuals with a
hearing loss in the future in that we are develOping new
methods for measuring and evaluating hearing impairments.

Inga few days you will be receiving a phone call from our
clinic to determine if you will be willing to participate
in this very important research study.

Sincerely,

William F. Rintelmann, Ph.D.
Associate Professor
Nicholas M. Hipskind
Clinical Assistant

f

 

 

 

 

 

 

 

 

 

 

IIIIIIIIIIIIIIIIIIIIIIIIIIIIIIII

"WWItmmmummluuwumuammnum

1293 030