0196/0202/90/1101/0056$02.

00/0
EAR A N D HEARING Vol. 1 I , No. I
Copyright 0 1990 by The Williams & Wilkins Co. Printed in U.S.A .

AMPLIFICATION AND AURAL REHABILITATION

A Critical Reevaluation of the Quantified Denver Scale of

Communication Function
Michael R. Tuley, Cynthia D. Mulrow, Christine Aguilar, and Ramon Velez

Health Services Research and Development Satellite Unit, Audie 1. Murphy Memorial Veterans’ Administration Hospital, San Antonio, Texas

ABSTRACT ministered in either a 25 item or 10 item version. It has

The Quantified Denver Scale of Communication Function (QDS) been recommended for use as a screening tool (Lichten-
is a 25 item questionnaire developed to measure communication stien, Bess, & Logan, 1988) as well as for assessment of
difficulties in adults with hearing impairment. This study reas- rehabilitation success (Ventry & Weinstein, 1982).
sessed the constructs, reliability, and validity of the scale, and Another measure is the Denver Scale of Communica-
developed a 5 item short version. The QDS was administered to tion Function, a 25 item scale which was specifically
238 elderly individuals (137 with and 101 without hearing loss). designed to measure communication dificulties that
Factor analysis using this sample identified only two subscale adults have as a result of hearing loss (Alpiner, Chevrett,
constructs as opposed to four originally proposed constructs. Glascoe, Metz, & Olsen, 1974). Early study of this measure
The validity of the new revised two-construct model was verified
by four independent investigators who labeled the two con-
in 10 middle-aged, hearing-impaired individuals revealed
structs as measuring self isolation and communication function. an overall scale test-retest reliability of 0.729 and individ-
The internal reliability of the revised scale was 0.97 and of both ual question test-retest reliabilities greater than 0.7
construct subscales was 0.95. Overall test-retest reliability was (McNeill, 1975). Kaplan, Feely, and Brown (1978) ex-
0.73. Validity examined by comparing the revised scale with panded the scale and reevaluated the test-retest reliability
another well-known handicap measure, the Hearing Handicap in 12 senior citizens, age 72 to 90. Overall reliability was
Inventory for the Elderly, was adequate: overall scale correlations 0.88 and question reliabilities ranged from 0 to 0.947. In
were 0.73 and subscale correlations ranged from 0.64 to 0.72. 1980, Schow and Nerbonne quantified a 25 item version
The accuracy of the revised QDS for discriminating between of the measure and found that increasing handicap as-
individuals with and without hearing loss was 73%. Stepwise sessed by the Quantified Denver Scale of Communication
discriminant analysis generated a 5 item short version scale
which contained two questions from the long communication
Function (QDS) was correlated with increasing PTA ( Y =
subscale and three from the long self-isolation subscale. The 0.58) hearing loss.
accuracy of the short QDS was 74%. We conclude that the The QDS warrants further study because it was specifi-
revised QDS is a reliable and valid scale that can be used to cally developed to measure communication dificulties,
assess self isolation and communication function in elderly indi- an area not well addressed by other handicap scales such
viduals with hearing loss, and that a new 5 item short version as the HHIE, and because early small pilot studies reported
performs as well as the original 25 item scale. reasonable reliability estimates and reasonable correla-
tions with hearing loss. With this in mind, we conducted
a larger study to: (1) reevaluate the exact areas or con-
structs measured by the QDS, (2) reexamine the reliability
and validity of the QDS, and (3) determine whether a
A variety of scales have been developed to measure short version of the QDS could be developed that could
different types of handicap in elderly individuals with be used as an efficient screening tool.
hearing loss (Giolas, Owens, Lamb, & Schubert, 1979;
High, Fairbanks, & Glorig, 1964; Noble & Atherley, 1970;
Schow & Nerbonne, 1982). A commonly used scale is the METHOD
Hearing Handicap Inventory for the Elderly (HHIE) which Participants for this study were recruited from the General
was designed to measure the social and emotional handi- Medicine Clinic at the Audie L. Murphy Memorial Veterans’
cap that elderly people with hearing impairment experi- Administration Hospital in San Antonio. Individuals over age
ence (Ventry & Weinstein, 1982). This scale can be ad- 64 who were not current hearing aid users who attended the

56
 Critical Reevaluation of the QDS Function 57

clinic during June 1987-April 1988 were screened for hearing within subscales did not appear to be measuring the same con-
impairment. All individuals who were found to be hearing im- struct. In particular, the 15 original family and social-vocational
paired ( n = 137) as well as an approximate 25% random sample items were confusing and were often labeled by the observers as
of individuals who were not hearing impaired ( n = 101) were measuring self or general communication function. Because of
included in the study. Characteristics of these patients ( n = 238) this identified substantial overlap, the observers proposed that
are presented in Table 1. fewer than four constructs were actually being measured by the
Hearing impairment was assessed using the Welch-Allyn au- questionnaire. Factor analysis confirmed that the original sub-
dioscope, a hand-held otoscope combined with an audiometer scales were not valid. Eleven of the 25 items had higher loadings
that delivers a 40 dB tone at frequencies of 500, 1000, 2000, and on subscales other than those originally proposed (Table 2).
4000 Hz (Frank & Petersen, 1987). For the purposes of this Subscale Development New subscales were then developed.
study, hearing impairment was defined conservatively as a better Exploratory factor analysis was performed using SAS. A scree
ear threshold of greater than 40 dB at 2000 Hz. All individuals plot was used to determine the number of subscales that should
who failed this screening test had their hearing loss confirmed be extracted (Cattell, 1978). Principal factor analysis with an
with pure-tone threshold tests conducted in audiometric test oblique rotation, a type of exploratory factor analysis, was used
booths by ASHA certified audiologists. to determine which items belonged in each subscale. An oblique
The QDS and the HHIE were given to study participants or nonorthogonal rotation was used because it was assumed that
before they were informed of the results of their hearing tests. the constructs would be correlated. The validity of the new
The QDS is a 25 item self-administered questionnaire which was subscales was tested by a multiple-groups confirmatory factor
originally reported to consist of four subscales. The original analysis using A Statistical Package for ITem ANalysis; ITAN
subscales were labeled family, self, social-vocational, and general (Gerbing & Hunter, 1987). The multiple-groups model assumes
communication function and consisted of 4, 5 , 11, and 5 items, unidimensionality, that is, that each construct is measured by
respectively. The measure uses a Likert scale that ranges from multiple items and that each item is part of only one construct
“1” for strongly disagree to “5” for strongly agree. Overall percent (Hunter & Gerbing, 1982). The fit of the model is then tested by
scores range from 0 to 100, where 100 suggests the individual comparing the predicted correlations generated from the multi-
perceives a significant handicap and 0 suggests no perceived ple-groups model to the observed correlations from the sample.
handicap. The HHIE is also a 25 item self-administered ques- (Exploratory factor analysis is used to generate a model when no
tionnaire which includes a social and an emotional subscale. previous model is proposed. Confirmatory factor analysis is then
Items are scored on a Likert scale and total scores range from 0 used to test how well the model fits the actual correlations of the
to 100, with 0 representing no handicap. items that were generated from the sample.) After factor analysis,
Data Analyses
Preliminary Analyses Initially, several different methods Table 2. Confirmatory factor analysis pattern coefficients of the original
were used to reassess the exact domains measured by the QDS. quantified Denver Scale of Communication Function.
These methods followed psychometric guidelines for selection,
interpretation, and evaluation of hearing handicap scales (De- Question Original
morest & Walden, 1984). Content and construct validity were Number Construct Family Self Social General
examined through an interactive process of content review with 1 Family 74 58 56 50
factor analysis. Four independent observers examined the origi- 2 Family 77 73 71 64
nal subscales for content validity. (These observers were an 3 Family 74 76* 72 68
audiologist and three primary care physicians. All four had 4 Family 71 63 63 60
previous clinical and research experience with hearing-impaired 5 Self 73 80 76 67
individuals.) Each observer independently evaluated each ques- 6 Self 74 81 80 75
tion by labeling them as measuring either family, self, social- 7 Self 76 77 82* 81
vocational, or general communication constructs. All four found 8 Self 76 84 76 71
that the subscales lacked face validity, that is, individual items 9 Self 72 83 75 71
10 Social 75 73* 80 84
11 Social 78 85* 80 73
12 Social 66 79 79 75
Table 1. Demographic and clinical characteristics of study subjects.
13 Social 65 78 80 76
Age 71.6 k 5.5 14 Social 60 70 71 67
Men 99% 15 Social 60 64 72 76*
Ethnicity 16 Social 75 73 80 84*
White 64% 17 Social 79 79 85 85
Hispanic 29% 18 Social 76 81 88 89*
Married 80% 19 Social 69 66 73 76*
Education (years) 10.3 k 4.0 20 Social 65 72 80 78
Retired 92% 21 General 71 75 81* 80
Abnormal visual acuity 26% 22 GeneraI 73 82 84’ 83
No. of comorbid diseases *
1.5 1.o 23 General 44 55 61 64
No. of medications 4.0 k 3.0 24 General 50 56 61 64
Pure-tone average’ 25 General 68 70 80’ 78
(500, 1000, 2000) 36.6 2 9.4
Indicates a higher pattern coefficient on factor other than original
* PTA only for those subjects with hearing impairment. construct.

Ear and Hearing, Vol. 11, No. 1, 1990

58 Tuley et al.

the face validity of the new subscales was evaluated by the four
Table 3. Exploratory and confirmatory analysis pattern coefficients on
investigators.
the Quantified Denver Scale of Communication Function.
The reliability and validity of the revised QDS and its new
subscales were then tested. Internal consistency was assessed by Exploratory Analysis
Confirmatory
Chronbach’s alpha (Chronbach, 1970). Test-retest reliability us- Analysis
ing Pearson’s product moment correlation coefficient was com-
puted for 64 subjects who had the QDS readministered 6 weeks Question Original Self Communi- Self Communi-
after the baseline measurement. (These 64 subjects represented Number Construct Isolation cation Isolation cation
a subset of the original 137 individuals who were identified as 2 Family 0.65 0.14 0.75 0.66
hearing impaired. Their demographic characteristics were similar 3 Family 0.67 0.14 0.77 0.68
to the original sample and their average PTA was 35.3 7.7.) 5 Self 0.66 0.17 0.79 0.70
Criterion validity was evaluated by correlating the revised QDS 6 Self 0.61 0.26 0.80 0.75
and its subscales with the HHIE and its subscales. Receiver 8 Self 0.77 0.08 0.83 0.71
Operating Characteristic (ROC) curves were used to determine 9 Self 0.80 0.03 0.82 0.69
the accuracy of the revised QDS for discriminating individuals 10 Social 0.86 -0.12 0.75 0.58
with hearing loss from those without hearing loss (Swets & 11 Social 0.75 0.14 0.86 0.74
Pickett, 1982). 12 Social 0.78 0.08 0.82 0.71
Development of Short Questionnaire Finally, stepwise dis- 13 Social 0.75 0.12 0.81 0.72
criminant analysis (Jennrich, 1977) was used to develop a short 14 Social 0.49 0.28 0.68 0.67
version of the QDS. Alpha was set at 0.15 for both the forward 23 General 0.33 0.31 0.57 0.58
and backward stepping; this value was chosen to guard against 1 Family 0.19 0.46 0.56 0.58
failure to include a significant variable in the equation. (Discrim- 4 Family 0.01 0.69 0.56 0.68
inant analysis is a multivariate technique that is used to classify 7 Self 0.28 0.61 0.76 0.83
observations into separate populations; i.e., hearing impaired 15 Social 0.13 0.65 0.63 0.74
from hearing.) The accuracy for correctly diagnosing those sub- 16 Social -0.03 0.91 0.68 0.88
jects with hearing impairment by the long and short version of 17 Social 0.19 0.74 0.76 0.88
the revised,QDS was evaluated by comparing the areas under 18 Social 0.24 0.71 0.78 0.89
the estimated Receiver Operating Characteristic (ROC) curves 19 Social -0.02 0.83 0.62 0.79
using a Z test (Metz, Wang, & Kronman, 1984). The area under 20 Social 0.33 0.52 0.73 0.77
an ROC curve measures the fraction of a sample that is correctly 21 General 0.33 0.54 0.73 0.80
diagnosed by the measure. The ROC curves were estimated by 22 General 0.43 0.49 0.80 0.82
maximum likelihood estimation from five cut points that were 24 General 0.03 0.63 0.52 0.65
chosen so that an accurate curve could be fit to the observed 25 General 0.03 0.82 0.67 0.82
points on both the long and short versions (Dorfman & Alf,
1969).
items with loadings near zero contributing little or nothing
to that factor or subscale. In the exploratory factor analy-
RESULTS sis, 12 items were identified as measuring one construct
and 13 items as measuring a second construct. All items
Two constructs or subscales were developed from the had pattern coefficients of 0.33 or larger. Eleven questions
exploratory and confirmatory factor analyses. As the orig- in the first construct had been labeled as measuring self-
inal family and social-vocational constructs had been iden- isolation and one as measuring communication function
tified as significantly overlapping with the original self and (question 23) by the majority of investigators. Eleven of
general communication constructs, the four investigators the 13 items in the second construct had been labeled as
labeled the two final constructs as self-isolation and com- measuring communication function by the majority of
munication. They confirmed the face validity of these two investigators. The remaining two items (questions 7 and
constructs by independently evaluating whether each of 15) of the second construct had been labeled by three of
the 25 items were measuring either self-isolation or com- four investigators as measuring self-isolation rather than
munication function. The investigators judged that all communication. Therefore, construct one was thought to
items could be labeled under one of the two proposed most appropriately represent self-isolation and construct
constructs. Regarding categorization of individual items, two communication function.
the investigators reached complete agreement on 10 and The model generated by the exploratory analysis was
majority agreement (3 of 4) on 14. One item (question 1) tested by confirmatory factor analysis. In this analysis,
was thought by two investigators to represent communi- one item (23) included in the self-isolation subscale for
cation function and by-the other two investigators to the exploratory analysis now had a pattern coefficient that
represent self-isolation. was higher on the communication subscale. (Three of the
Exploratory and confirmatory factor pattern coefficients four investigators had agreed that this item was measuring
of the items are shown in Table 3. The pattern coeffi- communication and not self-isolation.) This change re-
cients are regression weights that represent the amount sulted in a final model with high pattern coefficients for
that an individual item contributes to the underlying each item; all were above 0.58. Individual item correla-
factor. The coefficients range from - 1.O to 1.O with those tions predicted from the model were compared to the

Ear and Hearing, Vol. 11, No. 1, 1990

 Critical Reevaluation of the QDS Function 59

actual observed item correlations revealing an average A 5 item short questionnaire was generated using a
magnitude of the residuals of 0.04 which confirmed a good forward and backward stepwise discriminant analysis
fitting model (Table 4). The average magnitude of the re- where alpha was set at 0.15 to guard against a type I error.
siduals for the original model was much higher (0.54), Five items from the original questionnaire are contained
again indicating a poorer fitting model. in the final short version (Table 5). These items listed in
The internal reliability for both the self-isolation and order of their loading within the short version model were
the communication function subscales measured by questions 1, 13, 19, 10, and 14. All entered the model at
Chronbach’s alpha was 0.95. Reliability for the total mea- p < 0.08. Questions 1 and 19 are from the communication
sure was 0.97. In the 64 subjects who had the scale subscale and 10, 13, and 14 are from the self-isolation
readministered at 6 weeks, the test-retest reliability was subscale. (The five questions can be summed to form a
0.73. The QDS and HHIE had a correlation of 0.73 ( p < score that ranges from 5 for no perceived handicap to 25
0.0 1). Correlations between the subscales of the QDS and for significant handicap.) The reliability measured by
HHIE ranged from 0.64 to 0.72. The correlation between Chronbach’s alpha for the overall short scale was 0.82 and
self-isolation and communication function subscales of for the communication and self-isolation subscales 0.67
the QDS was 0.83. The accuracy of the QDS for correctly and 0.80, respectively. The accuracy of the short question-
discriminating individuals with hearing loss from those naire was 74% which was not significantly different from
without loss was 73%. the long questionnaire ( p = 0.87, Fig. 1).

Table 4. Observed correlations and residuals from Quantified Denver Scale*

11 8 12 9 13 6 5 3 10 2 14 18 17 16 7 25 22 21 19 20 15 4 24 1 23
~

11 -7 3 -3 1 8 0 0 -1 3 -3 -2 -1 -8 3 -7 5 3 -7 2 0 -2 -5 4 0
8 65 -5 10 -4 -4 -1 1 1 1 9 0 -7 -10 4 -4 5 -1 -6 0 -8 -2 -6 8 5
12 74 63 -6 14 3 -2 -3 -1 -4 1 -2 -3 -9 4 -4 8 3 -11 9 1 -10 0 -3 10
9 67 78 61 -4 -3 3 4 5 -5 -1 0 -2 -10 -3 -6 6 3 -8 -3 -8 -5 -4 4
13 71 64 81 63 2 -4 -5 0 -1 1 3 0 -8 3 -1 10 6 -7 10 2 -10 -1 -5
6 77 * 6 3 69 63 ~~ 68 2 2 -5 0 -5 3 4 1 12 -1 12 6 -3 7 3 2 3 1
5 68 65 63 68 60 66 -5 2 2 3 0 5 -5 4 -6 4 -1 -5 7 -3 4 -3 6
3 66 65 61 67 58 64 57 3 9 -6 0 -2 -5 0 -3 6 4 -4 5 -4 3 -3 8
10 64 63 61 67 62 56 61 62 -6 2 -6 -10 -16 -6 -13 -3 -1 -12 0 -4 -10 -10 0
2 68 64 58 57 60 61 62 68 51 0 -1 4 -7 -2 -5 8 1 -3 0 -5 3 -9 16 2
14 56 66 57 55 57 50 57 46 53 51 9 3 3 7 6 6 3 1 12 10 0 1 5 15
18 65 65 62 64 67 66 62 60 53 58 62 7 -2 -2 0 8 2 0 -2 1 -4 -3 -2 -2
17 66 57 61 61 63 66 66 57 48 62 56 85 -1 -3 -2 3 3 -1 1 -4 1 -4 4 -4
16 58 54 54 52 54 62 55 ~~ 55 41 51 55 76 76 0 7 -4 -5 5 1 -2 4 -2 3 -4
7 66 64 64 57 62 70 62 56 49 53 56 72 70 73 -2 1 0 -3 -1 2 6 -2 5 1
25 55 56 56 53 58 57 51 53 42 50 55 74 71 79 66 0 -5 -1 0 5 -2 4 -3 -2
22 67 75 67 66 69 70 61 61 52 63 55 81 75 68 69 68 5 -2 1 -1 -10 -1 -2 5
21 64 58 61 61 64 63 54 59 52 54 51 74 74 65 67 61 71 -1 2 -1 -5 0 -1 6
19 53 51 46 50 50 53 50 50 40 50 48 71 69 75 62 65 63 63 4 -4 3 2 3 -4
20 60 56 64 52 64 61 60 57 51 51 58 67 69 68 62 64 65 64 66 1 -7 -2 -6 8
15 56 46 54 45 55 55 48 46 45 44 54 67 62 63 63 66 60 59 55 58 0 3 -6 5
4 49 48 39 44 38 50 51 49 35 4a 41 56 61 64 62 54 46 50 57 45 51 1 24 -1 1
24 44 41 47 43 46 48 42 41 32 33 39 55 53 55 52 58 52 52 53 48 51 45 -9 12
1 48 50 38 45 37 42 46 47 39 54 39 49 55 54 52 45 45 46 49 38 37 63 29 -1 0
23 43 47 51 43 50 43 45 40 42 39 49 49 47 46 49 46 52 52 42 52 48 28 49 23
* Observed correlations are below the diagonaland residuals (observed-predictedcorrelations) are above the diagonal.

Table 5. Short version of the Revised Quantified Denver Scale of Communication Function.*
Strongly Slightly Slightly Strongly
Disagree Disagree Neither Agree Agree
c-1. The members of my family are annoyed 1 2 3 4 5
(1) with my loss of hearing.
s-2. I tend to be negative about life in gen- 1 2 3 4 5
(10) era1 because of my hearing loss.
s-3. Since I have trouble hearing I hesitate 1 2 3 4 5
(1 3 ) to meet new people.
s-4. I do not enjoy my job as much as I did 1 2 3 4 5
(1 4) before I began to lose my hearing.
c-5. Conversationin a noisy room prevents 1 2 3 4 5
(19) me from attempting to communicate
with others.
* Capital letfers superceding question numbers refer to the subscale to which that question belongs: C indicates communication subscale
questions; S indicates self-isolationsubscale questions. Numbers in parentheses correspond to the original question in the long version of the scale.

Ear and Hearing, Vol. 11, No. 1, 1990

60 Tuley et al.

tions 7 and 15 were placed under communication function

because of their high correlations ( Y > 0.65) with questions
16, 17, 18, 21, 22, and 25; and 18 and 25, respectively
(Table 3). The consistent high correlations with other well-
established communication questions suggests that study
individuals were responding to questions 7 and 15 as if
they were measuring the same construct. For this reason,
we believe they should remain under the communication
construct.
The validity of the scale was further confirmed by

0.2 I/ evaluating how well it correlated with another measure

which was specifically developed to assess handicap in
elderly hearing-impaired individuals and by evaluating
how well it identified individuals with hearing loss. In the
former analysis, the correlation of the RQDS with the
other handicap measure, the HHIE, was 0.73. This cor-
0 1 I I I I 1 relation suggests that the RQDS is a valid measure for
0 0.2 0.4 0.6 0.8 1
1 - Specificity assessing hearing handicap and that it is assessing handicap
areas that are different from the areas assessed by the
Figure 1. Receiver-operating curves for the long and short versions of HHIE. The HHIE specifically addresses emotional and
the QDS. The five points used to estimate the QDS curve were 50, 60, social handicap, and the RQDS addresses self-isolation
70, 80, and 90. The five points used for the QDS short curve were 11,
13, 15, 18, and 20.
and communication function. Moreover, a combination
of the two measures may be important to consider when
a comprehensive evaluation of hearing handicap is indi-
DISCUSSI0N cated. Finally, the overall accuracy of the RQDS for
correctly identifying individuals with hearing loss was
The results of this study suggest that the revised Quan- 73%. As handicap is not synonymous with hearing im-
tified Denver Scale of Communication Function (RQDS) pairment, a higher degree of accuracy was not expected.
is a reliable and valid measure for assessing hearing hand- The observed accuracy is consistent with a valid measure
icap in elderly individuals. The original scale was devel- for assessing handicap secondary to hearing impairment
oped to assess four areas of function which might be (Ventry & Weinstein, 1982).
affected by hearing impairment. These included family, The reliability of the RQDS was tested in two ways. A
self, social-vocational, and general communication abili- point in time estimate of internal consistency using Chron-
ties. After careful review of individual items, the face bach’s alpha was very high at 0.95. The test-retest reliabil-
validity of these four subscales was questioned by the four ity measured in 64 subjects after 6 weeks was lower at
investigators. Exploratory and confirmatory factor anal- 0.73. This estimate is typical of reliability estimates pre-
yses were undertaken to reevaluate the actual constructs viously reported for hearing handicap, self-report meas-
measured by the scale. These analyses showed two con- ures (Demorest & Walden, 1984). It may be biased down-
structs rather than four were being assessed by the scale. ward because of the long 6 week interval which elapsed
The face validity of these two constructs was verified by before the retesting. Moreover, the difference between the
the investigators who identified the two new subscales as point in time estimate of reliability and the test-retest
measuring self-isolation and communication function. estimate may reflect other changes in health status which
The original social-vocational and family subscale ques- may have occurred during the 6 week interval between
tions were evenly split between the two new self-isolation the test and retest (Demorest & Erdman, 1988). Regard-
and communication function subscales, which suggested less, both estimates are sufficiently high to suggest that the
that the original social-vocational and family categories scale is reliable enough for use in analyses of groups and
were unnecessary. of individuals. Furthermore, it appears stable enough over
The resulting simplified scale is superior to the old time to be used as an outcome measure in before-after
version because it represents a model that includes both studies.
content and statistical validity. The content validity and Lastly, a short version of the RQDS was developed
the statistical validity were congruent for all items except which consists of only 5 items as opposed to the original
three (questions 1, 7, and 15). Two investigators labeled 25. The short version includes 2 communication items
question 1 as measuring selfiisolation while the other two and 3 self-isolation items and thus can be used to measure
and the statistical model placed question 1 under com- both subscale areas. The accuracy of the short version for
munication function. Questions 7 and 15 were labeled by discriminatingbetween individualswith and without hear-
the majority of investigators as measuring self-isolation, ing impairment was approximately the same as the accu-
whereas the statistical model placed these questions under racy of the original 25 item version. These preliminary
communication function. In the statistical model, ques- results suggest that a short version RQDS is possible.

Ear and Hearing, Vol. 11, No. 1, 1990

 Critical Reevaluation of the QDS Function 61

In summary, we have utilized a variety of methods to Giolas T, Owens E, Lamb S, and Schubert E. Hearing performance inventory. J
Speech Hear Disord 1979;44:169-195.
demonstrate that the RQDS is a reliable and valid measure High W, Fairbanks G, and Glorig A. A scale for self assessment of hearing handicap.
that can be used to assess self-isolation and communica- J Speech Hear Disord 1964;12:215-230.
Hunter JF and Gerbing DW. Unidimensional measurement, second order factor
tion function in elderly individuals with hearing loss. We analysis and causal models. Res Organ Behav I982;4:267-320.
suggest that it be used along with additional handicap Jennrich RI. Stepwise discriminant analysis. In Enslein K, Ralston A, and Wilf H,
measures when a comprehensive evaluation of dysfunc- Eds. Statistical Methods for Digital Computers. New York: John Wiley and Sons,
1977.
tion secondary to hearing loss is needed. A new short Kaplan H, Feely J, and Brown J. A modified denver scale: test-retest reliability. J
version RQDS which is as accurate as the original long Acad Rehab Audio1 1978;11:15-32.
version may be useful as an efficient screening tool. Ad- Lichtenstien MJ, Bess FH, and Logan SA. Diagnostic performance of the hearing
handicap inventory for the elderly (screening version) against differing definitions
ditional research aimed at evaluating the relative ability of hearing loss. Ear Hear 1988;9208-211.
of these versions to detect small but clinically important McNeill MR. A study of the reliability of the denver scale of communication
function. Unpublished study, University of Denver, 1975.
changes in hearing handicap is needed. Metz CE, Wang P, and Kronman HB. A new approach for testing the significance
of differences between roc curves measured from correlated data. In Deconik F,
Ed. Information Processing in Medical Imaging. Boston: Martinus Nishoff, 1984:
References 432-445.
Noble WG and Atherley GRC. The hearing measurement scale: a questionnaire
Alpiner J. Chevrett W, Glascoe G, Mets M, and Olsen B. The denver scale of for the assessment of auditory disability. J Aud Res 1970;10:193-214.
communication function. Denver: University of Denver, 1974. Schow RL and Nerbonne MA. Hearing handicap and denver scales: applications,
Cattell RJ. The Scientific Use of Factor Analysis in Behavioral and Life Sciences. categories, interpretation. JARA 1980;13:66-77.
New York: Plenum Press, 1978. Schow R and Nerbonne M. Communication screening profile: use with elderly
Chronbach U. Essentials of Psychological Testing, 3rd ed. New York: Harper and clients. Ear Hear 1982;3:133-147.
Row, 1970. Swets JA and Pickett RM. Evaluation of diagnostic systems: methods from signal
Demorest ME and Erdman SA. Retest stability of the communication profile for detection theory. New York: Academic Press, 1982.
the hearing impaired. Ear Hear 1988;9:237-242. Ventry IM and Weinstein BE. The hearing handicap inventory for the elderly: a
Demorest ME and Walden BE. Psychometric principles in the selection, interpre- new tool. Ear Hear 1982;3:128-134.
tation, and evaluation of communication self-assessment inventories. J Speech
Hear Disord 1984;49:226-240. Acknowledgments: We thank James E. Endicott, Ph.D., Mary Gaul-
Dorfman DD and Alf E. Maximum-likelihood estimation of parameters of signal- Amatto, Pharm. D., and Helen Hazuda, Ph.D. for the computer program-
detection theory and determination of confidence intervals-rating-method data. rning and the editing assistance in the development of this manuscript.
J Math Psycho1 1969;6:487-496.
Frank T and Petersen DR. Accuracy of a 40 dB HL audioscope and audiometer Address reprint requests to Michael R. Tuley, Ambulatory Care (IIC),
screening for adults. Ear Hear 1987;8:180-183. Audie L. Murphy Memorial Veterans Administration Hospital, 7400 Mer-
Gerbing DW and Hunter JE. ITAN: A statistical package for item analysis. ton Minter Blvd., San Antonio, TX 78284.
(Available from D. W. Gerbing; School of Business Administration, Portland
State University, Portland, OR 97207-075 l), 1987. Received January 25, 1989; accepted July 25, 1989.

Ear and Hearing, Vol. 11, No. 1, 1990