MSU

LIBRARIES

V

 

 

\.
"‘ ". d . “‘23—.

RETURNING MATERIALS:
P1ace in book drop to
remove this checkout from
your record. FINES will
be charged if book is
returned after the date
stamped be10w.

 

 

fi

0,13?" 9 gr»
fit“ “wt-+913"

r'ficiv"
FV" J

. I

i .

 

 

 

LEGIBILITY OF CHILD RESISTANT BOTTLE
CAPS BY TWO METHODS OF MEASUREMENT

By

Mutune Wa Gitau

A THESIS

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

MASTER OF SCIENCE

School of Packaging

1988

ABSTRACT

LEGIBILITY OF CHILD RESISTANT BOTTLE CAPS
BY TWO METHODS OF MEASUREMENT

By

Mutune Wa Gitau

As a result of difficulties encountered by consumers
and physicians in reading the instructions or graphics on
medical labeling, this study was undertaken to determine a
method of measuring legibility. Child resistant bottle caps
were used as the medium. Two instruments, a polariscope and
a DMR visibility meter were used to determine legibility
differences among several variables such. as ‘print color
contrast, print size and visual acuity of the observer.

The comparison was determined by evaluating eight caps
with two different instruction messages using twenty
consumers as evaluators and a ten expert panel as a
reference control. The experts’ ratings were correlated
with those of the consumers using the two instruments.

Results indicated that color contrast, type size and
visual acuity of the observer significantly influenced
legibility of the two messages on these caps. There was a
high correlation between the experts’ ratings and the
consumers readings using the 'two instruments. Both

instruments can, therefore, be used to evaluate legibility.

This thesis is dedicated to my wife, Wangari, and daughter,
Wamaitha. Without their love, continual support, positive
encouragement and patience this thesis would not have been
possible. To my parents, for the foundation they gave me

through effort and hard work in putting me through school.

iii

ACKNOWLEDGMENTS

The author would like to express his appreciation and
gratitude to the following individuals:

Dr. Hugh Lockhart, School of Packaging, Michigan State
University, for his advice and guidance as major
professor.

Dr. Susan Selke, School of Packaging, Michigan State
University, for her help and direction and serving as a
committee member.

Mr. Joseph Kuszai, Art Department, Michigan State
University, for serving as a committee member.

Mr. Glen Pettit, School of Packaging, Michigan State
University, for his ability to organize and contact the
expert panel members for the legibility test.

Mr. Francis P. Tobolski, Design and Market Research
Laboratory (DMR), Carol Stream, Illinois, for kindly
allowing us to use DMR visibility meter for this study.

Dr. John Gill, Animal Science Department, Michigan State
University, for his guidance in experiment design and
statistical analysis methods.

Mr. Joshua G. Bagaka, Urban Studies Department, Michigan
State University, for his assistance in analyzing the
data. His countless hours of this work and genuine
concern will always be remembered.

Mr. Raymond. Cheung, School of’ Packaging, Michigan State
University, for his assistance in analyzing the data,
especially during a time of need.

Finally, special thanks to the students, staff members of
the School of Packaging, Michigan State University, and
the expert panel members who spared their time and
agreed to participate in this test. Last, but not
least, U.S.A.I.D. for the full scholarship granted for
this program.

iv

LIST OF

LIST OF

LIST OF

CHAPTER
1.

2.

MATERIALS AND APPARATUS.
3

TEST PROCEDURE . . . . . . . . . . .

TABLE OF CONTENTS

TABLES O O O O O O O O O O O O O O O O O 0
ILLUSTRATIONS O O O O O O O O O O O O O O O
FIGURES O O O O O O O O O O O O O O O I O 0
INTRODUCTION . . . . . . . . . . . . . . .
REVIEW OF LITERATURE . . . . . . . . . . .
2.1 Distance Method. . . . . . . . . . .
2.2 The Tachistoscope. . . . . . . . . .
2.3 The Focal Variator Method. . . . . .
2.4 Rate of Involuntary Blinking Method.
2.5 The Eye Movement Test. . . . . . . .
2.6 The Polariscope. . . . . . . . . . .
2.7 Luckiesh-Moss Visibility Meter . . .

.1 Materials. . . . . .
3.2 Major Apparatus. . .
3.3 Other Requirements .

4.1 Introduction . . . . . . . . .
4.2 Polariscope Procedure. .
4. 3 DMR Visibility Meter Procedure
4. 4 Expert Panel Procedure . . . .

DISCUSSION OF RESULTS. . . . . . . . . . .

5.1 Introduction . . . . . . . . . . . .
5.2 Experts’ Ratings and Consumer
Readings’ Correlation. . . . . . .
5.3 Polariscope and DMR Visibility Meter
Correlation. . . . . . . . . . . .
5.4 Effect of Message, Color Contrast,
Type Size and Visual Acuity on
Legibility . . . . . . . . . . . .

SUMMARY AND RECOMMENDATIONS. . . . . . . .

Page

vii

ix

70

77

91

Table of Contents (Continued)

LIST OF REFERENCES . . . . . . . . . . .

APPENDICES

A.
B.
C.
D.

EXPERT DATA 0 O I O I O I I O O O
CONSUMER DATA. . . . . . . . . .
RELIABILITY AND ANOVA PROGRAMS .

LEGIBILITY TEST DATA FOR 56- 76+ CONSUMER

AGE GROUP USING THE POLARISCOPE.

vi

Page

96

99
106
119

122

LIST OF TABLES

Page
Table

1. Polariscope Screen Rotation . . . . . . . 31
2. DMR Visibility Meter Wedge Rotation . . . 32

3. Cap Order of Presentation to the
Consumer for Both Polariscope and DMR
Visibility Meter. . . . . . . . . . . . . 43

4. Cap Order of Presentation to the
Experts Panel . . . . . . . . . . . . . . 50

5. Legibility Rating by Expert Panel vs.
Consumer 20/20 Readings on Polariscope
in Order of Ranking . . . . . . . . . . . 59

6. Legibility Rating by Expert Panel vs.
Consumer 20/30 Readings on Polariscope
in Order of Ranking . . . . . . . . . . . 60

7. Legibility Rating by Expert Panel vs.
Consumer 20/20 Readings on DMR Visibility
Meter in Order of Ranking . . . . . . . . 61

8. Legibility Rating by Expert Panel vs.
Consumer 20/30 Readings on DMR Visibility
Meter in Order of Ranking . . . . . . . . 62

9. Legibility Readings on Polariscope vs.
Readings on DMR Visibility Meter by
Consumer 20/20 in Order of Ranking. . . . 63

10. Legibility Readings on Polariscope vs.
Readings on DMR Visibility Meter by
Consumer 20/30 in Order of Ranking . . . . 64

11. Correlation Values and Statistical
Significance for Expert Panel Ratings and
Consumer Panel Machine Readings (Horvath
Test of Significance with Critical r =
0.707) . . . . . . . . . . . . . . . . . . 67

vii

List of Tables (Continued) Page

Table

12.

13.

14.

15.

16.

17.

18.

19.

20.

21.

22.

Main Treatments and Treatment
Combinations. . . . . . . . . . . . . . . 78

ANOVA Table for the Significant Main
EffeCtB 0 0 0 O 0 0 0 0 0 0 O 0 O O 0 0 O 79

Results of Tukey’s HSD Test for
Comparison of the Means of the Main
Effects for the Experts. . . . . . . . . . 82

Results of Tukey’s HSD Test for
Comparison of the Means of the Main
Effects for the Consumers Using P-sCOpe. . 83

Results of Tukey’s HSD Test for
Comparison of the Means of the Main
Effects for the Consumers Using DMR Meter. 84

Legibility Ratios of Means for Pairs of
variables 0 0 O O 0 O O O 0 O O O O O O I O 86

Type Size Legibility Ratios for Message I
and II means I O 0 O O 0 0 O O O 0 O 0 O O 87

Raw Data: Legibility Readings by 56-76+
Consumer Age Group Using the Polariscope . 124

Cap Text Legibility (6 pts) at Three
Levels of Visual Acuity (56-76+ years) . . 125

Cap Text legibility (8 pts) at Three
Levels of Visual Acuity (56-76+ years) . . 125

Cap and Bottle Symbol Legibility at Three
Levels of Visual Acuity (56—76+ Years) . . 126

viii

LIST OF ILLUSTRATIONS

Card (Dow

Hood. . .

ILLUSTRATION

1. Message I CRP Caps . . .

2. Message II CRP Caps. . .

3. Polariscope. . . . . . .

4. DMR Visibility Meter . .

50 Light Meter. 0 O O 0 O 0

6. Near Point Visual Acuity

7. Polariscope Covered by a

8. A Diagram Showing the Position of the
Observer in Relation to the P—sc0pe and
the TeSt‘Cap o o o o o o

9. Cap Presentation Order .

10. A Diagram Showing the Position of the
Observer Using DMR Visibility Meter in
Relation to the Test-cap o o o o o o o o

11. Legibility Test Samples.

ix

Page

27
29
30
33
35
36

40

42

44

46

127

Figure

LIST OF FIGURES

Bivariate Plot Showing Correlation
between the Experts and Polariscope by
Consumers with 20/20 Visual Acuity . . .

Bivariate Plot Showing Correlation
between the Experts and Polariscope by
Consumers with 20/30 Visual Acuity . . .

Bivariate Plot Showing Correlation
between the Experts and DMR Visibility
Meter by Consumers with 20/20 Visual
Acuity . . . . . . . . . . . . . . . . .

Bivariate Plot Showing Correlation
between the Experts and DMR Visibility
Meter by Consumers with 20/30 Visual
Acuity . . . . . . . . . . . . . . . . .

Bivariate Plot Showing Correlation
between the Polariscope and DMR
Visibility Meter by Consumers with 20/20
Visual Acuity. . . . . . . . . . . . . .

Bivariate Plot Showing Correlation
between the Polarisc0pe and DMR
Visibility Meter by Consumers with 20/30
Visual Acuity. . . . . . . . . . . . . .

Page

68

69

71

72

74

75

CHAPTER 1

mmoaucuou

The principal functions of a consumer package, briefly
stated, are: 1) to add utility or convenience to the
product, 2) protection of the product against all the
hazards and physical loss which may occur during the
distribution cycle, 3) protection for people against
dangerous or hazardous materials, and 4) communication.
While the first three functions essentially deal with the
technical aspects of packaging or its functional
performance, communication plays the major role of informing
the consumer about the product and in some cases about the
package, for example opening and closing instructions. Of
growing importance is the improvement of graphics for
marketing, and the emphasis on health and safety for the
consumer through voluntary and government standards and
regulations.

Communication on a package mainly appears as written
information or markings, e.g. symbols, signs, logos and
pictograms, which may have different interpretations
depending on the package and its targeted users. Some of

this form of communication may be required by regulations.

2

All the printed informational material that accompanies the
product is referred to as labeling, while the label is that
"labeling" that is affixed to the immediate container (25).
It is in this area of communication on the immediate
container where legibility of the written information is
critical. This is so because for written information to be
communicated efficiently, it has to be legible so that the
message can be read (and interpreted) correctly and absorbed
in the shortest time possible. The label on a primary
package that goes to the point of use is the most important
in this case. Factors such as typography, quality of print,
color, substrate texture, layout and the shape of the
package all convey the message or communication (14, 15).

Legibility as defined by Webster’s Unabridged
Dictionary is "the quality or state of being legible, . . .
the influence of typeface on the legibility of print" (26).
Miles. A. Tinker also defines legibility as concerned "with
perceiving letters, words or symbols; or the ease and speed
of reading continuous textual material" (22). Legibility
therefore deals with coordination of typographical factors,
inherent in letters, symbols and copy material, which affect
the ease and speed of reading and can only be defined in
terms of a specific method of approach to the study of the
problem.

The serious nature of label or package legibility has

been newly recognized in two major areas:

3
(i) WWWeflm - this is

concerned. with the legibility (readability) of use
instructions and warnings on the label. The labeling
requirements for hazardous materials are regulated by three
bodies: the US Department of Transportation (DOT) under the
Title 49 of the code of Federal Regulations (CFR) in parts
171-179, the Consumer Product Safety Commission (CPSC) Title
16 CFR part 1700 of the Poison Prevention Packaging Act, and
the Environmental Protection Agency (EPA) under regulation
40 CFR 162.10 and in section 162.16 regarding Child
Resistant Packaging (CRP).

(ii) Medical_lahgling - Fast, accurate identification
of drugs is critical in a crisis. Of particular concern
recently was the problem of medication errors in hospitals
by anesthesiologists during surgery which occurred due to
incorrect identification of injectable drug products (4).
This led to the formation of Subcommittee D10.34 on
Identification of Pharmaceutical Drug Pruduct Containers of
the American Society for Testing and Materials (ASTM). This
subcommittee of ASTM Committee D-10 on Packaging has
addressed the issue of legibility in the "Standard
Specification for Labels for Small-Volume (Less Than 100 Ml)
Parenteral Drug Containers, D4267-83" (1). This includes
label legibility requirements, legibility tests, type size
requirements and requirements dealing with the omientation
of the message on the container. The specification requires

a minimum type size of not less than legible 6-point type

4
for the proprietary or established name of the drug and the

numerals indicating the amount of drug per unit.

The legibility test in the specification requires that
the proprietary and established drug names, and the amount
of drug per unit, be legible in a light of 20 foot candles
at a distance of 19.7 inches (500 mm) to a person with 20/20
unaided or corrected vision. This test was chosen partly
because many drug labels were difficult to read and partly
because of the dim light found in hospitals. This
specification was expected to alleviate the previous problem
of identification in the hospitals.

The other area of increasing interest is the legibility
of child—resistant bottle caps. These caps are designed to
protect small children from a.dangerous product. However,
if the package labeling does not inform the adult how to use
the package, the protective function may be lost. So, for
lack: of effective' communication, the protective function
becomes inoperative. Child-resistant packaging caps (CRP)
in the market today need to be legible but many consumers
have reported difficulty reading the white-on-white print on
many bottle caps. This problem is particularly prevalent
among the elderly population. In its report on Innovative
Child Resistant Packaging Systems (16), the School of
Packaging reported that in the past, messages on child
resistant caps were often printed in colored ink. From
talks with caps producers and user companies, it was learned

that this practice was diminished by manufacturers because

5
of the cost of operation. and the poor quality of the

printing. However, industry members interviewed agreed that
the caps instructions and package graphics could be printed
economically by using non-contact printing methods, where
the printing ink would cost about one cent per pint. This
would produce an acceptable printing that would remain neat
and legible, avoiding the chipping or rubbing off the
package during shipment or repeated use, which has been a
problem.

The objective of this research was to devise a method
or system of measuring the legibility of labeling. Child
resistant bottle caps were chosen as the study medium. The

labeling on these caps was the main visual stimulus in this

test. To evaluate the legibility and to compare the
methods, the study involved two groups. One was an expert
panel (professional people working in graphics trades). The

other group was consumers in two age subgroups: 18-25 years
and 30-54 years. These two groups measured the legibility
of child resistant caps by the use of two mechanical devices
which were selected for this study--the polariscope and the
Design & Market Research Laboratory (DMR) visibility meter.
The results of the research are expected to help in
answering the following questions.
1. Do the instrument readings give the same result as the

experts?

6

How do the two instruments compare? The results will
help determine whether there is any significant
differences between the two devices used.

Effect of contrast and type size on the legibility of
the message.

Effect of visual acuity of the observer on the
legibility of the message.

The message content and understanding of meaning were

not measured, only the ability to detect and read the words

under a controlled amount of illumination.

CHAPTER 2

LIIEEAIQEEflELEE

A major goal of this work has been to establish a
convenient and reliable method for determining the
legibility of package label copy. Some methods have been
used previously and described in the literature. Seven such

methods are described in the following pages.

2.1. maimfilhgd

This method. has found considerable use in studying
legibility, particularly visibility or perceptibility at a
distance (23). The apparatus consists of a wooden rail
about three meters long placed before the subject, slanted
downward at about a 15 degree angle. There is a headrest at
the upper end of the rail. The stimulus material (letters,
words etc.) is placed in a small, well illuminated car which
can be moved any desired distance along the rail. The
stimulus material is placed near the far end of the rail and
the car is moved towards the subject by steps of about 10
centimeters at a time, thereby allowing the subject to read
as .much of the material as possible at each step. .A
modification. of this. method was developed to study the

legibility of large print for advertisements or highway

7

8

signs. Here the "stimulus material is placed in a fixed
position and the subject walks slowly toward it until he
just reads it correctly" (23), and the distance is recorded.
This method measures the distance from the eyes at
which print can be perceived accurately. It has been found
useful in studies of relative legibility of letters of the
alphabet and digits, of specific letters in different type
faces, and the effects of brightness-contrast between print
and paper surface. However, Tinker states that this method
"has little or no validity for investigating such
typographical factors as leading, line width,
determination of optimal type size for reading
continuous text, or optimal type faces." (23)
However, it has been used to measure the legibility of print
for road signs, bill boards and poster advertising, which
involves the use of large print. Apart from this, the

application of results obtained by this method to the

legibility of print has not been ascertained.
2.2. The Iachistggggpg (I-SQszpe.L__o_1;_1he..§_b_QLL-;Expg§iine
flatbed

M. A. Tinker describes the short-exposure method as the
method for measuring "legibility by determining the speed of
accurately perceiving printed symbols." (23). The apparatus
is called a tachistoscope. This is a mechanical device used
to evaluate surface designs of a package. Leslie Barton (3)
defines the tachistoscope as an instrument which measures
legibility, recognition, attention and memory value which
may be correlated with shelf visibility. Barton says that

this is a useful testing device where individual word

9
legibility is in question. Donald Morich (13) describes the

tachistoscope as one used to measure the impact of salient
visual elements of a package and recommends it for most
marketing-packaging evaluations, particularly as an impact
measure. This device is comprised of a shutter that, for a
desired fraction of a second, will allow a subject to view a
test sample. The device precisely controls the amount of
time a stimulus (i.e. a package) is exposed to the subject
or respondent, thus ensuring equal exposure time.

Barton identifies three types of tachistoscopes used in
Packaging--slide projection, mechanical and light flash (3).
Slide projection is the most common type of tachistoscope in
use today. It is a still projector with a shutter similar
to that used in a camera. The slide projector flashes the
image on a screen and the duration is controlled by changing
the setting on the shutter. The triggering device is the
most important physical consideration. The tachistoscope
can be mounted to a normal slide projector. Advantages
include group viewing, flexibility as to physical sample
size, and the increased attention of the subject when
showing a lighted area in a dark room. One disadvantage,
though, is not seeing an actual package. The actual testing
should expose each subject to a preset number of exposures
starting at a speed of 1/150 seconds and graduate down to
speeds around 2 seconds. After each exposure the respondent
reports what he saw according to a pre-established question

guideline. People do become more proficient as the test

10

continues, so the order of samples should. be uniformly
varied and a dummy sample used initially for warm-up. Lee
Swope (17) emphasizes the use of a test group with varying
eyesight as this would tend to simulate the real market
place. He says the test reveals the impact measure on
brand-name recognition, product description, play back,
recognition. of symbols/logos and. other salient. graphics.
This can then be used to compare design alternatives such as
shape, color and type. Swope goes on to say that this
technique is ideal in comparing test package alternatives as
it produces consistent results in a test-retest situation
and is able to discriminate among test alternatives.
Similarly, if the test packages are capable of producing
different levels of impact, this testing procedure will
measure the difference.

Another form of tachistoscope in use is similar to a
revolving blackboard on which a multitude of objects are
placed and the viewer’s response recorded. The viewing
period is usually in seconds rather than fractions of
seconds. In this case, there is a variation in lighting,
and objects (packages) are viewed, rather than images as in
the case of slide projection method of tachistoscope. The
third type of tachistoscope described by Barton uses a flash
of light on an object (firing a flash of light of accurate
duration by electrical impulses on a magnetic tape). Flash
can also be originated by construction of a tachistoscope

using a high speed photolight which can achieve exposures of

 

11

one millionth of a second. The major advantage is the
accuracy achieved at short durations; objects (packages)
are viewed as opposed to slide projector images (3).

The tachistoscope is useful as an impact measure,
according to Barton, but does not reveal the legibility of
the graphics on a package particularly as they pertain to
textual information on the packages (3). This instrument is
important for such features as brand name recognition, logos
and symbols, which play a major role in marketing of a
product. In areas where detailed information is important
to a consumer or a professional, such as in medical or
pharmaceutical labels/packages, the tachistoscope cannot be
used. This is because the exposure time is very limited,
time here is a controlled factor, and the level of
illumination is changing greatly in a short time. Alf
Nelson, the inventor of the polariscope for package graphics
evaluation states that

"legibility cannot be accurately measured during

changes of illumination, because the time lag

necessitatedby the pupillary adjustments of the

eye under varying levels of illumination creates

false reading results" (18).

M. A. Tinker reported that

"the relationship between tachistoscopic reading

and ordinary reading of continuous textual

material is so small that conclusions concerning

legibility of continuous printed text from
tachistoscopic results must be made with caution"

(23).

In summary, this method is only useful for measuring the

impact and recognition of letters and digits, specific

12
letters in different type faces and the effect of variation

in brightness contrast between print and paper.

2.3- W

This method consists of a system of lenses which are
related to each other such that a visual stimulus (print)
may be projected upon a ground glass screen in any degree of
clearness from an unrecognizable blur to clear definition of
focus and thus the degree of clearness can be measured (23).
To facilitate motion, the lenses are interconnected in such
a way that they travel in opposite directions. As one lens
increases the size of the image, the other reduces its size.
This results in a blurred image on the ground glass having
the same size as the clear image would be.

No image is present on the ground glass at the start of
the test, but as the subject turns the wheel of the
apparatus, the image appears and comes into focus. The
respondent or subject reports as soon as he perceives a
meaningful visual form. Scale value (”I the apparatus is
also recorded, which indicates the degree of clearness of
the visual task at every step. A zero scale value indicates
maximum clearness.

This method may be used to determine legibility by
measuring how far a letter can be thrown out of focus and
still be recognized. The technique is however limited to
the investigation of the relative legibility of letters of
the alphabet, of digits and of specific letters in different

type faces. The apparatus has been reported by Tinker (23)

13

to yield precise measurements of relative legibility of the

above which should have high validity.

2.4- WWLM

Luckiesh and Moss promoted blink rate as a measure of
readability (legibility) (23). ‘The studies were concerned
with ease of seeing and the effects of varying typographical
factors. The rate of involuntary blinking during reading is
assumed to be inversely proportional to ease of seeing.
This means that the easier the visual task, the fewer the
blinks for a set period of reading. Therefore any visual
task (i.e. typography) which is read with greater ease
should produce fewer blinks. Similarly, the rate of
involuntary blinking decreases as the level of illumination
increases, while it is greater for the lower levels of
illumination.

In studying typographical factors, the eye blinks are
counted by direct observation for a period of five minutes
while the subject is reading continuous text printed in each
set up (i.e. 6 pt or 12 pt type etc.), while the
illumination is held constant for all the tests (23). The
blinks are counted and recorded by the experimenter seated
to the side and slightly behind the subject. Tinker reports
that, to study the effects of illumination level on reading,
the blinks are recorded. during the first and. last five
minutes of approximately an hour of continuous reading under

each light intensity level (23).

 

14
Tinker evaluated the reliability and validity of

legibility measurement by this technique during reading by
using 74 and 64 subjects in two experiments. For adjacent
five minute periods, the reliability coefficients were
fairly high, .74 for the 74 subjects and .92 for the 64
subjects. For periods of 10 to 15 minutes a similar trend
was observed. When an interval of 20 minutes of reading was
substituted, the reliability dropped to .49 to .56 which was
marginal and indicated inconsistency of the blink rates
(23). It is suggested that to minimize the effects of
individual differences, sizable groups of subjects should be
used in order to have a higher reliability especially where
group comparisons are concerned.

The validity of this method in measuring legibility has
not been clearly ascertained even though the results
comparing the reading of 12 pt Book type and 7 pt the
Newsprint by Tinker appear convincing. The data indicated
that the Newsprint was easier to read than the Book type
where the scores were recorded from 60 readers whose
frequency of blinking was recorded at 5 minutes intervals
for a total of 10 minutes for both typefaces. It is further
reported that none of the results from Luckiesh’s laboratory
were evaluated statistically (23). This brings into
question the use of the rate of blinking as a criterion of
ease of seeing or legibility since it is doubtful that the
frequency of reflex blinking can be accepted as a valid

measure because this can also be affected by other

 

15
psychological and physiological factors which might affect

the results.

2.5. The Exe_MsxamenLlTaat

The eye movement test indicates where the eye is
travelling on a package, and the sequence in which the
different graphic elements are being observed. This test is
important as an analysis of eye movement patterns which can
determine how well the overall graphics layout and
typography attract or hold the eye, while it yields
information on why one typographical arrangement is more or
less legible than another. Tinker reports that "eye-
movement records tend to be reliable and valid measures of
legibility of print" (23).

There are different techniques which are used to record
eye movements in reading. One form is that of a single
movie camera fitted with a special close-up lens which takes
pictures of the eye and transmits them to a computer
controlled unit which bounces a beam of infrared light off
the retina, thus tracking the eye movement physically as the
subject views the test package (17). Each movement of the
eyes produces a change in the beam of reflected light so
that horizontal and vertical movements are recorded on the
film. Tinker mentions that the eye movements are readily
distinguished from the head movements on the film record
(23). This is also referred to as the corneal reflection
method. The test shows what the subjects observed in a

given amount of time and in a certain sequence. The less

16

important graphic elements that may interfere with those
elements which are essential to package communication are
highlighted by this test, and this may result in corrective
measures for redesigning the package . The major
disadvantage of this test is that it does not show the
quality of the graphic elements observed in detail. This is
where a tachistoscopic test is helpful to measure
noticeability of individual graphic elements.

The other form of eye movement test commonly used is
the electrical method, which depends upon an electric
potential difference which exists between the cornea and the
retina of living eyes (23). The movements are recorded by
placing electrodes on the skin near the eye and attaching
the electrodes to an amplifying system and an ink-writing
oscillograph or a photographic device. This method has
gained more credibility because it is much more flexible
than the corneal reflection method especially because the
subjects (or reader’s) head is not held rigidly in one
position and records for longer periods of reading can be
obtained.

Results obtained by Tinker for short paragraphs. found
reliabilities from .66 to .89; for reading 20 to 25 lines;
the mean reliability was .81 (23). He therefore reported
that those measures had satisfactory reliability for
studying legibility of print and all eye-movement measures
provided valuable supplementary information in investigating

legibility. However, it is Tinker’s opinion that these

17

measures and results can only' be correlated. to reading
performance since they do not show the degree of legibility
of individual graphic elements but largely deals with a

subject’s perception and reading ability.

2.6. Emmi—EJ92211-

Leslie Barton describes the polariscope as an
instrument which can be used to measure the various degrees
of visibility or legibility of the different components of a
surface design (3). The one used by Barton for his
legibility study was an adaptation of the polariscope based
on the original Alf Nelson (18) invention which was referred
to in the literature as "Polarascope" visual tester. The
polariscope used for the study reported here was designed by
General Radio Company (5).

Barton’s description of the operation of the
polariscope is summarized as follows: Polariscope is based
on the properties of polarized light where two light
polarizing screens are used. The two screens are the
polarizer and the analyzer. Light is polarized by the first
screen, the polarizer, and is passed through the analyzer.
When the screen polarizing planes .are parallel to each
other, the maximum amount of light is allowed. to pass
through and when the planes are perpendicular the opposite
is true as all light passing through the polarizer would be
absorbed by the analyzer. Both of these results depend upon
the relative orientation of the screens. Each plane

polarizer carries a degree scale calibrated from zero to

18

plus and minus ninety degrees for determining the degrees of
rotation (3). The normal light source furnished with the
polariscope is an incandescent lamp contained in a metal
housing with a diffuser fitted to it. This light source is
used when light transmitted through a material is used to
evaluate stress patterns.

The polariscope has been widely used for dynamic and
static stress analysis in the field of mechanical research.
Of importance in this area is photo-elastic stress analysis,
an experimental method of determining where and to what
degree stress concentrations will exist in a plane section
under known loading conditions. The method has been applied
successfully to design problems ranging from bolt heads to
battleships (5). More important to a packager is the use of
' the polariscope in evaluating seal integrity in pouches and
strains in glass by observing color differences of
transmitted lights which are related to stress
concentrations. A further use in packaging has been the
adaptation of the polariscope to evaluation of the
visibility, or legibility of label elements, as described by
Barton (3).

Lockhart and Michel (8) describe the adaptation of the
General Radio equipment in summary as follows: The label
element is placed between the diffuser and the polarizer
screen on a platform where it is slanted to an angle for
easy viewing. A source of illumination is directed at the

label element at this angle from a lighting source. Once

19
the light falls on the label, it is then reflected through

the polarizing screen and on to the analyzer screen and this
light is viewed by the subject rotating the analyzer screen.

In comparing the tachistoscope with the polariscope,
Barton stated that the tachistoscope tests revealed the
attention-getting characteristics of a package as observed
by a shopper when he gets a glimpse of the package, while
the polariscopic tests revealed the order in which the
subject would perceive the surface design after his
attention was brought to it and he stopped and studied it
for a few seconds (3). Barton reports that specific
information is not seen as a whole through the polariscope
as is the case with tachistoscope; it is usually seen in
parts.

Alf Nelson, the inventor of the polariscope application
claimed that eyesight differences would not affect the
results because all surface designs were rated comparatively
by each individual (18). However, tests conducted by
Lockhart and Michel, of the School of Packaging, Michigan
State University (8) showed that people with 20/30 vision
needed considerably more light to see a specified message
than those with 20/20.

Using the polariscope, Lockhart and Michel tested 48
subjects, 81% of whom had 20/20 vision and the rest with
20/30 (8). They were tested on two package forms, a bottle
with a printed label and an ampoule with message silk

screened on the surface. For both packages, as the type

— 1
P...

20

size got smaller, the degrees of rotation increased--showing
that more light was required to read the message. In a
second test by Lockhart and Michel, there were eight
variations of black on white printed labels comprised of two
messages, two type styles (Garamond and Helios) and three
type sizes of 5, 6, and 10 points. After developing a split
plot experiment utilizing analysis of variance, the results
showed the effect of the difference of the two label
messages with the two different type styles--with the
Garamond being harder to read than Helios. Similarly, the
subjects with 20/30 vision had greater difficulty reading
the labels than did the subjects with 20/20 vision.

The outcome of this experiment showed that the
polariscope can measure differences in legibility between
type sizes and type styles. The authors say that this
method could also be adapted to pass-fail testing as well as
comparative testing on label design as follows: since the
experiment involved medical labels, it was the two authors’
belief that

"the legibility test should take into account that

the medical practitioner is expecting a particular

message on the label; he is comparing the message

on the label with the message in his mind--not
seeking to decode a totally unfamiliar message."

(8)

One of the most important advantages of the polariscope
is its ability to determine which design elements will be
most legible to the observer and to aid in finding design
criteria for choosing the package surface design. The only

major disadvantage pointed out by Barton is that, "the

21

polariscopic tests indicate nothing as to how the design
will perform among others as in a store display" (3). This
is true because polariscopic tests involve single packages

which are normally viewed in parts and in isolation.

2 . 7 - meawwwwmt

The Luckiesh-Moss visibility' meter 'was developed. by
Luckiesh and Moss as a means of measuring visibility under
controlled conditions of illumination, thus obtaining a
foundation of knowledge upon which to specify light and
lighting for specific visual tasks (10). This meter is a
binocular device that has two colorless filters in parallel,
one for each eye, which are optical circular gradients.
Each filter varies from almost clear to very dark, and when
rotated simultaneously in front of the eyes, can alter the
brightness--contrast of the object whose visibility is to be
measured. The amount of rotation is marked by a numerical
circular scale associated with the filter. The instrument
is held in approximately the same position that eye glasses
are worn, and with the use of a finger, a knob is slowly
turned until the visual object or its fine detail is legible
to the observer, at which point the reading is recorded
(10).

The meter scale is called relative visibility (RV) and
it has a range of 1 unit to 20 units. A second scale on
this instrument is that of’ relative foot. candles which

ranges from 1 to 1000. By definition, the scale term

22

"relative" visibility can be "absolute" visibility, while
the scale of relative foot candles is related to the amount
of light required to achieve a given visibility (12). The
foot-candle scale is not of interest for this investigation,
but the relative visibility scale is.

In designing the scale and calibrating the visibility
meter, some assumptions were made by Luckiesh and Moss.
They pointed out that,

"critical visual tasks are performed. either' at

distances within arm’s reach, as in reading, or at

much greater distances, as in driving an

automobile. The ophthalmologist usually assumes

that distances of about 14 inches and 20 feet,
respectively, are representative of these two

cardinal distances." (12)

Therefore, the visibility scale can be calibrated for either
near-vision or distant vision, with respect to the detection
of the presence of an object and its critical resolution.
This scale is calibrated in terms of the threshold size of
Luckiesh and Moss standard parallel-bar test. This test
uses a standardized object which consists of two parallel
bars of certain width and spaced a distance apart equal to
the width. The dimension of the bars and space are related
to the viewing distance in such a way that the space/bar
dimensions and the viewing distance establish a known visual
angle. The threshold size is the size that is just visible
under 10 foot candles illumination by a person with normal
vision (12). For the scale design, and calibration object

dimensions and viewing distances, combinations were chosen

to given visual angles ranging from 1 minute subtended angle

23

to 20 minutes subtended angle. Subjects of normal vision
then viewed the objects through the meter and rotated the
filters until the objects were just visible. The scale was
them marked in terms of the visual angles 1 minute to 20
minutes.

Tinker points out that since the meter is calibrated in
terms of threshold size, a visibility scale value of "1"
indicates the standard test-object (parallel bars) whose
critical detail subtends a visual angle of one minute.
Similarly, a scale value of "2," two minutes, etc. (23).
Thus, the scale values of relative visibility are directly
proportional to the visual size of the standard test-objects
under threshold conditions.

When the instrument is calibrated and ready for use
under standardized conditions, it is possible to measure the
visibility of any object. A standard condition for viewing
the objects is provided by the simple method of having the
field surrounding the object either black or of constant
brightness.

Among the studies conducted using this meter is that of
relative visibility of type faces, type forms, size of type,
and effects of variation in ‘brightness contrast between
print and paper. Further studies by Luckiesh and Moss
explore the relationship of illumination intensity to type
size. It was reported that,

"about 3 1/2 times as much illumination was

required to make 6 point Bodoni type as visible as
12 point. Deficiencies in type sizes between 6

24

and 12 point can be compensated for by increases
in illumination." (10) -

The same study found out that the visibility of 3-point-type
was close to "1," which indicated that this was about the
smallest type readable by persons with average normal vision
under an illumination of 10 foot candles. One drawback of
these measurements as argued by Tinker (23) is that one does
not know what scale value corresponds to optimal legibility
when investigating size of type. The author continues to
say that for studying the effects of variation in type face,
in leading or in line width, visibility measures are of
little value. Nevertheless, situations ‘where 'visibility
scores are useful include studies of the effects of
brightness contrast between print and paper, and the
relative legibility of letters of the alphabet, digits and
other isolated symbols.

The Design and Market Research visibility meter (DMR)
which was used for this study was designed by Francis P.
Tobolski (24). The meter ‘was designed using the same
principles used in the original Luckiesh-Moss visibility
meter. The difference from the original Luckiesh-Moss meter
is that the DMR meter was redesigned for the use of
measuring legibility/visibility and therefore the scale for

relative foot candles has been eliminated amd in its place

 

is the scale of relative visibility. The scale begins at 1
and ends at 20 just like the original meter. Another
modification of this meter is the provision of a handle,

which a subject or viewer can use while putting the meter in

25
front of the eyes and with the use of the other hand (or

fingers), can rotate the disks comfortably. This
modification clearly improves the usability of this device
because of the efficient handling device.

One of the precautions to 'be taken while using a
visibility meter, according to Luckiesh (10), concerns
extraneous light, especially from sources and areas in the
visual field surrounding the object or field. If allowed to
enter the windows of this meter, it can add to the veil of
haze, thereby affecting the visibility measurement. This
can be prevented by either screening the visibility meter
from extraneous light or by maintaining a constant level of
illumination. For the purpose of this study, the latter was

chosen with illumination maintained at an intensity of 50

foot candles.

CHAPTER 3

MATERIALMNWABAHS

3.1. Materials.

Testing for the legibility of messages on child
resistant caps was performed on two types of cap system of
similar (plastic) material, but with different message
content. The message, which consisted of opening and
closing instructions, was emboss printed on the top surface
of the caps. The emboss printed message was white on a
white background and all the caps were white in color. For
the purpose of this study, the lettering of half of the caps
used for this study were colored black in order to compare
the effect of contrast differences on legibility. For the
purposes of simplicity in the presentation and discussion of
results, the following message categories and designations
will be used.

Message 1 - In this category were 2 paired (four) caps
of two sizes (33 and 22mm continuous thread caps) with the
following instructions: "Push Down & Turn, To Open,"
including two directional arrows; and "Close Tightly." (See
illustration No. 1) This message appeared in bold upper

case sans serif type style on both pairs of the test caps.

26

‘27

ILLUSTRATION NO. 1
MESSAGE 1 CRP CAPS

$§0M¢e
3 CLOSE a

, \naum I

’0 ovk‘

 

THE CAPS ON THE LEFT ARE IN 12PT TYPE SIZE WHILE THOSE ON
THE RIGHT ARE IN 8PT.

28

The message type size was 12 pts on the large pair and 8 pts
on. the smaller' pair. Although. all the test. caps ‘were
originally supplied emboss printed white-on-white, the
lettering on one cap of each pair was colored black in order
to create a color contrast for testing purposes. These caps
were supplied by Kerr Glass Manufacturing Corporation,
Lancaster, Penn.

Messangl - Same pairs and cap sizes as above, but
with the following message instructions: "Close Tightly,"
and. "While Pushing Down Turn" including two directional
arrows. (See illustration No. 2) The message on the larger
pair was emboss printed in bold upper case sans serif type
style in 8 pts while the smaller pair appeared in 6 pts.
These caps ‘were supplied by’ Owens Illinois, Brookville,

Penn.

3.2. Major Apparatus

Two types of mechanical devices were used to determine
the legibility of messages on the child resistant caps. The
main purpose of using the two devices was to compare the
legibility measurements obtained with these devices and to
correlate the results of each with the expert panel’s
evaluation as the control, with the ultimate goal of
recommending the most suitable method. of’ measuring
legibility. They are the following:

(i) Polarisggpg_- Polariscope Type 1534-A, manufactured
by General Radio Company, Cambridge 39, Massachusetts, was

used (See illustration No. 3) The major components were the

29

ILLUSTRATION NO. 2
MESSAGE 2 CRP CAPS

 

THE CAPS ON THE LEFT ARE IN 8PT TYPE SIZE WHILE THOSE ON
THE RIGHT ARE IN 6PT.

30

ILLUSTRATION NO. 3
POLARISCOPE

 

 

THE POLARIZING SCREEN AND THE ANALYZER WITH THEIR SCALES ARE
IN THE FOREGROUND. THE CRP CAP IS LOCATED ON THE WHITE BLOCK
BETWEEN THE POLARIZING SCREEN AND THE DIFFUSER
(IN THE BACK GROUND).

31
Polarizer and Analyzer Type 1534-PI, both with a scale of

degrees of rotation from 0 to 90 degrees. The following

table shows the Polariscope screen rotation (9).

Table 1

Polariscope Screen Rotation

 

Angular Setting Degrees Light Conditions
Polarizer Analyzer
Screen 1 Screen 2
90' 0' Dark
90' 10' Lighter
90' 45' Still Lighter
90' 90' Maximum Light

 

Table 1 shows the light transmission when the analyzer
(screen 2) is rotated from 0 to 90'. Assuming the polarizer
is fixed at 90' when the analyzer reaches 90', its axis of
polarization is parallel to that of the polarizer and the
maximum amount of light is transmitted.

(ii) Illumination. A controlled source of illumination
for the polariscope was provided by a General Electric (GE)
high intensity bulb of 40 watts (120V), with white

reflector.

 

(iii)
meter - The binocular device used for this study was

supplied by Design a Market Research (DMR) Laboratory

32

Corporation, Carol Stream, Illinois, (See illustration No.
4). This instrument is based on the use of a neutral
density' wedge and. an opaque to transparent. wedge. 'The
visibility scale runs from 1 to 20' (visual angle in
minutes). The source of illumination was the same as that
used with the polariscope. The following table shows the
visibility' meter’s opaque to transparent wedge rotation,

when rotated from 20 to 1.

Table 2

DMR Visibility Meter Wedge Rotation

 

 

Visual Angle Light Conditions
20 Semi-opaque
15 Lighter
10 Still Lighter
1 Maximum Light
(Transparent)

 

3.3. Other Requirements

(i) W1: - This was used for the purpose of
measuring the amount of illumination used in this test. The
amount used for testing purposes was 50 foot candles. This
level of illumination was arrived at as a result of a
general survey conducted in the offices of the School of

Packaging and one department Store in East Lansing, where

33

ILLUSTRATION NO. 4
DMR VISIBILITY METER AND A CLOSE-UP SHOWING THE SCALE AND
THUMB WHEEL ON THE RIGHT

 

34

illumination was recorded at an average of 50 foot candles.
The light meter used had a range of 0—250 foot candles and
was manufactured by General Electric Corporation, Type DO-
78, Model J-55. No. F-2, (See Illustration No. 5).

(ii) Near Point Visual Acuity Card - Visual acuity is
commonly defined as the ability to distinguish fine details
(11). Visual acuity may be high or low for a person, which
can be attributed to different degrees of eye-defectiveness,
different levels of brightness or illumination, and
different brightness-contrasts between object and
background. However, it is entirely, or at least
overwhelmingly associated with eyesight. Visual acuity in
this test is important because eyesight or vision influences
the ability to perceive, recognize or read a given visual
message such as print under a controlled level of lighting
or illumination.

To measure visual acuity, Ii near point visual acuity
card from Dow Corning Ophthalmics Inc., Norfolk, Virginia,
was used, (See illustration No. 6). The instruction was to
hold the card in good light 16 inches from the eyes. The
Dow Card follows the Snellen System (11) of rating vision,
where:

d actual distance
Visual acuity = --- = ------------------
D normal distance

"d" is the distance in feet at which a given line of

letters is barely recognizable by any subject, while "D" is

the distance in feet at which the same line of letters is

35

ILLUSTRATION NO. 5
LIGHT METER

 

36

ILLUSTRATION NO. 6
NEAR POINT VISUAL ACUITY CARD (DOW CARD)

.5
'"vSDKHN OCVRZ

HCZSV DORNK

‘ , ONHDR szcv

Dow Comang _
Opnmclmxcs

un-

 

 

 

37

barely recognizable by a person with normal vision. "d" is
always the distance at which the test-chart is actually
viewed. The Dow Card has Distance Equivalent ranging from
20/200 (in 26 pts of assorted letters of the alphabet in
upper case) to 20/20 (in 3 pts). The Dow Card is devised to
be used at a distance of 16 inches. Luckiesh (11) emphasizes
that one could make a chart to be used at any specified
distance but Luckiesh and Moss pointed out that,
"critical visual tasks are performed. either at
distances within arm’s reach, as in reading, or at
much greater distances, as in driving an
automobile. The ophthalmologist usually assumes
that distance of about 14 inches and 20 feet,
respectively, are representative of these two
cardinal distances" (12).
As an example, if at a distance of 16 inches a person can
barely recognize the 3 pts letters in the test chart, he is
said to have average normal vision which is expressed as
20/20 vision. If one can barely distinguish the 26 pts
letters (first line on the chart), his vision is 20/200.
This means he is just able to recognize at 20 feet what a
person with normal vision can recognize at 200 feet.

Therefore threshold size for him is 10 times normal and his

visual acuity is rated as 10% of average normal (11).

CHAPTER 4

TESWDLIBE.

4 . 1 . .lntroduciign

Tests were conducted to determine the legibility of
white-on-white and black-on-white typographic messages on
child—resistant caps. Two devices were used in conducting
this test--the polariscope and the DMR visibility meter.
Two other objectives were to determine if these two devices
gave the same information on legibility and whether or not
they gave the same result as would expert evaluators. In
these tests, two classes of consumers (based on age
subgroups) evaluated eight CR closures with the polariscope
and DMR visibility meter. The first age subgroup was made
up of ten students from the School of Packaging, ages 18 to
25 years. The second age subgroup consisted of ten staff
members from the School of Packaging ages 30 to 54 years. A
third group was made up of an expert panel consisting of ten
professionals in graphics trades related to legibility, and
this group was used as a standard or reference control.
Visual acuity of each group was measured with the Dow Card.
In this test, the illumination was held constant at 50 foot

candles.

38

39

4-2. W

Ten subjects of the first age sub-group (18-25 years)
were used in this test. Before the beginning of the tests,
every subject read a near point visual acuity card using
both eyes and wearing such vision correction as was normal
for them. This visual acuity was then recorded on the test
form (see sample Test Form No. 1, page 51). 70% had a 20/20
visual acuity while the rest (30%) had a 20/30 acuity. The
other ten subjects tested on this device were from the 30-54
years age group, 70% of whom had 20/20 vision and 30% 20/30
vision.

The polariscope used for this test was placed on a
table and covered by a hood (see illustration No. 7) This
hood was a corrugated box with a circular opening on the top
to allow for a controlled light source, and an opening at
one end so that the viewer could rotate the analyzer screen
of the polariscope to view the object. The inside of the
corrugated box was lined with chrome-coated paper board to
help prevent absorption of light by the corrugated board,
thereby maximizing the illumination intensity. The lighting
or illumination was held constant throughout the test at 50
foot candles. Only the test source of lighting was used for
this test, and therefore the rest of the lights in the
laboratory were turned off during the duration of this test.
The test source of lighting was a General Electric 40 watt

bulb which was mounted in a lamp with white reflector.

4O

ILLUSTRATION NO. 7
POLARISCOPE COVERED BY A HOOD (CORRUGATED BOX)

 

 

41

At the beginning of the test, the polariscope screens
were closed so that the closures could not be seen. This
means that the analyzer (screen 2) was at zero degrees while
the polarizer (screen 1) was at ninety degrees. In the test
room, the subject was seated in front of the polariscope
which focused his/her attention on the instrument screens.
The distance from the eyes to the target (or the test cap)
was maintained at 16 inches (See illustration No. 8).
Before the actual tests were conducted, a trial test was
performed in order to familiarize the subject with what was
expected. of him/her. A. sample cap was placed in the
illuminated space. The subject was instructed to rotate the
first screen facing him/her (the analyzer) slowly and
consistently' in. one direction (either clockwise or
anticlockwise) with his hands until a message on the cap was
legible. The subject was then instructed to stop the
rotation and report what he/she could read. Subjects were
not instructed to read in. any special order, but were
encouraged to indicate when they could read the whole
message. The cap message was not revealed to the subject.

Once the subject read the message, the numerical
reading on the scale attached to the outer diameter of the
analyzer (screen 2) was recorded by the test monitor. This
scale started from zero (with no visibility) to 90 degrees,
the maximum allowable light transmission. After
familiarization with the evaluation procedure, the sample

cap was replaced by the test cap and the above procedure was

42

ILLUSTRATION NO. 8
A DIAGRAM SHOWING THE POSITION OF THE OBSERVER IN RELATION
TO THE P-SCOPE AND THE TEST CAP.

 

 

Analyzer
Angle 4 Polarizer Cap Position
70" ‘
Observer TF5 I p q ._ .1
' I, h .4 h \ I s a
r x
I \\ \

 

 

 

 

 

 

 

 

 

 

 

 

 

 

1.0inch 11:11“: _. Er

P1054

inches

 

3. 5 inches

 

16 inches ——9

43

repeated. The reading was a direct mechanical reading as no
electrical or other' devices were used toi determine the
reading. Each result was recorded on a test form (see
sample Test Form No. 1, page 51).

The eight caps (see illustration No. 9) consisting of
four caps with message I and the other four with message 11
were presented to the subject in a fixed order (see Table 3)
and each cap was viewed once. One set of the caps with
message I was mixed with the other set of message II and
numbers 1 to 8 were assigned to them. The caps were then
viewed one at a time in the same sequence by all the test

subjects.

Table 3

Cap Order of Presentation to the Consumers on Both
Polariscope and DMR Visibility Meter

 

 

Cap No. Color Size Message Type
1 Black-on-white 12pts I
2 White-on-white 6pts II
3 White-on-white 12pts I
4 Black-on-white 8pts II
5 White-on-white 8pts I
6 Black-on-white 6pts II
7 Black-on-white 8pts I

8 White-on-white 8pts II

 

44

ILLUSTRATION NO. 9
CAP PRESENTATION ORDER
1 TO 8 — PRESENTED TO CONSUMERS
1 TO 16" PRESENTED TO THE EXPERTS

‘i
ii
,ri
E
Ii
.4
It“
3.

 

45
4.3. pug visibility Meter_2recedure

The same subjects that were tested with the polariscope
also took part in the visibility meter test. The test was
individually administered under the same illumination of 50
foot candles as that of the polariscope. The visibility
meter (see illustration No. 4 and 10) was held by a subject
in front of the eyes, in approximately the same position
that eye glasses are worn. On the table, 16 inches away
from the subject’s eyes a cap was placed on an L-shaped
rigid polystyrene foam material such that it had a 45°
inclination to help the subject read the message with ease.

At the beginning of the test, the density wedges were
at the position of maximum opacity. In this case, the
visual angle scale (in minutes) was at 20. Initially a
trial test was conducted to familiarize the subject with the
use of the meter and to build up confidence in its usage.
The same sample cap was used with all the subjects. Once
the sample cap was placed. on the designated. spot, the
subject was instructed to rotate the thumb wheel on the
visibility meter until he/she could see the message. There
was no particular order of reading the message, but the
subjects were encouraged to stop rotating the knob once they
perceived the whole message.

As in the polariscopic procedure, the message on the
cap was not revealed to the subject. Once the message was
legible to the subject, the numerical reading (between 1 and

20) on the meter was recorded. After the familiarization

46

ILLUSTRATION NO. 10
A DIAGRAM SHOWING THE POSITION OF THE OBSERVER USING DMR
VISIBILITY METER IN RELATION TO THE TEST CAP.

DMR Visibility Meter

l

'Cap Position

 

 

Angle Observer
65° <

I

 

 

\

 

 

 

 

.1p-—-—-r b‘

l—
l__

1.3ft.

47
with the sample cap, it was replaced by the test cap and the

same order of presentation (of cap number one to number
eight) was followed (see Table 3). The visual acuity of the
subjects was only recorded once, during the polariscopic
test. Both tests were conducted consecutively with a break
of about ten minutes to reduce the fatigue that might have
been associated with the first test and to further diminish
the memory of what the subject had seen in the previous
test. The results were recorded on a test form (see Sample

Test Form 1, page 51).

4 . 4 . WHWQQQQLQ

An expert panel consisting of ten professionals was
created as a standard whose evaluation could be compared to
that of the consumers using the two devices. These experts
were drawn from the professional field of graphics trades.
They had a wide range of experience in this field, ranging
from thirteen to thirty years, and their ages ranged from
thirty two to sixty years. All of them had attended
college.

The expert panel was asked to observe the caps and to
give a numerical rating for legibility. Unlike the
consumers, they did not use the two devices, but rather each
cap was rated for legibility on a scale of 1 to 10. Ten was
the highest level of legibility while one was the least. It
was expected that this method of rating the caps would give
some objective results which could be used in correlating

the consumers data from the two devices with that of the

48

expert judges as a reference standard. Initially before the
test every expert was requested to fill out a questionnaire
which included education/academic level, years of experience
in a relevant field and title (see Sample Questionnaire No.
1, page 53).

The test was conducted in an ordinary office in the
School of Packaging where the curtains were open and the
level of illumination maintained at 50 foot candles. This
setting was important because it represented a typical
reading environment such as a home or office. Some of the
office lighting was deliberately turned off to maintain the
desired level of illumination.

At the beginning of the test each expert was
familiarized with the subject matter by taking him/her for a
visit to the test room consumers would use and informing
him/her why the School of Packaging was conducting this
study. Examples of child resistant bottle caps were shown
in the laboratory with their varied messages and the
objectives of the study on legibility were clearly defined.
Each expert was also shown the two devices that the
consumers would use in evaluating the legibility and how
they would use them. In doing this, the importance of their
contribution to this study was emphasized and the
application of the results was explained.

In the testing room each expert was informed that
sixteen caps would be rated for legibility, on a scale of 1

to 10, one at a time. Each cap was evaluated twice for

49
legibility but the experts did not know this. They thought

they were evaluating sixteen different caps. This is unlike
the consumers who evaluated each cap once while using each
of the two devices. The caps were presented to the experts
in a fixed order (see illustration No. 9 and Table 4). The
procedure was to have the expert seated at the table while
each test cap from one to sixteen was presented one by one
by the test monitor. There was no limit of time in which
the experts could view any cap. The only emphasis was to
place the cap 14-16 inches away from the eyes as this is
usually the normal reading distance. The test was
individually administered and the content of the message was
not revealed to the experts before the test. After the
expert rated each cap, he/she recorded the rating on a test
form (see Sample Test Form No. 2, page 52). Before the
beginning of the test, each expert read a Near Point Visual
Acuity' Card and the visual acuity was recorded. on the
aforementioned form. After evaluating all the caps, each
expert was shown all of them together and relevant comments

noted.

50
Table 4

Cap Order of Presentation to the Expert Panel

 

 

Cap No. Color Size Message Type
1 Black-on-white 12pts I
2 White-on-white 6pts 11
3 White-on-white 12pts I
4 Black-on-white 8pts 11
5 White-on-white 8pts I
6 Black-on-white 6pts 11
7 Black-on-white 8pts I
8 White-on-white 8pts II
9 Black-on-white 12pts I

10 White-on-white 6pts II
11 White-on-white 12pts I
12 Black-on-white 8pts II
13 White-on-white 8pts I
14 Black-on-white 6pts 11
15 Black-on-white 8pts I
16 White-on-white 8pts 11

 

51
SAMUWMNE

WW

Measuremsmuestha

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

W
Manners
No: Visual Acuity:
Age Group: Check One: Sex:
____ 18-23
____ 30-55
Bolerisseps newsmumumx
Readings: Degrees of Readings: Visual
Qap_NQg Rotation Ca . Angle in Minutes
1 1
2 2
3 3
4 4
5 5
6 6
7 7

 

 

 

No:

52
SAW

Michiganjtateflnixsrsm
MW

Meesmremewiezihiliuwum

DatLShseLlEl
£1.31).ng

Visual Acuity

 

 

9.32.119... WWW ' ' J.

1
2

10
11
12
13
14
15
16
17

18

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Generalflmments :

53
SAWW

Waxing
ElmsrLPanel
would like some information about you.
Name:
Work Address: City/Zip:

Phone: ( )
Sex:
Age:

Title:

Education/academic Level:

Years of experience in printing/graphicsltypography:

Have you ever taken part in a research involving an
expert judge/panel? If yes, what type?

Would you be willing to participate in another research
panel at some later date?

CHAPTER 5

DISQQSSIQN_QE_EE§QLI§

5.1. Introduetien

In analyzing the results, the following questions
relating to the objectives of this study will be discussed:

(i) Do the instruments give the same results as the
experts? In other words, what is the correlation between
the experts’ rating of legibility on the eight caps with
that of the consumers using the two instruments?

(ii) How do the two instruments compare?

(iii) What is the effect of color contrast, type size
and visual acuity on the legibility of the two messages for
(a) the experts and (b) the consumers using the two

instruments?

5.2. Corr eletien Between the Exper.s..fiatins.and.th§
Censumers.Rsadin1s_!ains.the.lwQ.lnatrumenta

The raw data for the experts’ ratings of the sixteen
cap presentation and the average rating of the eight cap
variations is recorded in Appendix A page 90-100, while that
of the consumers using the two instruments is recorded in

Appendix B page 106-108.

54

55
11W

Before the correlation between instruments and experts
was obtained, the data for the experts was analyzed for an
estimate of reliability (or the degree of internal
consistency) among the ten experts and their overall
reliability for the sixteen caps.

In evaluating the expert data for reliability the
method used for estimation was internal consistency by
Cronbach Coefficient Alpha (a) (27). SPSS-X RELEASE 2.2
Program was used for this analysis. The following formula
was used in determining the statistic rxx (which represents

reliability):

where n - number of items in test (e.g. 16 caps)

813 = variance of a single item (e.g. one cap)

M
II

Summation sign indicating that 8:3 is summed over
all items
3,4 = variance of the total test.

The reliability coefficients were reported for Standardized
Item Alpha, which controls for variance of the measurements
and is taken as the true estimate of reliability. A
reliability coefficient of "0" would indicate a lack of any
internal consistency among the experts, while "1.00" would

indicate complete agreement. 'The reliability coefficients

56
between pairs of experts (1 to 10, 2 to 10 etc. for all

pairs) and the overall reliability for the ten experts is
reported in Appendix A page 101.

The reliability coefficients between pairs ranged from
-1.64 to .90, and the overall reliability among the experts
was .83.

In their discussion of Reliability and Test Use,
Mehrens and Lehmann (27) claim that if a measure is to be
used to help make predictions about individuals, then it
should be more reliable (high alpha value) than if it is to
be used to make predictions about groups of people. It is
also the two authors’ opinion that standardized tests used
to assist in making decisions about "individuals" should
have reliability coefficients of at least .85, while for
"group decisions," a reliability coefficient of about .65
may suffice. Similarly, Miles Tinker (23) says that for
group comparisons, such as ordinarily employed in legibility
studies, a reliability should not be less than .50;
preferably, it should be .60 or above. The author continues
to say that if one is dealing with individual diagnosis, the
reliability coefficient should be at least .80, and
preferably .90 or above.

For the purpose of this study, the measures or rating
obtained from the experts will be used to help make
decisions about legibility by using the consumers with 20/20
and 20/30 visual acuities using the two instruments.

Therefore the overall reliability coefficient of .83

57
obtained from the ten experts is highly acceptable by the

above standard and indicates a very high agreement or
consistency among the experts. This internal consistency
estimate will then be of value for interpreting the
legibility scores of the CRP caps obtained from the
consumers using the two instruments, which means that the
experts can be relied upon as a reference control or
standard for this test.

For the internal consistency method of estimating
reliability, sources of error which may affect the estimates
obtained are: sampling error, random error within the test,
test length, speed, group homogeneity, item difficulty and
objectivity, according to Mehrens and Lehmann (27). They
further report that, the more heterogeneous the group, the
higher the reliability and the more subjectively a measure
is scored, the lower the reliability of the measure.

The expert panel was assembled from practitioners in
graphic arts in order to reduce as much as possible the
effect of subjectivity. Also, the panel was instructed
about the purpose of the evaluation and they were asked to
exercise their professional judgement as a further effort to
reduce the effect of subjectivity. In the reliability
analysis, a correlation matrix was also obtained. This data
reinforces the results of the reliability coefficients, but
was not used for this study. It is recorded in Appendix A,

page 102 (between pairs of experts and all pairs).

58
Egrrsletien_hetaeen.ths.smnerta_and.the.instrumsnta

Correlation coefficients (which measure how close a
relationship is to a straight-line fit) were obtained
between the experts’ ratings and consumers’ readings using
the polariscope and DMR visibility meter. The data used to
evaluate the correlation are reported in Tables 5-10 in
order rankings for the above mentioned categories. The caps
were ranked within each group from the most legible to least
legible and whenever there was a tie in two values, the
standard deviation value was used as the tie breaker, with
the smaller deviation assumed to represent greater
legibility. These rankings were used for comparisons
between the experts and the consumers using the two
instruments and for comparisons between the two instruments
in order to find out if there was a pattern in the ranking
order within these groups.

The method used to compute the correlation was based
upon the Product-Moment Correlation (19). MSTAT Version 3
Program was used for this analysis. The following formula
was used to determine the statistic r (Correlation
Coefficient):

NZXY - EXEY
{Tnzx= - (EXI’I [NZY' - (EYI'I
where N = number of caps (which were eight)
X and Y = ratio measures taken on N.
An example of the mathematical computation of

correlation is shown in Appendix B page 110.

559

lable 5
Legibility Rating by Expert Panel vs. Consuaer 20/20 Readings on
Polariscope in Order Rankings

 

 

 

 

Expert Panel Polariscope
Hessage Rank Cap Rating2 Color Size Rank cap Reading2 Color Size
OrderI No. (Pts) Order No. (Pts)
1. 5 4 R/R 8 1. 5 9 k/R O
2. 3 6 R/R 12 2. 3 9 R/k 12
1
3. 7 7 a/u 8 3. 7 6 O/R 8
4. 1 9 O/R 12 4. 1 5 all 12
1. 2 2 MIN 6 1. 2 14 R/R 6
2. O 4 MIR 8 2. O 10 III 8
II
3. 6 6 s/u 6 3. 6 8 O]! 6
4. 4 8 O/R a 4. 4 6 DIN 8

 

1. Ranking Order is free 1 = least legible to 4 = aost legible.
2. Free Appendix A, page 100, seen rating for expert panel.

3. Froa Appendix 8, page 107, seen reading for subjects using polariscope.

64)

iable 6
Legibility Rating by Expert Panel vs. Consuaer 20/30
Readings on Polariscope

 

 

 

 

Expert Panel Polariscope
Message Rank Cap Rating2 Color Size Rank Cap Reading2 Color Size
Order1 No. (Pts) Order No. (Pts)
1. S 4 R/R O 1. 5 17 All 8
2. 3 6 All 12 2. 3 13 all 12
1
3. 7 7 all 8 3. 7 11 ll! 8
4. l 9 AIR 12 4. 1 8 DIN 12
1. 2 2 All 6 1. 2 24 AIR 6
2. 8 4 III 8 2. 6 15 all 6
ll
3. 6 6 all 6 3. O 14 R/R O
4. 4 8 all 8 4. 4 11 All 8

 

l. Ranking Order is free 1 : least legible to 4 = eost legible.
2. Free Appendix A, page 100, seen rating for expert panel.

3. Free Appendix 3, page 107, seen reading for subjects using polariscope.

€51

7able 7
Legibility Rating by Expert Panel vs. Consuaer 20/20
Readings on ORR Visibility hater

 

 

 

 

Expert Panel ORR hater
Ressage Rank Cap Rating! Color Size Rank Cap Reading2 Color Size
Order1 No. (Pts) Order No. (Pts)
1. S 4 III 0 1. 5 11 III 8
2. 3 6 III 12 2. 3 l4 R/R 12
I
3. 7 7 O/R 8 3. 7 17 O/R 8
4. 1 9 All 12 4. 1 19 Elk 12
1. 2 2 RI! 6 1. 2 10 III 6
2. O 4 ill 8 2. O 13 Ill 8
ll
3. 6 6 ale 6 3. 6 14 CIR 6
4. 4 0 CIR 8 4. 4 17 Ill 0

 

1. Ranking Order is free 1 = least legible to 4 = aost legible.
2. Froa Appendix A, page 100, eean rating for expert panel.

3. Free Appendix B, page 108, seen reading for subjects using ORR visibility peter.

(52

iable O
Legibility Rating by Expert Panel vs. Consuaer 20/30
Readings on ORR Visibility Hater

 

 

 

 

Expert Panel OAR Rater
Ressage Rank Cap Rating2 Color Size Rank Cap Reading2 Color Size
OrderI No. (Pts) Order No. (Pts)
1. 5 4 AIR 0 1. 5 10 AIR 0
2. 3 6 All 12 2. 3 13 I]! 12
I .
3. 7 7 Oil 0 3. 7 15 all 8
4. 1 9 O]! 12 4. 1 19 AIR 12
l. 2 2 AIR 6 1. 2 9 R/R 6
2. 8 4 All 8 2. O 11 I/R O
11
3. 6 6 CIR 6 3. 6 13 RIP 6
4. 4 0 CIR 8 4. 4 16 11v 8

 

1. Ranking Order is free 1 = least legible to 4 = cost legible.
2. Prop Appendix A, page 100, seen rating for expert panel.

3. Free Appendix C, page 100, seen reading for subjects using ORR visibility aeter.

(i3

iable 9
Legibility Readings on Polariscope vs. Readings on OAR Visibility
Aeter by Consuaer 20/20 in Order Rankings

 

 

 

 

Polariscope OAR later
Aessage Rank Cap Rating2 Color Size Rank Cap Readingz Color Size
Order! No. (Pts) Order No. (Pts)
1. 5 9 VII 8 1. 5 11 V/k 8
2. 3 9 VII 12 2. 3 14 I/V 12
I
3. 7 6 CIA 8 3. 7 17 S/R 8
4. l 5 all 12 4. 1 19 O/k 12
1. 2 14 Ali 6 1. 2 10 V/A 6
2. O 10 V/R O 2. 8 13 AIR 8
11
3. 6 8 all 6 3. 6 14 AIR 6
4. 4 6 DIR B 4. 4 17 DIR B

 

l. Ranking Order is froa 1 = least legible to 4 = aost legible.
2. Froa Appendix 0, page 107, aean reading for subjects using polariscope.

3. Free Appendix 0, page 100, aean reading for subjects using OAR visibility eater.

(i4

iable 10
Legibility Readings on Polariscope vs. Readings on OAR Visibility
Hater by Consuaer 20/30 in Order Rankings

 

 

 

 

Polariscope OAR Hater
Aassage Rank Cap Ratingz Color Size Rank Cap Reading2 Color Size
Order1 No. (Pts) Order No. (Pts)
1. 5 17 All 0 1. 5 10 HIV 8
2. 3 13 AIR 12 2. 3 13 VII 12
1
3. 7 11 Elk O 3. 7 15 O/V O
4. 1 8 CIA 12 4. 1 19 Old 12
1. 2 24 VII 6 1. 2 9 HIV 6
2. 6 15 O/k 6 2. O 11 HIV 8
ll
3. O 14 AIR 8 3. 6 13 81V 6
4. 4 11 O/R O 4. 4 16 O/V O

 

1. Ranking Order is froa = least legible to 4 : aost legible.
2. Froa Appendix 0, page 107, seen reading for subjects using polariscope.

3. Free Appendix 0, page 108, aean reading for subjects using OAR visibility eater.

65

A correlation coefficient of 1.00 indicates a perfect
positive relationship or agreement while ‘that of. -1.00
indicates a perfect negative relationship or complete
disagreement. In the discussion on correlation Theodore
Horvath (19) claims that before the result can be
interpreted, the r coefficient must be tested for
significance. The test of significance establishes whether,
at some confidence level, the present outcome should be
viewed as the result of a sampling error. A test of
significance for the r coefficient was computed before the
interpretation of the results was made. The r coefficient
was tested by comparison with the r critical value at the
0.05 significance level obtained from Table H, page 360, of
"Basic: Statistics for Behavioral. Sciences" (Theodore
Horvath) (20).

The following correlations were tested for r critical:

(i) Experts vs. Consumers 20/20 using polariscope

(ii) Experts vs. Consumers 20/30 using polariscope
(iii) Experts vs. Consumers 20/20 using DMR meter

(iv) Experts vs. Consumers 20/30 using DMR meter

(v) Polariscope vs. DMR meter using Consumers 20/20

(iv) Polariscope vs. DMR meter using Consumers 20/30

All the null hypotheses are stated as: no difference/no
effect. The null hypotheses are rejected when the
calculated r equals or exceeds the critical value obtained
from Table H. These hypotheses are tested at the 0.05

significance level. The value of the statistic r, r

66

critical and the decision on statistically significant
difference are recorded in Table 11.

Reviewing the results of this study from Table 11, item
(1), Experts vs. polariscope 20/20, shows a strong negative
relationship between the expert ratings and the readings for
consumers with 20/20 visual acuity on the polariscope.
This relationship is such that the higher ratings of
legibility by the experts is associated with lower readings
on the polariscope by the consumers with 20/20 visual
acuity. The same is true of number (2), Experts vs.
polariscope 20/30. Figures 1 and 2 show bivariate plots for
the data given in Tables 5 and 6. This strong correlation
is further corroborated by Tables 5 and 6 where the caps’
order ranking by consumers with 20/20 visual acuity is the
same as that of the experts. The order ranking changes
slightly for the 20/30 consumer group when reading message
II (Table 6). These subjects saw the 8 point, white-on-
white print as slightly easier to read than the black on
white 6 point print.

Table 11, item 3 and 4, experts vs. DMR visibility
meter 20/20 and 20/30 shows a strong positive correlation
between the experts’ rating of legibility and the DMR
visibility meter readings when viewed by the consumers with
20/20 and 20/30 visual acuity. The relationship is such
that the higher ratings of legibility by the experts are
associated with higher readings on the DMR visibility meter

by the consumers. This relationship is corroborated by

67

Table 11
Correlation Values and Statistical Significance for Expert
Panel Ratings and Consumer Panel Machine Readings (Horvath
Test of Significance with r Critical = 0.707).

 

 

 

Statistical
Significance

Source of Correlation r Coefficient at 0.05
significance

level

1. Experts vs. M; 20/20 —0.950 Yes

2. Experts vs. My 20/30 -0.951 Yes

3. Experts vs. M; 20/20 0.966 Yes

4. Experts vs. M2 20/30 0.974 Yes

5. M1 vs. M: 20/20 -0.911 Yes

df = N-2 = 8-2 = 6 (8 caps were observed).

M; = polariscope viewed by Consumers with the given visual

acuity.
M2 = DMR visibility meter viewed by Consumers with the given

visual acuity.

P-Scope Reading (degrees of rotation)

68

Figure 1

Bivariate Plat showing correlation between the
expert's rating and polariscope reading by
consumers with 20/20 visual acuity.

 

 

 

16—

14— '-

12-4

10—

8—

5..

4_

2i

0 I I I r I l r I l j

0 2 4 6 8 10
Expert's Rating (on a scale of 1 — 10)

Correlation = -—O.950
Intercept = 15.160
Slape = —1.180

P-Scope Reading (degrees of rotation)

26
24.4
22-

204

69

Figure 2

Bivariate Plot showing correlation between the
expert's rating and polariscope reading by
consumers with 20/30 visual acuity.

 

 

I I I I I

0 2 4 6 8 10

Expert's Rating (on a scale of 1 - 10)

Correlation - —0.951
Intercept = 25.590
Slope = -1.993

70

Tables 5 and 6 where the caps’ order rankings for consumers
with the two visual acuities are in total agreement with
that of the experts. Figures 3 and 4 show bivariate plots
for the data given in Tables 7 and 8 illustrating these
correlations. It is evident from the bivariate plots that a
stronger relationship exists between the experts and DMR
visibility meter 20/30 than 20/20, as shown by the closeness
of values to the regression line.

From the analysis of Table 11 and the subsequent
discussions on correlation it can be stated that the DMR
visibility meter readings agree well with the ratings given
by the experts. This is reinforced by the order rankings
for the caps in Tables 7 and 8 in which consumer rankings
follow exactly those of the experts. Since the experts’
reliability was high, the high degree of correlation of
experts with consumers was taken as at least partial
validation for examining the correlation between the
instruments.

5.3. Compgrigga_LQggrelationl of the Pala§i§992e_and*DMR
1.1th

From the correlations reported in Table 11, there is a
very strong negative relationship between the polariscope
and DMR visibility meter at both visual acuities. This
relationship is such that the higher readings of legibility
by consumers on polariscope is associated with lower
readings on the DMR visibility meter. The negative

relationship is explained by the difference in scales of

DMR Reading (Visual Angle in Minutes)

71

Figure 3

Bivariate Plot showing correlation between the
expert's rating and DMR visibility meter reading I
by consumers with 20/20 visual acuity.

 

 

 

4-
2..
o U l I T 1 1 T T T l
O 2 4 6 8 10
Expert's Rating (on a scale of 1. — 10)
Correlation = 0.966
Intercept = 6.900

Slope 1 .300

DMR Reading (Visual Angle in Minutes)

20—

18-1

72

Figure 4

Bivariate Plot showing correlation between the
expert's rating and DMR visibility meter reading
by consumers with 20/ 30 visual acuity.

 

 

 

infrir l
2 4 6 8 10

Expert's Rating (an a scale of 1 - 10)

Correlation = 0.974
Intercept = 5.200
Slope = 1.400

73

these instruments. The scales are reversed, with a high
legibility value for polariscope is associated with a low
number (small rotation) while on the DMR visibility meter,
it is associated with a high number (large rotation of the
density wedge). The polariscope’s scale ranges from 0-90
degrees while that of the DMR visibility meter is 1-20 and
therefore is much more compressed than the polariscope
scale.

Tables 9 and 10 show the order ranking for legibility
of the eight caps for both instruments, with the consumers
group 20/20 and 20/30. The cap order ranking for both
instruments is the same for the consumers 20/20 group and
this is reinforced.tnr the high correlation (-0.911) while
that of the 20/30 is slightly lower (-0.905) since the order
ranking changes slightly. Once again, it is for message II,
with this group that the ranking is different for the
polariscope than for either experts or the DMR visibility
meter.

Figures 5 and 6 show the negative linear relationship
between the polariscope and DMR visibility meter for the
data from Table 9 and 10, for consumers with the two visual
acuities. From the correlations obtained between the
experts and both of the instruments, it is evident that both
can be used to evaluate legibility since their scale
readings agree with the ratings obtained from the experts
with the same relative numerical order of the caps despite

the reversed scales. The question that arises then, is how

DMR Reading (Visual Angle in Minutes)

74

Figure 5

Bivariate Plot showing correlation between the
Polariscope reading and DMR visibility meter
reading by consumers with 20/20 visual acuity.

 

 

 

5..
4._I
2..
O I 1 T I I I I l I I 1 I I I
O 2 4 6 8 10 12 14
P-Scope Reading (degrees of rotation)
Correlation = —0.91 1
Intercept = 22.640
Slope = —0.987

DMR Reading (Visual Angle in Minutes)

 

75

Figure 6

Bivariate Plot showing correlation between the
Polariscope reading and DMR visibility meter reading
by consumers with 20/30 visual acuity.

 

 

j T I

I l' I r I I r I
2 4 6 8

1 I T

11'] I l l 1']
1012141618 20 22 24

P-Scope Reading (degrees of rotation)

Correlation
Intercept
Slope

-0.905
22.010
-0.620

76

to rate their performance. Performance of the two
instruments can be evaluated and rated only if their
individual reliabilities are established, according to run
Joshua G. Bagaka (2), Statistical Consultant, Research and
Evaluation Services, at Michigan State University. After
computing their reliabilities or consistency, the two
instruments can then be compared and rated. Reliability
analysis requires at least two measures of the same thing
from a random population. This was done for the experts,
but the legibility test of the eight CRP caps was performed
only once for each instrument and the data collected was not
adequate to evaluate the instruments' reliability. It is
therefore not feasible to compare the two instruments’
performance with the available data and this should be a
subject for another investigation.

The polariscope has the advantage of an extended scale
(1-90 degrees) unlike the DMR visibility meter’s which is
compressed (1-20 visual angle in minutes). Due to the
compressed scale, more light tends to be allowed in per
increment of rotation, while viewing a target. In
polariscope the light transmission increase tends to be more
gradual which results in a relatively long scale and small
change in light per increment (greater sensitivity).
Consumers interviewed immediately after using the two
instruments tended to be more comfortable with the
polariscope than with the DMR visibility meter. They felt

it was easier to manipulate and the wide viewing screen did

77

not strain the users as did the DMR visibility meter with
its binocular structure. One advantage of the DMR
visibility meter over the polariscope is that it can be
administered to consumers in a wider variety of situations
because it is lighter in. weight and less bulky. The
polariscope is designed to be used primarily in the
laboratory where illumination can easily be controlled.

5.4- Ef£§9t_gfnmessasellgglor Qantrastllllxn§_§iaelsnd_hx

wisuallAcuitxlQn_themhegibilitx_af_thelQRBlQapslfarlthe
ExberialandlQansumers.lysinslth§.1sallnatrumentsl

The statistical evaluation of the above effects was
based on a Split-Plot Factorial design, Type SPF-pru.q
design (6) utilizing analysis of variance (ANOVA). SAS.A1
Program was used for this analysis. There were four
treatments or main effects and eleven treatment combinations
or interactions of the main effects for the consumers using
the two instruments. In the case of the experts, there were
three main treatments and four treatment combinations for
the examination of interaction effects. Table 12 shows the
main treatments .and the treatment combination for the
experts and consumers in this test. Visual acuity for the
experts was not important in this test. The layout or block
diagram of this design (SPF - 222.4 Design) and its

structural model (6), is shown in Appendix B, page 109.

78

Table 12
Main Treatments and Treatment Combinations

 

Source Treatments Possible Combinations

 

Message (A) AB, AC, AD, ABC, ABD ACD and ABCD

 

Visual (B) BC, BD, and BCD
Consumers

Color (C) CD

Size (D) -

Message (A) AB, AC, and ABC
Experts Color (B) BC

Size (0) -

 

The ANOVA computational procedure (6) was used to
evaluate the significance of the main and treatment
combinations in Table 12. The ANOVA table for the Experts
is shown in Appendix A, page 103, while that of the
consumers using the two instruments is shown in Appendix B,
page 111, for the polariscope and page 115 for the DMR
visibility meter. In this test, all the main effects were
significant (except for visual acuity on the DMR visibility
meter) and are reported in Table 13 in summary form.

In calculating the F value, all the null hypotheses
were stated as: no difference/no effect.

The null hypotheses are rejected when the probality > F
value is equal to or less than 0.05. Table 13 shows that

all the main effects were significant for both the experts

79

Table 13
ANOVA Table for the Significant Main Effects

 

 

 

 

 

Statistical
Significance
Test Group Source of DF ANOVA SS F value at 0.05
Variation significance
level
Message 1 45.00 22.59 Yes
Expertsl Color 1 248.51 124.77 Yes
Size 2 129.06 32.40 Yes
Message 1 333.51 11.62 Yes
Visual 1 971.80 33.85 Yes
Consumers2
(polariscope) Color 1 888.31 30.94 Yes
Size 2 810.51 14.11 Yes
Message 1 126.03 10.77 Yes
Consumers3 Visual 1 43.89 3.75 No
(DMR
Visibility Color 1 960.40 82.08 Yes
Meter)
Size 2 432.85 18.50 Yes
1. All main effects significant at 0.9999 confidence level.

2.

Message significant at 0.9992 confidence level. All the
other effects significant at 0.9999 confidence level.

Message significant at 0.9987 confidence level. For
visual acuity, a = .0547, so this effect should be noted
along with the main effects. All the other effects

significant at 0.9999.

80

and. the consumers apart from ‘visual acuity on. the DMR
visibility meter. This means that message, color contrast,
size and visual acuity differences affected the legibility,
while the interactions did not significantly' affect the
legibility.

When ANOVA shows significance between two treatments it
is not necessary to do further testing as interpretations
can be made directly, but when the ANOVA shows significance
for three or more treatments, it is necessary to do a post-F
comparison procedure in order to identify which treatments
differ significantly. There are many such tests and
Turkey’s HSD (Honestly Significant Difference) test (21) is
one of them. According to Theodore Horvath (21), Tukey’s
HSD procedure would be applied by determining the critical
difference between independent variable level totals or
pairwise comparisons among means. SAS.A1 Program was used
for this analysis. A comparison involving two means is
declared to be significant if it exceeds HSD, which is given

by

 

where q is obtained from the distribution of the
studentized range statistic as given in Table
D.7.1 of Statistical methods in Education and
Psychology, Glass/Hopkins. (7)
a = significance level, in this case, 0.05

v = degree of Freedom

81

MSE Mean Square Error Term

n number of scores in each group
An example of the mathematical computation of HSD is shown
in Appendix A page 105.

Even though the ANOVA values for message, color and
visual acuity were significant (see Table 13), and allowed
direct interpretation immediately, Tukey HSD analysis
results are also reported for them (Tables 14, 15 and 16).
These results support the ANOVA ‘values, but are infact
redundant and not requiredn However, there ‘were three
treatments for type size which were significant and for
these ANOVA alone did not identify the individual treatments
that were significantly different. It was therefore
necessary to do a Tukey HSD test to determine which levels
between the three type sizes differed significantly.
Turkey’s Studentized Range (HSD) test for experts and
consumers is reported in Tables 14, 15, and 16 in summary
form. The legibility ratios of the means for the pairs of
variables are reported in Tables 17 and 18. The full
analysis is shown in Appendix A, page 103 to 105 for the
experts, and Appendix B, page 112 to 114 for consumers using
polariscope and page 116 to 118 for consumers using the DMR
visibility meter. The null hypothesis adopted for this test
was: no difference/no effect. The hypotheses were tested at
the 0.05 significance level. Any hypothesis was rejected if

the difference between the means exceeded the critical value

82

Table 14
Results of Tukey’s HSD Test for Comparison of the Means of
the Main Effects for the Experts

 

 

 

 

 

 

Difference Statistical

Source of Between HSD‘ Significance
Comparison Means Means Statistic at .05 Level
Message 1 6.3375

1.500 0.6290 Yes
Message 2 4.8375
Color 1 - B/W 7.3500

3.5250 0.6290 Yes
Color 2 - W/W 3.8250
Size 1 - 6pts -

1.5375 1.5102 Yes
Size 2 - 8pts -
Size 1 - 6pts -

3.5750 1.5102 Yes
Size 3 - 12pts -
Size 2 - 8pts -

2.0375 1.5102 Yes

Size 3 - 12pts -

 

# The BSD statistic for message and color reported from the
"Minimum Significance Difference" values shown in Appendix A
page 104. The HSD statistic for the type sizes is
calculated using the BSD formula by, using the values shown

in Appendix A, page 105.

83

Table 15
Results of Tukey’s HSD Test for Comparison of the Means of
the Main Effects for the Consumers Using Polariscope

 

 

 

 

 

 

 

Difference Statistical

Source of Between HSD‘ Significance
Comparison Means Means Statistic at .05 Level
Message 1 8.4375

2.8875 1.6746 Yes
Message 2 11.3250
Color 1 - B/W 7.5250

4.7120 1.6746 Yes
Color 2 - W/W 12.2375
Visual 1 - 20/20 8.2679

5.3779 1.8271 Yes
Visual 2 - 20/30 13.6458
Size 1 - 6pts -

6.7948 5.6746 Yes
Size 2 — 8pts —
Size 1 - 6pts -

6.1000 5.6746 Yes
Size 3 - 12pts -
Size 2 - 8pts -

. 1.7625 5.6746 No

12pts

Size 3

 

‘ The HSD statistic for message, color and visual is
reported from the "Minimum Significance Difference" values,
shown in Appendix B, page 112 to 113. The HSD statistic for
the type sizes is calculated using the BSD formula by using

the values shown in Appendix B, page 114.

84
Table 16

Results of Tukey’s HSD Test for Comparison of the Means of
the Main Effects for the Consumers Using
DMR Visibility Meter

 

 

 

 

 

 

 

Difference Statistical

Source of Between HSD' Significance
Comparison Means Means Statistic at .05 Level
Message 1 14.5625

1.7750 1.0690 Yes
Message 2 12.7875
Color 1 - B/W 16.1250

4.9000 1.0690 Yes
Color 2 - W/W 11.2250
Visual 1 - 20/20 14.0179

1.1429 1.1664 No
Visual 2 - 20/30 12.8750
Size 1 - 6pts -

3.9937 3.6227 Yes
Size 2 - 8pts -
Size 1 - 6pts -

4.6500 3.6227 Yes
Size 3 - 12pts -
Size 2 - 8pts -

3.7937 3.6227 Yes

Size 3 12pts

 

* The BBB statistic
reported from the

shown in Appendix B, page 116 to 117.

for message, color

"Minimum Significance Difference"

and visual

is

values

The HSD statistic for

the type sizes is calculated using the BSD formula by using

the values shown in Appendix B, page 118.

85

of the studentized range or HSD. Therefore the comparison
involving the two means was declared significant.

An examination of Tables 14, 15, and 16, indicates that
the three pairwise comparisons of the two messages are
statistically significantly different. Tables 17 and 18
show the ratios of legibility which demonstrated the
practical importance of the differences. Message I was
almost 1.3 times more legible to the experts than message II
(see Table 17). This difference is logical when one
considers that the message content and message layouts were
different on the two sets of caps evaluated. The difference
is greater with the polariscope than the DMR visibility
meter (1.3 and 1.1 times respectively). The data in
Appendix A, page 100 on experts’ average ratings reveals
that, as the message II became harder to read, the standard
deviation increased.

Color contrast greatly affected the legibility of CRP
caps as the results of this test showed (see Table 17). For
the experts, it was twice as easy to read the caps printed
in 'black-on-white as the white-on-white. This is also
supported by the results in Table 14 where the difference
between the two colors was greatly significant. This was
also true for the consumers using the two instruments
(Tables 15, 16 and 17). It was 1.6 and 1.5 times easier to
read the print in black-on-white than white-on-white with

the polariscope and DMR visibility meter respectively, and

86

 

 

 

Table 17
Legibility Ratios‘ of the Means for the Pairs of Variables
Legibility
Experts Polariscope DMR Visibility

Variables Ratio Ratio Meter Ratio
Message 1.3x 1.3x 1.1x

Color

Contrast 2.0x 1.6X 1.5x

Visual

ACUity - 107x 1.1X

 

For message,

the ratio is:

IMW
legibility of message II

For color contrast, the ratio is:

leaibiliLLgLBlscltzmflhite
legibility of White-on-White

For visual acuity,

legibilithQLZQLZQ
legibility for 20/30

the ratio is:

87

Table 18
Type Size Legibility Ratios'

for Message I and II Means

 

 

 

 

Message I Message 11
Color Size Mean Ratio Color Size Mean Ratio
W/W 8pts 4 W/W 6pts 2
1.5x 2.0x
W/W 12pts 6 WIN 8pts 4
Experts
B/W 8pts 7 B/W 6pts 6
1.3x 1.3x
B/W 12pts 9 B/W 8pts 8
W/W 8pts 6 WIN 6pts 14
0.67X 1.4x
W/W 12pts 9 W/W 8pts 10
Polariscope
B/W 8pts 6 B/W 6pts 8
1.1x 1.3x
B/W 12pts 5 B/W 8pts 6
W/W 8pts 17 W/W 6pts 10
0.82X 1.3x
DMR W/W 12pts 14 W/W 8pts 13
visibility
meter B/W 8pts 17 B/W 6pts 14
1.1x 1.2x
B/W 12pts 19 B/W 8pts 17

 

‘ For type size,

the ratio is:

lejjbliliWW ' r '
legibility of smaller type size

88

the difference was greater in the polariscope than the DMR
visibility meter.

Legibility was affected by visual acuity of the
consumers tested using the polariscope ‘but not the DMR
visibility meter. In general, Tables 9 and 10 for the
polariscope and DMR visibility meter order rankings show
that the consumers with 20/20 visual acuity found it easier
to read the two messages than those with 20/30. Table 15,
for polariscope, shows a significant difference of the two
acuities. According to the legibility ratios in Table 17,
the consumer group with 20/20 visual acuity found it easier
to read the messages on the CRP caps by more than one and a
half times. For the DMR visibility meter the difference
between the two acuities was almost significant at the 0.05
level (See the ANOVA Table in Appendix B, page 115). This
is reinforced by the ratio on Table 17, which shows the
consumer group with 20/20 visual acuity able to read the cap
messages 1.1 times easier than those with 20/30. The reason
for this small difference between the two acuities‘ can
probably be accounted for by the compressed scale of this
instrument. The readings tended to be very close together
and this was reflected by the very small difference between
the means of the two visual acuities.

The effect of type size on legibility was evaluated. A
comparison was made between 6 and 8 pts, 6 and 12 pts and 8
and. 12 ‘pts (Tables 14, 15, 16 and 18) to see whether

significant differences existed. There is a significant

 

89
difference between 6 pts and 8 pts for the experts and the

consumers with the two instruments. Table 18 shows a high
ratio of legibility for message II which ranged from 1.2 to
2 ‘times for the experts, jpolariscope and. DMR ‘visibility
meter. The 8 pt size was easier to read than the 6 pt size
under the same conditions. The greatest difference is shown
between 6 pts and 12 pts for the experts and the consumers
with the DMR visibility meter (See Tables 14 and 16). This
may not mean alot since we are comparing 12 pts in message I
and 6 pts in message II whose type faces, message content
and layout were different, but in general these differences
have been demonstrated, especially where uniform type sizes
have been used (8). The difference between 8 pts and 12 pts
is significant for the experts and consumers using the DMR
visibility meter, but not the polariscope. In the two cases
above, the 12 pt message was easier to read than the 8 pt
message, but in the case for polariscope, the two type size
difference did not significantly affect the legibility
results. This can be attributed to the closeness of values
obtained for both 12 and 8 pt sizes. This is further
supported by the order rankings on Table 9 where the ratings
or scores for 8 and 12 pts were very close. The size
anomaly between 8 and 12 pts is shown on Table 18, where the
ratio for the white-on-white variation for both instruments
is less than 1, which means that 12 pts was less legible
than 8 pts. The effect of glare could have contributed to

this small difference ratio in legibility, although this

,“

90

argument is not supported by the white-on-white for message
II. However, these two cases further reinforces that a lack

of contrast can affect legibility.

CHAPTER 6

§UMMWEX_A§D_RECOMMEEDAIIQNfi

 

One of the objectives of this research was to
investigate whether the two instruments’ readings gave the
same results as the expert panel. The expert panel was used
as the standard. or reference control and therefore its
reliability or internal consistency had to be proved before
any correlation could be done. This evaluation showed a
high experts’ reliability ().80) which meant that they could
be used as a reference control against which the consumers’
readings could be correlated. The results showed a very
high correlation (>.90) between the experts’ ratings and the
readings obtained. by the consumers using the two
instruments. The correlations tended to be higher with the
DMR visibility’ meter which had. a jpositive linear
relationship, than the polariscope which had a negative
linear relationship. Both instruments gave results similar
to the experts and therefore either one can be used to
evaluate legibility of visual elements for packaging.

When compared against each other, the two instruments
show a very high correlation (>.90) with negative slope.
The negative relationship is explained by the reversed

scales of the two instruments. Because it is hard to rate

91

92

their performance without knowledge of their reliability, it
is recommended that the two instruments be evaluated for
reliability. With a known reliability it would then be
possible to evaluate the two instruments’ validity
independently since the validity of a test is influenced by
its reliability (27). There are some advantages offered by
the polariscope which makes it superior to the DMR
visibility meter.

The extended scale of the polariscope makes it more
sensitive to changes which are not recorded in detail by the
DMR visibility meter because of its compressed scale. One
of the questions raised by this study is why the visual
acuity of the observers did not have significant difference
for the DMR visibility meter as it did for the polariscope.
This can probably be attributed to the DMR visibility
meter’s lower sensitivity because of its compressed scale
which is reflected by the closeness in the values obtained
for legibility. This is also reinforced by the fact that
visual acuity was almost significant as the results of the
ANOVA show, which is also supported by the small ratio
(1.1x) of the two acuities. 0n the other hand, an
unpublished report on the use of polariscope for measuring
legibility by Lockhart and Michel (8) showed clear
differences between 20/20 and 20/30 visual acuities when
testing for the legibility of visual elements on an ampoule
and a vial. A general observation during this study was

that subjects were more comfortable using the polariscope

 

 

93
than the DMR visibility meter. This can be attributed to

the polariscope’s wide viewing screen, which was easier to
manipulate. Subjects with bi-focals found it easier to use
than the DMR visibility meter whose binocular viewing
filters were small and restrictive.

With the results obtained with the polariscope, it is
now easy to formulate a guideline that can be used by the
manufacturers of CRP caps in testing legibility for the caps
intended for consumers within the range of 20/20 to 20/30
vision.

Message, type size and color' contrast affected the
legibility of the CRP cap label copy. Results showed the
two messages statistically significantly different for the
experts and the consumers with both instruments. However,
message effect, which is affected by message content and
layout was not evaluated. in this study‘ although. it 'was
reflected in these results. A further study of message
effect on legibility needs to be done.

Legibility was also affected by the type size. As the
type got smaller, the message became hard to read.
Legibility was severely diminished when the type was small
(6 pts) and lacked contrast. The results showed a
statistically significant difference between the black-on-
white print and the white-on-white, ranging from one and a
half to two times for the legibility ratio. A color
contrast for these point sizes would greatly alleviate the

current problem of legibility as this study' has shown.

 

94

Alternatively the caps can be manufactured in any other
color apart from the pmesent white-on-white embossing, and
then "reverse-printed" to offer a high contrast. It would
be very hard to give an optimum type size for legibility
purposes without an in depth research, but size 6 to 12pts
were fairly legible, especially when colored black. Size of
print will mainly depend on the size of the cap and the
amount of information required.

Visual acuity of the observers affected the way they
rated the legibility. The consumer group with 20/20 vision
were able to read the two messages more easily than the
20/30 visual group. 30% of the two test groups had 20/30
vision the question that arises is whether tests for
legibility should be conducted at 20/20. A study on
legibility conducted by the School of Packaging for an age
group between 56 and 76 years found a sizeable group with
20/30 and 20/40 visual acuity (See Appendix D, page 122).
These two acuity levels should also be included in future
legibility studies because they are so much a part of the
age group reported to have difficulty with the readability
of the CRP caps. This also raises the question of age
effect on legibility which was not covered by this study.
Note that in Tables 20 and 21 the regular decrease of
legibility with increase in age in every visual acuity group
suggests an age effect which, if tested statistically may

prove to be significant.

 

95

SummmLEesammendatims

1.

Further studies of legibility on the elderly population
because it was not included in this study. The message
and age effect was not evaluated in this study, but its
influence on legibility was observed. Further studies
should be done to evaluate the effects.

During these studies, the order of presentation of the
CRP caps to the test subjects should be fully
randomized to avoid a learning effect which may affect
the results.

Reverse-printing of the CRP caps by manufacturers for
effective contrast to improve the legibility.

Further studies to evaluate the reliability and
validity of the two instruments should be done in order

to compare their performance.

LI ST 0F REFERENCES

8.

10.

REFERENCES

A.S.T.M., Spegifiggtion No. .D4261-83, "Standard
Specification for Labels for Small-Volume (Less Than
100 Ml) Parenteral Drug Containers," (Amgrigan*§pgiety
for Testingmandmflnawterialg), Philadelphia, Pa., 1983,
pp. 1-2.

Bagaka, Joshua, G., Statistical Consultant, Research
and Evaluation Services, Dept. of Urban Studies,
interviewed by Mutune Wa Gitau, Michigan State
University, E. Lansing, Michigan, 12.30 pm., June 13,
1988.

Barton. Leslie w., Algessarisgnmafmthelfielahiseenemaith
the TachistssCOpe for EvainatiamL_§urfse§amDesiznslmaf
gagsumermmgagkages, Thesis for the Degree of M.S.,
Michigan State University, 1968, pp. 4-32.

 

Bowdish, John, F., "Drug Labeling: A Challenge to

ASTM-". Afilmlfitandardizatieamfleas. August 1984. pp- 22-
23.

General Radio Company, Eglariscgpe, Cambridge, Mass. ,
Nov. 1952, p. 2.

Glass, Gene, V., and Hopkins, Kenneth, D., "Split-Plot
Design." Statisticalm”.nethgdalllin_l_fiducaticnl_mand
Egyghglggy, Englewood Cliffs, N.J., Prentice-Hall,

Ibid., Table D.7.

Lockhart, Hugh, E. , Dr. , Professor, and Michel, Peg,
Special ist , School of Packaging , Michigan State
University. The Techistaseepe__and__29iariscene.lfgr
Measurins_lmhegihilitx, Report prepared for ASTM
Committee D10, Denver, Colorado, October 12, 1983, p.
1.

 

Ibid., pp. 5.

Luckiesh, Matthew, "Meaning and Measurement of

Visibility," Light, Visionmandmggeing, D. Van Nostrand
Co., New York, NY. 1944, pp. 177-193.

96

11.

12.

13.

14.

15.

16.

17.

18.

19.

20.

21.

22.

23.

24.

25.

97

Ibid., pp. 139-1570

Luckiesh, M., and Moss, Frank, E., "Visibility of

Objects." Wins. D. Van Nostrand Coo.
New York, NY., 1937’ pp. 168-1800

Morich, D., "Using the Tachistoscope, Semantic
Differential and Preference Tests in Package Design
Assessment," Handbggk 9f Package Designl_Resea£Qh.

Wiley-Interscience, New York, NY., 1981, pp. 128-131.

Oborne, David, J., "Man-Man Communication: Words and

Symbols," WA, Wiley-Interscience, New
York, NY., 1982, pp. 58.

Patterson, Donald, J., and Tinker, Miles, A., "Speed of

Reading and Typography." 1121M a..ke..f£xpe._3.eada_bl.e.
Harper & Brothers Publishers, New York, 1940, pp. 1-12.

School of Packaging, Michigan State University,

LQBWMWWo Final
Report to Consumer Product Safety Commission (CPSC),
April 28, 1988, pp. 38-39.

Swope, Lee, "Visionmetric Testing: How a Package

Communi cates . " Hendbae.k___qf__2ack_as.e..__fiesiaa_' Research.
Wiley-Interscience, New York, NY., 1981, pp. 142-143.

"The Eyes Have It," Mode_rn_£aW, Vol. 34, No.2,
October, 1960, pp. 124-125.

Theodore, Horvath, "Product-Moment Correlation,‘ Basic

StatisticsMhaxiWs. Little. Brown and
Company, Boston, Toronto, 1985, pp. 262-264.

Ibid., pp. 360.
Ibid., pp. 223-2250

Tinker, Miles. A., Legibilitzfigf_£rin_, The Iowa State
University Press, Ames, Iowa, 1963, pp. 7-8.

Ibid., pp. 9-310

Tobolski, Francis, P., Carol Stream, Illinois, DMR
W, Letter sent to Lockhart, Hugh, E.,

Dr., Professor, School of ZPackaging, Michigan. State
University, August 25, 1987.

United States Pharmaggpeia (USP), Twenty—First
Revision, Washington DC: The United States

Pharmacopeial Convention, 1985, p. 24.

 

26.

27.

98

Webster ’ s._1hird..._.u_e.w___..._1_nterna..t.i.an.al-.12.igt_i_enary of the
Emeuas.e.llaahridaad. Springfield. Massu G & C-
Merrian Co., 1976.

 

William, Mehrens, A., and Lehmann, Irvin, J.,
" Re 1 i ab 1 l i ty and Va 1 i d i ty . " Mess.uremenmndixaluatign

in.-....-E.dncatianmaaLBsthlex . Ho 1 t . Ri nehart and
Winston, New York, Chicago, 1975, pp. 87-109.

APPENDIX A:

EXPERT PANEL DATA

89!.DALA;.LQQIQ111191841199.91.13g.§ngtt.flangl

 

Cap Ho. and Rating on 1 Scale of 1-10

 

 

Visual
No. Sex Age Acuity 1 3 4 5 6 7 9 10 11 12 13 14 15 16
1 H 40 20/20 8 7 4 5 3 7 8 2 7 4 6 3 7 4
2 H 40 20/20 9 S 8 4 6 7 9 2 7 8 4 6 7 6
3 H 60 20/30 8 4 7 2 3 5 8 1 5 6 2 4 5 3
4 H 32 20/20 10 10 10 4 9 10 10 2 9 10 5 9 10 9
5 H 42 20/20 10 7 9 3 5 6 10 3 6 7 3 S 8 5
6 F 55 20/20 10 4 9 2 7 9 10 1 4 9 2 B 4 1
7 F 45 20/20 9 5 7 3 7 7 9 3 5 7 4 6 7 4
8 H 41 20/20 9 S 8 3 5 7 9 4 5 7 3 5 5 3
9 H 44 20/20 10 6 7 3 5 7 10 1 6 6 4 5 6 4
10 H 37 20/20 8 5 9 2 7 8 10 1 4 8 3 6 8 3

 

5T9

 

100

 

 

 

 

MeeWMWsMimrtjml
Experts Cap No. and Rating*
No. 1 2 3 4 5 6 7 8
1 8.0 1.5 7.0 4.0 5.5 3.0 7.0 3.5
2 9.0 2.0 6.0 8.0 4.0 6.0 7.0 5.5
3 8.0 1.0 4.5 6.5 2.0 3.5 5.0 3.0
4 10.0 3.5 9.5 10.0 4.5 9.0 10.0 7.0
5 10.0 3.0 6.5 8.0 3.0 5.0 7.0 5.0
6 10.0 1.0 4.0 9.0 2.0 7.5 6.5 1.0
7 9.0 3.5 5.0 7.0 3.5 6.5 7.0 4.0
8 9.0 3.5 5.0 7.5 3.0 5.0 6.0 3.5
9 10.0 1.0 6.0 6.5 3.5 5.0 6.5 4.0
10 9.0 1.5 4.5 8.5 2.5 6.5 8.0 3.5
Mean 9.0 2.0 6.0 8.0 4.0 6.0 . 7.0 4.0

 

Std. Dev. 0.9 1.1 1.6 1.7 1.3 1.8 1.5 1.6

 

* Caps 1, 3, 5 and 7 Message I

Caps 2, 4, 6 and 8

Message II

1(71.

Buddhism.lituna.lx§ia;§aale.llatiml
humming.Epithgientllatria
(Reported in Standardized ltee Alpha)

 

Expert

 

Expert 1 2 3 4 5 6 7 8 9 10

2 .5362 -

3 .5336 .8025 -

4 .7028 .7265 .2864 -

5 -.0259 .3840 .7718 ’1.6397 -

6 .6582 .7897 .7718 .8549 -.2236 -

7 .5921 .6115 .6282 .6599 .2362 .9045 °

8 .2204 .7845 .6914 .3955 .5526 .3479 -.1959 -

9 .8402 .5169 .4226 .8689 -.9076 .8287 .7346 .2988 -

10 .2149 .7183 .2970 .0000 .4825 .0000 -.3492 .6002 -.2937 -

 

Overall Reliability Coefficient - 10 itees : Alpha = .8295 Standardized Itee Alpha : .8278

ll)?!

mammamazfimamiml
Mandamus

 

Expert

 

Expert 1 2 3 4 5 6 7 8 9 10
1 1.0000
2 .3663 1.0000
3 .3639 .6701 1.0000
4 .5417 .5705 .1671 1.0000
5 -.0128 .2376 .6284 -.4505 1.0000
6 .4905 .6524 .3374 .7465 -.1006 1.0000
7 .4206 .4404 .4580 .4924 .1339 .8256 1.0000
8 .1239 .6454 .5283 .2465 .3818 .2106 -.0892 1.0000
9 .7244 .3485 .2679 .7682 -.3122 .7075 .5805 .1757 1.0000

10 .1204 .5604 .1744 .0000 .3180 .0000 -.1487 .4287 -.1280 1.0000

 

103

SAS .
ANALXSLSMMANMEQCMBEJQRWIS

DEPENDENT VARIABLE: SCALE

Source: DF SUM OF SQUARES MEAN SQUARE
Model: 7 380.98125000 54.42589286
Error: 72 143.40625000 1.99175347
Corrected Total: 79 524.38750000
F VALUE PR>F R-SQUARE C.V.
27.33 0.0001 0.726526 25.2581
ROOT MSE SCALE MEAN
1.41129496 5.58750000
SOURCE: DF ANOVA SS F VALUE PR>F
MESSAGE 1 45.00000000 29.59 0.0001
COLOR 1 248.51250000 124.77 0.0001
SIZE 2 129.05625000 23.40 0.0001
MESSAGE*COLOR 1 0.00000000 0.00 1.0000
MESSAGEXSIZE 2 0.00000000 . .
COLOR¥SIZE 2 0.10625000 0.03 0.9737
SAS
AMALX§l§m9E_XAElANQ2.2EQQEDHEE_EQELIHE_§XE§EI§
MAINLEEEEQI§21E§I

CLASS LEVEL INFORMATION

CLASS LEVELS VALUES
MESSAGE 2 1 2
COLOR 2 1 2
SIZE 3 1 2 3

NUMBER OF OBSERVATIONS IN DATA SET = 80

 

104

SAS

ANALYSIWARIWQBDUBEMEQMHEJXBEBIS

MAIN-..EEE_EQ_T§_JIE.§1
TUKEY’S STUDENTIZED RANGE (HSD) TEST FOR VARIABLE: SCALE

NOTE: THIS TEST CONTROLS THE TYPE I EXPERIMENTWISE ERROR
RATE, BUT GENERALLY HAS A HIGHER TYPE II ERROR RATE THAN
REGWQ

ALPHA=0.05 DF=72 MSE=1.99175
CRITICAL VALUE OF STUDENTIZED RANGE=2.819
MINIMUM SIGNIFICANT DIFFERENCE=.62909

MEANS WITH THE SAME LETTER ARE NOT SIGNIFICANTLY DIFFERENT.

TUKEY GROUPING MEAN N ME§§AQE
A 6.3375 40 1
B 4.8375 40 2
SAS

ANALXS1.35.-..QE_YAB.IANQ.EWEBQWQED.QBE-FW
MAINMEEELEIiIESI
TUKEY’S STUDENTIZED RANGE (HSD) TEST FOR VARIABLE: SCALE
NOTE: THIS TEST CONTROLS THE TYPE I EXPERIMENTWISE ERROR
RATE, BUT GENERALLY HAS A HIGHER TYPE II ERROR RATE THAN
REGWQ

ALPHA=0.05 DF=72 MSE=1.99175
CRITICAL VALUE OF STUDENTIZED RANGE=2.819
MINIMUM SIGNIFICANT DIFFERENCE=.62909
MEANS WITH THE SAME LETTER ARE NOT SIGNIFICANTLY DIFFERENT.
TUKEY GROUPING MEAN N COLOR
A 7.3500 40 1

B 3.8250 40 2

105

SAS

ANALXSLSMQE_¥AEIANQE.REQQEDUBE.EQRLIHE_EXREEI§
MALHHEEEEQTSLIESI
TUKEY’S STUDENTIZED RANGE (HSD) TEST FOR VARIABLE: SCALE
NOTE: THIS TEST CONTROLS THE TYPE I EXPERIMENTWISE ERROR
RATE

ALPHA=0.05 CONFIDENCE=O.95 DF=72 MSE=1.99175
CRITICAL VALUE OF STUDENTIZED RANGE:3.384

COMPARISONS SIGNIFICANT AT THE 0.05 LEVEL ARE INDICATED BY

’88!’
SIMULTANEOUS SIMULTANEOUS
LOWER DIFFERENCE UPPER
SIQE CONFIDENCE BETWEEN CONFIDENCE
OOMPARISON LIMIT MEANS LIMIT
3 -2 1.1126 2.0375 2.9624 88*
3 -1 2.5070 3.5750 4.6430 $38
2 -3 -2.9624 -2.0375 -1.1126 84*
2 -1 0.6126 1.5375 2.4624 4*!
1 -3 -4.6430 -3.5750 -2.5070 3*:
l -2 -2.4624 -1.5375 -0.6126 xxx

Artisan e f aJaihenatieaLQeWtiWefismflfiugt
CelQLQQntrssmmtheJmW

 

 

 

MSE
BSD - qa. y ———————
\ n
a = 0.05
V = 72
99" = 2°819 > given in Appendix A, page 105
MSE = 1.99175
n = 40
.1
1.99175
HSD = 2.819 .......
\ 40

l
E?
10

APPENDIX B:

CONSUMER DATA

 

§M§.UHEB..DMA
35H DATA: LEGIBILIIY gfififllfisfi 87 IHE~QQNSUHE3§ USING PQLAQIfiCOEfi AHD QHR VI§I§ILIII Hgtgg

 

POLARISCOPE 0H8 VISIBILITY HEIER

 

Cap Ho.9 1 Degrees of Rotation Cap Ho. 4 Visual Angle in Hinutes

 

 

 

Ho. Age Group Sex Visual Acuity 1 2 3 4 5 6 7 8 1 2 3 4 5 6 7 8
1 18-25 yrs F 20130 6 19 13 8 8 11 9 10 20 11 14 20 13 16 20 15
2 18-25 yrs F 20/20 3 12 17 7 5 10 7 11 14 8 11 15 12 11 16 12
3 18-25 yrs H 20/20 4 11 5 5 7 6 5 6 20 15 18 20 15 19 20 16
4 18-25 yrs F 20/30 3 9 5 5 l4 7 5 10 20 12 18 20 13 17 18 14
5 18-25 yrs F 20/20 3 9 4 4 5 5 4 5 20 14 18 20 15 19 20 18
6 18-25 yrs H 20/20 5 12 10 6 7 7 4 8 20 11 17 17 11 13 17 15
7 18-25 yrs F 20/20 4 11 7 6 8 8 7 7 17 11 15 17 12 12 13 12
8 18-25 yrs F 20/20 5 12 9 6 8 7 6 10 20 9 11 15 10 12 13 11
9 18-25 yrs F 20/20 5 11 5 4 9 8 5 7 15 8 13 13 11 11 14 11
10 18-25 yrs H 20/30 4 17 8 7 13 9 6 5 20 8 12 12 10 10 12 10
11 30-54 yrs F 20/20 4 18 9 7 11 10 9 12 16 6 11 11 8 10 11 9
12 30-54 yrs F 20/20 5 18 11 10 13 11 7 14 20 2 6 11 4 11 11 8
13 30-54 yrs H 20/30 17 42 26 22 35 28 22 32 13 4 5 11 3 7 8 4
14 30-54 yrs F 20/20 5 l6 9 8 10 9 8 10 20 9 13 20 11 15 20 14
15 30-54 yrs H 20/20 7 21 12 9 17 11 8 17 20 5 11 16 7 11 14 9
16 30-54 yrs F 20/30 5 13 7 5 10 8 4 10 20 9 15 15 6 13 15 12
17 30-54 yrs H 20/20 4 17 6 3 9 4 5 7 20 14 14 20 13 16 20 14
18 30-54 yrs H 20/20 3 13 7 6 10 8 7 10 20 11 17 20 14 20 20 17
19 30-54 yrs F 20/30 10 40 15 16 20 23 15 19 20 8 10 15 11 12 17 10

20 30-54 yrs H 20/20 5 25 7 5 8 7 6 9 20 9 14 20 9 13 20 12

 

7 Caps Ho. 1, 3, 5 and 7 = Hessage I and Caps Ho. 2, 4, 6 and 8 = Hessage ll

1()f5

107

 

 

 

WAMGSWS
MW
20/20 20/30
Cap No. Mean STD. DEV. Mean STD. DEV.
1 05 1.1 08 3.8
2 14 1.5 24 10.8
3 09 3.4 13 6.8
4 06 1.8 11 5.1
5 09 1.9 17 7.9
6 08 2.1 15 6.2
7 06 1.4 11 5.6
8 10 2.8 14 7.0

 

* The mean readings have been pooled for the two Age groups.

108

 

 

 

 

 

WJEAMHQSIJQMOIZO’S AND 20130’8 CONSUMERS
USING DMR VISIBILITY MEIER
20/20 20/30
Cap No. Mean STD. DEV. Mean STD. DEV.
1 19 2.1 19 2.1
2 10 3.4 09 2.4
3 14 3.3 13 4.1
4 17 3.1 16 3.5
5 11 2.8 10 2.9
6 14 3.6 13 3.5
7 17 3.7 15 4.5
8 13 3.1 11 3.4

 

* The mean readings have been pooled for the two Age groups.

109

LAXQHTWORMBLQQKMDIAQEAM“Egfimfififiznguxgmlfififizfiz2.4) DESIGN
USED.EQBMANQXAWINLIHISLRESEARCH

 

 

 

 

 

 

 

 

 

 

A MESSAGE I MESSAGE II
C B/W W/W B/W W/W
D 8pts 12pts 8pts 12pts 6pts 8pts 6pts 8pts
B 81 SI SI 81 S1 $1 81 S1
82 $2 82 82 82 82 $2 82
S3 S3 S3 S3 S3 S3 S3 83
S4 S4 S4 S4 S4 S4 S4 S4
85 85 SS 85 85 SS S5 85
$6 86 S6 86 S6 36 S6 86
20/20 S7 S7 S7 S7 S7 S7 S7 S7
88 88 S8 88 S8 88 $8 $8
89 S9 89 S9 S9 S9 39 S9
810 810 S10 810 810 S10 810 $10
811 $11 $11 $11 $11 811 S11 811
$12 $12 812 S12 $12 $12 812 812
813 S13 $13 $13 813 $13 813 813
$14 814 $14 $14 $14 814 814 $14
81 S1 SI 81 S1 81 81 S1
82 82 82 $2 82 S2 82 82
20/30 S3 83 S3 S3 S3 S3 S3 S3
S4 S4 S4 S4 S4 S4 S4 S4
85 S5 S5 S5 85 S5 S5 S5
S6 86 86 S6 S6 S6 86 S6
Treatmentlexels: A - Message
C - Color SPF - ACD.B
D - Size
B - Visual Acuity

Shrueturalmtlddel :
lek1 = u+ai+ya+6i+Bi+(a6)ik+(a6)ii+(aB)ia+laYG)iai+
(GYBlika+(a60)111+1096071k11+1y6)xi+(yfllei+
(y681k11+(6B)ii+Eo(ijkl)

where p = Grand mean of treatment populations.
as = Effect of treatment i of message, which is
constant for all subjects within treatment
population i.
ya = Effect of treatment k of visual acuity.
61 = Effect of treatment 1 of color.
0: = Effect of treatment j of size.
eo(ijkl) = Experimental error, which is independent of

other E’s and is normally distributed with
mean = 0 and variance = 08'.

110

Anisemplegfjsthemeticawpatstignmmelati9n.lr.1.
ferjzsperLsJatina_cf_LeaihilLWanaumem_2MQ

MdWiwlfig.lami.seese

Variable X: Experts’ rating of Cap legibility in mean

order ranking.

Variable Y: Consumers 20/20 readings of legibility by

polariscope in mean order ranking.

 

 

 

 

Cap No. X Y X3 Y2 XY
5 4 9 16 81 36
3 6 9 36 81 54
7 7 6 49 36 42
1 9 5 81 25 45
2 2 14 4 196 28
8 4 10 16 100 40
6 6 8 36 64 48
4 8 6 64 36 48
2X46 2Y6? 2X3302 2Y3619 £XY341
N = 8
NZXY - EXZY
r: —————————————————————————————————

 

fF—Nizxz - (2)111] [szz - lZYl’]
(87(3411- (46)(67)

- 0.9496

'00950

 

111
SAS

ANALXWSISIQE.-..VABIANQE-EEQQEDUEE_EQB...-Q.QN§UMEB§._LL§.I.N. .
BQLABJSQQEE

DEPENDENT VARIABLE: SCALE

Source: DF

Model: 15

Error: 144

Corrected Total: 159

F VALUE PR>F

6.72 0.0001

ROOT MSE

5.35824724

SOURCE:

MESSAGE

VISUAL

COLOR

SIZE

MESSAGEtVISUAL
MESSAGE¥COLOR
MESSAGE*SIZE
MESSAGE*VISUAL*COLOR
MESSAGEiVISUAL*SIZE
MESSAGE*COLOR*SIZE
MESSAGEXVISU*COLO*SIZE
VISUALXCOLOR
VISUAL*SIZE
VISUAL‘COLOR*SIZE
COLOR¥SIZE

U
m

NNNHHOHHNHHNHHH

SUM OF SQUARES
2892.38660714

4134.35714286

7026.74375000
R-SQUARE C.V.
0.411625 54.2264

SCALE MEAN
9.88125000

ANOVA SS F VALUE
333.50625000 11.62
971.80029762 33.85
888.30625000 30.94
810.50625000 14.11

19.65744048 0.68
16.25625000 0.57
0.00000000 .
1.18125000 0.04
0.00000000 .
0.00000000 .
7.53363095 .
30.28601190 1.05
68.50267857 1.19
3.18363095 0.06
102.95625000 1.79

MEAN SQUARE
192.82577381

28.71081349

PR>F

0.0008
0.0001
0.0001
0.0001
0.4094
0.4530

0.8395

0.3061
0.3063
0.9461
0.1702

 

112

SAS
WMIAHQLRRWQBMSQMEBWQ
WWW

CLASS LEVEL INFORMATION

CLASS LEVELS VALUES
MESSAGE 2 1 2
VISUAL ACUITY '2 l 2
COLOR 2 l 2
SIZE 3 1 2 3
NUMBER OF OBSERVATIONS IN DATA SET = 160
SAS
WWI S 0 VAMMMRQQEDEBEJQLQQNSMQUEBSMG
PQLAEISQMIW 013.113.5111

TUKEY’S STUDENTIZED RANGE (HSD) TEST FOR VARIABLE: SCALE
NOTE: THIS TEST CONTROLS THE TYPE I EXPERIMENTWISE ERROR
RATE, BUT GENERALLY HAS A HIGHER TYPE II ERROR RATE THAN
REGWQ
ALPHA=0.05 DF=144 MSE=28.7108
CRITICAL VALUE OF STUDENTIZED RANGE=2.795
MINIMUM SIGNIFICANT DIFFERENCE=1.6746
MEANS WITH THE SAME LETTER ARE NOT SIGNIFICANTLY DIFFERENT.
TUKEY GROUPING MEAN N MESSAOE
A 11.3250 80 2

B 8.4375 80 1

113

SAS
AM I VARMQWflfiJfimNSLME 85.113133
2.912431% OEE___MA- ILEEEEQLILSMJESII
TUKEY’S STUDENTIZED RANGE (HSD) TEST FOR VARIABLE: SCALE
NOTE: THIS TEST CONTROLS THE TYPE I EXPERIMENTWISE ERROR
RATE, BUT GENERALLY HAS A HIGHER TYPE II ERROR RATE THAN
REGWQ

ALPHA=0.05 DF=144 MSE=28.7108
CRITICAL VALUE OF STUDENTIZED RANGE=2.795
MINIMUM SIGNIFICANT DIFFERENCE=1.8271

WARNING: CELL SIZES ARE NOT EQUAL.
HARMONIC MEAN OF CELL SIZES=67.2

MEANS WITH THE SAME LETTER ARE NOT SIGNIFICANTLY DIFFERENT.

TUKEY GROUPING MEAN N VLSOAL ACUITY
A 13.6458 48 2
B 8.2679 112 1
SAS
8.01M filial—0 VABLMEflOMCJE HBE...FQB._QQN_UMS ER_§__._1._.US NG
EQLABLSEWNJEEEQMLSI

TUKEY’S STUDENTIZED RANGE (HSD) TEST FOR VARIABLE: SCALE
NOTE: THIS TEST CONTROLS THE TYPE I EXPERIMENTWISE ERROR
RATE, BUT GENERALLY HAS A HIGHER TYPE II ERROR RATE THAN
REGWQ
ALPHA=0.05 DF=144 MSE=28.7108
CRITICAL VALUE OF STUDENTIZED RANGE=2.795
MINIMUM SIGNIFICANT DIFFERENCE=1.6746
MEANS WITH THE SAME LETTER ARE NOT SIGNIFICANTLY DIFFERENT.
TUKEY GROUPING MEAN N OOLOR
A 12.2375 80 2

B 7.5250 80 l

114

SAS
ANALYSISMETAHQEJBQQEDIJBEWEQE“-.®N§.11MEB§211§_LNQ

BQLARISQQREWFJEEEQISIE8I
TUKEY’S STUDENTIZED RANGE (HSD) TEST FOR VARIABLE: SCALE

NOTE: THIS TEST CONTROLS THE TYPE I EXPERIMENTWISE ERROR
RATE

ALPHA=0.05 CONFIDENCE=0.95 DF=144 MSE=28.7108
CRITICAL VALUE OF STUDENTIZED RANGE=3.349

COMPARISONS SIGNIFICANT AT THE 0.05 LEVEL ARE INDICATED BY

’883’
SIMULTANEOUS SIMULTANEOUS
LOWER DIFFERENCE UPPER
SIZE CONFIDENCE BETWEEN CONFIDENCE
OOMPAEISON LIMIT MEANS LIMIT
1 -2 1.8802 4.3375 6.7948 33*
1 -3 3.2625 6.1000 8.9375 *3!
2 -1 -607948 “403375 -108802 ‘**
2 -3 -0.6948 1.7625 4.2198
3 -1 -8.9375 -6.1000 -3.2625 38*
3 -2 -4.2198 -1.7625 0.6948

115

SAS

ANAHSLSLWMQQEDLLEEEEQELQQNSUMEESJSLNLDMR
WE.

DEPENDENT VARIABLE: SCALE

Source: DF

Model: 15

Error: 144

Corrected Total: 159

F VALUE PR>F

8.49 0.0001

ROOT MSE

3.42068096

SOURCE:

MESSAGE

VISUAL

COLOR

SIZE

MESSAGE*VISUAL
MESSAGEXCOLOR
MESSAGE*SIZE
MESSAGE¥VISUALICOLOR
MESSAGE¥VISUAL¥SIZE
MESSAGE*COLOR*SIZE
MESSAGEXVISU*COLO*SIZE
VISUAL*COLOR
VISUAL*SIZE
VISUAL*COLOR*SIZE
COLOR*SIZE

U
m

NNNHOOCONNHHNHH

SUM OF SQUARES
1490.14761905
1684.95238095
3175.10000000

0.469323 25.0141
SCALE MEAN
13.67500000
ANOVA SS F VALUE
126.02500000 10.77
43.88571429 3.75
960.40000000 82.08
432.85000000 18.50
0.86785714 0.07
18.22500000 1.56
0.00000000 .
1.00119048 0.09
0.00000000 .
0.00000000 .
0.00000000 .
1.21904762 0.10
5.62619048 0.24
3.43809524 0.15
9.80000000 0.42

MEAN SQUARE
99.34317460

11.70105820

PR>F

0.0013
0.0547
0.0001
0.0001
0.7858
0.2140

0.7703

0.7473
0.7866
0.8635
0.6587

116

SAS
W.MWWJQNSUMEEWQ
WWWJAW

CLASS LEVEL INFORMATION

CLASS LEVELS VALUES
MESSAGE 2 1 2
VISUAL ACUITY 2 l 2
COLOR 2 1 2
SIZE 3 1 2 3

NUMBER OF OBSERVATIONS IN DATA SET = 160

SAS
ANALYSILQLMAELAELREQQWE-QQNSLJLMEILSWQSJNG
DMRJISIEILITLMEIER..._...: “MAINJEEEQISMIESI
TUKEY’S STUDENTIZED RANGE (HSD) TEST FOR VARIABLE: SCALE
NOTE: THIS TEST CONTROLS THE TYPE I EXPERIMENTWISE ERROR
RATE, BUT GENERALLY HAS A HIGHER TYPE II ERROR RATE THAN
REGWQ

ALPHA=0.05 DF=144 MSE=11.7095
CRITICAL VALUE OF STUDENTIZED RANGE=2.795
MINIMUM SIGNIFICANT DIFFERENCE=1.069
MEANS WITH THE SAME LETTER ARE NOT SIGNIFICANTLY DIFFERENT.
TUKEY GROUPING MEAN N MESSAOE
A 14.5625 80 1

B 12.7875 80 2

 

 

117

SAS '

ANALYSISLQELXAEIANQELBEQQEQUEELEQRMQQNSHMEBSMUSLNG

QMELYISIBILIIXMMEIERL:_MAINMEEEEQI§LIE§I
TUKEY’S STUDENTIZED RANGE (HSD) TEST FOR VARIABLE: SCALE

NOTE: THIS TEST CONTROLS THE TYPE I EXPERIMENTWISE ERROR
RATE, BUT GENERALLY HAS A HIGHER TYPE II ERROR RATE THAN
REGWQ

ALPHA=0.05 DF=144 MSE=11.7011
CRITICAL VALUE OF STUDENTIZED RANGE=2.795
MINIMUM SIGNIFICANT DIFFERENCE=1.1664

WARNING: CELL SIZES ARE NOT EQUAL.
HARMONIC MEAN OF CELL SIZES=67.2

MEANS WITH THE SAME LETTER ARE NOT SIGNIFICANTLY DIFFERENT.

TUKEY GROUPING MEAN N YIfiQALWAQQLIX
A 14.0179 112 1
A 12.8750 48 2
SAS

ANALISISMQEWXARIANQELERQQEDQEELEQEMQQNSQMEE§.Q§INQ

DMRLXISIBILIIXMMEIEE_:LMAIN_EEEEQI§MIE§I
TUKEY’S STUDENTIZED RANGE (HSD) TEST FOR VARIABLE: SCALE

NOTE: THIS TEST CONTROLS THE TYPE I EXPERIMENTWISE ERROR
RATE, BUT GENERALLY HAS A HIGHER TYPE II ERROR RATE THAN
REGWQ

ALPHA=0.05 DF=144 MSE=11.7011
CRITICAL VALUE OF STUDENTIZED RANGE=2.795
MINIMUM SIGNIFICANT DIFFERENCE=1.069
MEANS WITH THE SAME LETTER ARE NOT SIGNIFICANTLY DIFFERENT.
TUKEY GROUPING MEAN N OOLOB
A 16.1250 80 1

B 11.2250 80 2

 

 

118

SAS
ANALX.§..I§..Q.E..Y.AB.LANQE..EBQQEDQEE.EQE-MLSDMMERHSLNG
MILL/1.51.8.111.1..11YLMEIEEMLIIALN..EPLEEQISIESI
TUKEY’S STUDENTIZED RANGE (HSD) TEST FOR VARIABLE: SCALE
NOTE: THIS TEST CONTROLS THE TYPE I EXPERIMENTWISE ERROR
RATE

ALPHA=0.05 CONFIDENCE=0.95 DF=144 MSE=11.7011
CRITICAL VALUE OF STUDENTIZED RANGE=3.349

COMPARISONS SIGNIFICANT AT THE 0.05 LEVEL ARE INDICATED BY

 

’833’
SIMULTANEOUS SIMULTANEOUS
LOWER DIFFERENCE UPPER
SI E CONFIDENCE BETWEEN CONFIDENCE
COMPARISON LIMIT MEANS LIMIT
3 —2 0.6563 2.2250 3.7937 3*:
3 —1 2.8386 4.6500 6.4614 *8!
2 -3 -3.7937 -2.2250 -0.6563 38*
1 -3 -6.4614 -4.6500 -2.8386 ***

APPENDIX C:

PROGRAMS FOR COMPUTING RELIABILITY AND ANOVA

 

 

SP§§ PRQ§RAH [QR CDHPUTINQ THE EXPERIIEAQEL’S RELIAQILIIII

1 0 FILE HANDLE HHSZ]HAHE='RELIA8 DAT A’

2 0 DATA LIST FILE=HH62

3 0 [CAPID 1-2 EXPERT1 T0 EXPERTIO 3‘22

THE ABOVE DATA LIST STATEHEHT HILL READ 1 RECORDS FROH FILE HHGZ

VARIA8LE REC START END FDRHAT HIDTH DEC

CAPID 1 1 2 F 2 0
EXPERTI 1 3 4 F 2 0
EXPERTZ 1 5 6 F 2 0
EXPERT3 1 7 8 F 2 0
EXPERT4 l 9 10 F 2 0
EXPERTS 1 11 12 F 2 0
EXPERT6 1 13 14 F 2 0
EXPERTT 1 15 16 F 2 0
EXPERT8 1 17 18 F 2 0
EXPERT9 1 19 20 E 2 0
EXPERTIO 1 21 22 F 2 0

END OF DATALIST TABLE.

4 0 RELIABILITY VARIABLES=EXPERT1 70 EXPERTIO/
5 0 SCALE(RATIH61)=EXPERTI EXPERTZ]

6 0 SCALE(RATIHC2)=EXPERT1 EXPERT3]

7 0 SCALE(RATIHG3)=EXPERT1 EXPERT4]

8 0 SCALE(RATIH84)=EXPERTI EXPERTS]

9 0 SCALE(RAT1NCS)=EXPERTI EXPERT6]
10 0 SCALE(RATIHG6)=EXPERII EXPERT7/
11 0 SCALE(RATIHGT)=EXPERTI EXPERTB]
12 0 SCALE(RATIN68)=EXPERT1 EXPERT9]
13 0 SCALE(RATIHG9)=EXPERTI EXPERTIO]
14 0 SCALE(RATIN810)=EXPER72 EXPERT3]
15 0 SCALE(RATIH811)=EXPERT2 EXPERT4]
l6 0 SCALE(RATIH812)=EXPER72 EXPERTS]
17 0 SCALE(RATIH813)=EXPERT2 EXPERT6]
18 0 SCALE(RATIHCI4)=EXPER72 EXPERT7I
l9 0 SCALE(RATIH615)=EXPER72 EXPERTB/
20 0 SCALE(RATIH816)=EXPERT2 EXPERT9/
21 0 SCALE(RATIH817)=EXPERT2 EXPERTIO]
22 0 8CALE(RATIH818)=EXPERT3 EXPERT4/
23 0 8CALE(RATIHGI9)=EXPERT3 EXPERTS]
24 0 SCALE(RATIH820)=EXPERT3 EXPERTS]
25 0 SCALE(RATIH821)=EXPERT3 EXPERT7]
26 0 SCALE(RATIH822)=EXPERT3 EXPERTE]
27 0 SCALE(RATIH823)=EXPERT3 EXPERT9]
28 0 SCALE(RATIH824)=EXPERT3 EXPERTlO]
29 0 SCALE(RATIHG25)=EXPERT4 EXPERTS]
30 0 8CALE(RATIH626)=EXPERT4 EXPERT6]
31 0 SCALE(RATIH827)=EXPERT4 EXPERT7]
32 0 SCALE(RATIH628)=EXPERT4 EXPERTB]
33 0 SCALE(RATIH829)=EXPERT4 EXPERT9]

1.1!)

34
35
36
37
38
39
4O
41
42
43
44
45
46
47
48
49
50
51
52

O
0
0
0
0
0
0
0
0
O
0
O
0
0
0
0
0
0
0

173()

SCALE(RATIHS30)=EXPERT4 EXPERTIO]
SCALE(RATIHS31)=EXPERTS EXPERT6]
SCALE(RATIH832)=EXPERTS EXPERT7]
SCALE(RATIH833)=EXPERTS EXPERTO]
SCALE(RATIH834)=EXPERTS EXPERT9/
SCALE(RATIH835)=EXPERTS EXPERTIO]
SCALE(RATIHC36)=EXPERT6 EXPERT7/
SCALE(RATIHC37)=EXPERT6 EXPERTO/
SCALE(RATIH838)=EXPERT6 EXPERT9]
SCALE(RATIHG39)=EXPERT6 EXPERTIO]
SCALE(RATIHG40)=EXPERT7 EXPERTO]
SCALE(RATIH841)=EXPERT7 EXPERT9]
SCALE(RATIH842)=EXPERT7 EXPERTIO]
SCALE(RATIN843)=EXPER78 EXPERI9]
SCALE(RATINC44)=EXPERT8 EXPERTIO]
SCALE(RATIH64S)=EXPERT9 EXPERTIO]
SCALE(RATIHS)=EXPERTI TO EXPERTIO]
HODEL=ALPHA
STATISTICS 3

344444 HETHOD 2 (COVARIAHCE HATRIX) HILL BE USED FOR THIS ANALYSIS 444344

 

121
§fi§ ANOVA PROQILOH FOR 2139111”: I (Tl-SCOPE), HAOHILII 2 (DHLHEIER) AND THE EXPERIJMEL

CHS FILEDEP SIIAU DISH CDHSUHER DATA A;
DATA ALL;
IHFTLE SIIAU;
INPUT 10 1-2 CAPID 3 HACHIHE 4 scALE 6-7 ASE 9 VISUAL 11;
COLOR : 1;
IF CAPID E0 2 on CAPID E0 3 0R CAPID EA 5 0R CAPID EU 8 THEN coLoA : 2;
HESSAGE = 1;
IF CAPID EA 2 on CAPID Ea 4 on CAPID E0 6 0A CAPID EU 6 THEN HESSAGE=2;
SIZE = 1;
IF CAPID Ed 4 0R CAPID E0 5 on CAPID EA 7 on CAPID E0 8 THEN SIZE :2;
IF CAPID EA 1 0R CAPID E0 3 THEN SIZE : 3;
DATA HACHINEl; SEI ALL;
1F mean: = I;
PROC ANOVA;

cLAss nESSASE VISUAL COLOR SIZE;

HODEL scALE = HESSAGE VISUAL COLOR SIZE
HESSAGEFVISUAL AESSASEtcoLoA HESSAGEFSIZE
HESSAGE‘VISUALFCOLOR HESSAGEFVISUAL*SIZE
AESSASEtcoLoAtSIZE HESSAGEFVISUALFCOLDRFSI2E
VISUALtcoLoA VISUALFSIZE VISUALFCDLORFSIZE
COLORFSIZE;

HEARS HESSAGE VISUAL COLOR SIZE
HESSAGEFVISUAL AESSASEtcoLoAt HESSASEISIZE
HESSAGEFVISUALFCOLOR AESSASEtVISUALtSIZE
AESSASEtcoLORISIZE HESSASEIVISUALtcoLoAISIZE
VISUALxcoLoA VISUALFSIZE VISUALFCOLDRFSTZE
coLoAtSIZE / TUKEY;

DATA HACHIHE2; SET ALL; *

1F mum: = 2;

PROC ANOVA;

cLASS HESSAGE VISUAL COLOR SIZE;

HODEL scALE = HESSAGE VISUAL COLOR SIZE
HESSAGE‘VISUAL AEssncEtcoLoA HESSAGEFSIZE
AESSASEIVISUALtcoLon AESSASEtVISUALtSIZE
HESSAGEFCDLORFSIZE HESSAGE‘VISUAL‘CDLDR‘SIZE
VISUALtcoLoa VISUALFSIZE VISUALFCDLORFSIZE
COLDRFSIZE;

HEARS HESSACE VISUAL COLOR SIZE
AESSASEIVISUAL HESSASEtcoLon HESSASEISIZE
HESSASEFVISUALFCDLOR HESSASEFVISUALFSIZE
HESSACEFCOLORFSIZE HESSAGEFVISUALFCDLDRFSIZE
VISUALtcoLUA VISUALFSIZE VISUALtcoLontSIZE
coLoAxSIZE / IUIEI;

HOTE: IHFILE SIIAU IS EILE EerAI IAIA AI
nnoc ANOVA:

CLASS HESSASE COLOR SIZE

HDDEL scALE : AESSASE coLoA SIZE
nESSASEtcoLoA HESSAEEISIZE coLoAtSIZE
HESSASEtcoLoA SIZE;

HEARS HESSACE coLoA SIZE
HESSASEtcoLoA HESSAGEFSIZE coLoAISIZE
HESSASEtcoLUA SIZE ITUHEV;

 

A.
h-

 

APPENDIX D:

LEGIBILITY TEST DATA FOR 56-76+
CONSUMER AGE GROUP USING THE POLARISCOPE

11.5.9.1ELL.I.T.X~IE.§I__E.Q_R..§§;1§.+.MCQNSMMMQELQRQULUSING.-...'T.‘.T..T.E.

During the later stages of the main study it became
important to know how elderly people perceived label copy
and indexing arrows on child resistant caps and bottles in
order to obtain information for the Innovative Child
Resistant Packaging Systems grant study. A test of
legibility was conducted using the polariscope only, with
message 11 caps (nos. 2, 4, 6 and 8) and subjects in the 56-
76+ age group. The subjects were residents of the Lansing
area, all living independently in their own homes. The caps
were presented in the same manner as in the main study. The
individual results appear in Table 19, page 124 which
contains raw data for the cap arrow, bottle arrow and the
cap message. Note that in this age category, there were 31%
who had visual acuity of 20/40.

Tables 20 and 21 show the results for the three age
groups 18-25, 30-55 and 56-76+. A statistical analysis has
not been done on the data to evaluate an age effect, but the
regular' decrease in legibility with age in each visual
acuity level suggests the possibility of an age effect on
the legibility measurements.

Table 22 contains the results when the age 56-76+
subjects observed indexing arrows on the caps and bottles of
snap cap child resistant packages (see Illustration 11).
This extension of the legibility test indicates that color
contrast can be extremely helpful in improving legibility

(or visibility) of important symbols. Note that for the

122

123

20/40 visual acuity, the color contrast provides a four fold
improvement in legibility. Actually this improvement is
greater than four fold because two of the four subjects with
visual acuity of 20/40 could not see the bottle arrow at any
polariscope reading (see Table 19). For these individuals,
the maximum value of 90 degrees was assigned for

computational purposes.

124
Table 19

MWIDATA;JEQIBWEAQLWYLQQ:.LG.$..QQ.N§QMEE.AQE..._QR_-QIIF
U.SING.IHE..BQLAEISCOREJ.LQEQREE§W_QEWBQTAIIQNI

 

Cap Arrow1 Bottle Arrowz Cap Message’

 

 

 

 

W/W B/W W/W B/W W/W B/W
Age Visual

No. Sex Group Acuity 1 3 2 4 1 3 2 4 8 2 4 6
1 M 55-65 20/20 10 15 7 6 40 20 7 10 10 15 7 10
2 F 71-75 20/20 20 10 7 10 32 25 15 9 13 35 14 15
3 M 66-70 20/20 14 13 3 5 24 17 4 5 6 13 5 6
4 M 71-75 20/20 17 12 18 10 45 20 12 17 20 50 14 18
5 F 66-70 20/20 11 8 7 8 36 28 13 15 15 40 15 13
6 M 66-70 20/20 6 6 6 5 42 12 10 9 16 33 9 15
7 F 66-70 20/30 8 11 8 7 32 23 8 10 18 21 13 13
8 M 76+ 20/30 25 19 20 13 25 25 26 12 25 55 19 28
9 F 55-65 20/30 12 8 4 5 t 20 9 7 50 8 17 45

10 M 76+ 20/40 15 12 8 11 t 33 18 15 35 t 27 41

11 F 76+ 20/40 13 13 12 10 25 20 12 14 49 t 29 35

12 F 76+ 20/40 12 8 10 5 45 20 12 12 25 55 17 24

13 F 76+ 20/40 13 17 11 12 t 25 16 15 25 8 19 24

1. Cap arrows nos. 1 and 2 are similar arrows with one in
white-on-white (W/W), and the other in knack-on-white
(B/W). This is the same for cap arrow nos. 3 and 4.
(See Illustration 11, page 127).

2. Bottle arrows nos. 1 and 2 are similar arrows with one
in W/W and the other in B/W. This is the same for
bottle arrow nos. 3 and 4. (See Illustration 11, page
127).

3. Cap message nos. 8 and 4 are two similar caps in 8 pts

(*1

Nos. 2 and 6 are
(See Illustration

with one in W/W and the other in B/W.
in 6 pts and are also in W/W and B/W.
2, page 29).

The subject could not see or read the visual object at
the maximum screen rotation. The maximum rotational
value (90 degrees) was inserted in the calculation for
the subjects.

‘1735

 

 

 

Table 20
CAP TEXT LEGIBILITY 16 PTS) A7 THREE.LE!EL§.OE
Vl§0£L.8£0111.156:l§1_1EGGS1
20/20 20/30 20/40
HIV 8/8 R/R 8/8 R/R Ill

 

Age Group Hean Std. Dev. Hean Std. Dev. Hean Std. Dev. Hean Std. Dev. Hean Std. Dev. Hean Std. Dev.

 

18-25 11.0 1.1 7.0 1.6 15.0 5.3 9.0 2.0 - - - -
30-55 16.0 1.8 9.0 2.5 32.0 16.2 20.0 10.4 - - - -

56-76+ 31.0 14.4 13.0 4.3 55.0 34.5 29.0 16.0 81.0 17.5 31.0 8.4

 

Table 21
CAP IExI LEGIBILITY (a PIS) AT IAAEE LEVILS or
11901L.A£0111.156:161.1§SR§1

 

20/20 20]30 20]40

 

H/H OIH H/H OIH IIH OIH

 

Age Group Hean Std. Dev. Hean Std. Dev. Hean Std. Dev. Hean Std. Dev. Hean Std. Dev. Hean Std. Dev.

 

18-25 8.0 2.2 5.0 1.2 8.0 2.9 7.0 1.6 - - - -
30-55 11.0 3.4 7.0 2.4 20.0 11.1 14.0 8.6 - - - -

56-76+ 13.0 4.9 11.0 4.2 31.0 16.8 16.0 3.1 34.0 11.4 23.0 5.9

 

 

 

 

11265
Table 22
CAP All!) BOTTLE SYHIDLIEAIDILHIMOBELLEVELME
115118.859le .6._§___-7 * YEARS).
20/20 20130 20/40
All 8/H H/H 8/I H/R 8/8

 

Age Group Hean Std. Dev. Hean Std. Dev. Hean Std. Dev. Hean Std. Dev. Hean Std. Dev. Hean Std. Dev.

 

13.0 5.1 8.0 5.1 15.0 8.9 11.0 8.3 13.0 1.3 10.0 1.7

 

CAP
ARROH 11.0 3.3 7.0 2.4 13.0 5.7 8.0 4.2 13.0 3.7 10.0 3.2

37.0 7.7 10.0 4.1 49.0 35.7 14.0 10.1 63.0 32.8 15.0 3.1 ‘
OOTTLE

ARROH 20.0 5.7 11.0 4.4 23.0 2.5 10.0 2.5 25.0 6.2 14.0 1.4

 

 

127
ILLUSTRATION NO. 11

LEGIBILITY TEST SAMPLES

 

Closure text

 

 

 

 

 

 

k /

Translucent bottle White bottle

 

TRTE U

HICHIGRN S NIV. LIBRRRIES
llNlUllllllHllUll lllll lllUlHlHlllllllllllHllHl
3129 7482

300089