Journal of Statistics Education

ISSN: (Print) 1069-1898 (Online) Journal homepage: https://www.tandfonline.com/loi/ujse20

Assistive Technologies for Second-Year Statistics

Students who are Blind

Robert J. Erhardt & Michael P. Shuman

To cite this article: Robert J. Erhardt & Michael P. Shuman (2015) Assistive Technologies for
Second-Year Statistics Students who are Blind, Journal of Statistics Education, 23:2, , DOI:
10.1080/10691898.2015.11889733

To link to this article: https://doi.org/10.1080/10691898.2015.11889733

Published online: 29 Aug 2017.

Submit your article to this journal

Article views: 365

View related articles

View Crossmark data

Citing articles: 1 View citing articles

Full Terms & Conditions of access and use can be found at

https://www.tandfonline.com/action/journalInformation?journalCode=ujse21
 Journal of Statistics Education, Volume 23, Number 2, (2015)

Assistive Technologies for Second-Year Statistics Students who are Blind

Robert J. Erhardt

Michael P. Shuman

Wake Forest University

Journal of Statistics Education Volume 23, Number 2 (2015)

www.amstat.org/publications/jse/v23n2/erhardt.pdf

Copyright c 2015 by Robert J. Erhardt and Michael P. Shuman, all rights reserved. This
text may be freely shared among individuals, but it may not be republished in any medium
without express written consent from the authors and advance notification of the editor.

Key Words: Learning assistance; Statistical models; Tactile image; Visual impairment.

Abstract

At Wake Forest University, a student who is blind enrolled in a second course in statistics.
The course covered simple and multiple regression, model diagnostics, model selection,
data visualization, and elementary logistic regression. These topics required that the stu-
dent both interpret and produce three sets of materials: mathematical writing, computer
programming, and visual displays of data. While we did find scattered resources for blind
students taking mathematics courses or introductory statistics courses, we found no com-
plete account of teaching statistical modeling to students who are blind. We also discov-
ered some challenges in stitching together multiple partial solutions. This paper outlines
our specific approach. We relied heavily on integrating the use of multiple existing tech-
nologies. Specifically, this paper will detail the extensive use of screen readers, LATEX, a
modified use of R and the BrailleR package, a desktop Braille embosser, and a modified
classroom approach.

1
 Journal of Statistics Education, Volume 23, Number 2, (2015)

1. Introduction

1.1 Background

The Americans with Disabilities Act of 1990 (ADA 1990) as amended in the Americans
with Disabilities Act Amendment Act (ADAAA) of 2008 and Section 504 of the Rehabil-
itation Act of 1973 (Section 504 1973) are two pieces of national, civil rights legislation
intended to prevent the discrimination of individuals based on their disability. At the post-
secondary educational level, colleges and universities are obligated to provide “reasonable
accommodations” that would allow students with disabilities the opportunity to participate
in the college or university’s services and programs, including classroom readings, science
labs, etc. Without guidelines and policies provided by the ADA and Section 504 that en-
sure equal access to educational programs, students with disabilities would face barriers
to equal access in postsecondary institutions, and be prohibited from participating fully in
college and university life.

According to the ADA, a disability is considered any physical or mental impairment that
substantially limits one or more major life activities. Most colleges and universities have a
Disability Support Services (DSS) office on campus to help coordinate services and “rea-
sonable accommodations” for students with disabilities. At Wake Forest University, the
DSS office is integrated into the larger Learning Assistance Center and Disability Services
(LAC), which is a general academic support office that, in addition to coordinating ser-
vices for students with disabilities, provides academic coaching and peer tutoring to all
Wake Forest University students. LAC staff members work closely with university faculty
and staff members to facilitate the implementation of accommodations on campus and in
classrooms.

To request academic accommodations, students must present recent documentation of their

disability to the LAC to determine their eligibility to receive services. Then, the LAC staff,
in consultation with other university professionals, (a) review the disability documentation
presented by students and (b) determine which accommodations are considered reasonable.
Examples of reasonable accommodations could include the following: priority registration;
extended time for in-class tests and writing assignments; sign language interpreters; course
substitutions; classroom relocation; assistive technology; or a reduced course load. It is im-
portant that academic accommodations should not substantially alter or modify essential
requirements. Academic accommodations are intended to provide equal access to educa-
tional opportunities, not compromise the integrity of course content.

At Wake Forest University, approximately 7% of students have disclosed some type of

disability and receive academic accommodations in the classroom. The majority of these

2
 Journal of Statistics Education, Volume 23, Number 2, (2015)

students have a diagnosis of a learning disorder or attention deficit / hyperactivity disorder,

and the most frequently provided accommodations for these students are extended time
and a low distraction environment for tests and exams. Neither accommodation requires
much effort from the faculty member at Wake Forest University. Students with disabilities
at Wake Forest University are advised about course selection similarly to students without
disabilities. In other words, all students are encouraged to enroll in courses that satisfy
Divisional requirements at Wake Forest University, are counted toward their major, and/or
satisfy their own personal interests.

At Wake Forest University, a female student presented documentation of a significant vi-

sual impairment to the LAC. Her eye condition, congenital microphthalmia, results in re-
duced visual acuity. In this specific case, the student has significantly reduced visibility
in one eye and a prosthetic eye in the other. While she is able detect light, and she can
identify high contrasts, she has no usable vision for reading. She used Braille throughout
most of her K–12 education and had a moderately high level of knowledge with assistive
technology when she entered the university. She enrolled at Wake Forest University with
the intention of majoring in mathematical business or a similar field, having previously
excelled in her high school AP statistics class.

According to the National Center for Educational Statistics (NCES, Raue and Lewis 2011),
90% of public 4-year and 59% of private 4-year institutions report enrolling 1 or more stu-
dents with visual impairments. Furthermore, students with visual impairments make up 3%
of the total population of students who disclose disabilities among all types of institutions.
Therefore, it is likely that an instructor will have a student with a visual impairment at
some point in his or her teaching career, particularly instructors who teach frequently at
larger institutions. In addition, 87% of public 4-year and 62% of private 4-year institutions
provided adaptive equipment and technology for their students with disabilities.

Readers may wonder about the proper use of terms such as blind versus visually impaired.
For the particular student discussed in this paper, the term blind is most appropriate, and
is not considered offensive. Indeed, one of the major national advocacy groups calls it-
self the National Federation of the Blind (https://nfb.org//). The term visually impaired
encompasses a wide range of visual disabilities and limitations. Many visually impaired
students can manage reading ordinary text with minimal accommodations. Given the par-
ticular disability of the student at Wake Forest University, the Learning Assistance Center
at Wake Forest University suggests that faculty use the term blind to properly convey her
visual impairment. We have therefore chosen to title this paper “Assistive Technologies for
Second-Year Statistics Students who are Blind.”

There has been research and discussion on the challenges of teaching the first statistics

3
 Journal of Statistics Education, Volume 23, Number 2, (2015)

course to blind students. Gibson and Darron (1999) discuss the use of physical manip-
ulatives—such as modeling clay and raised line drawing kits—and how they can be put
to use to convey things like the shape of distributions. Mechan et. al. (1993) similarly
discuss these manipulatives, but also share tips and pitfalls on verbalization (i.e. how to
dictate mathematical writing in a clear, unambiguous way) and testing strategies. Vita and
Kataoka (2014) demonstrate a creative approach to teaching about probability and random
walks with a tactile model. Marson et. al. (2013) provide a more recent and comprehensive
account of teaching introductory statistics to students who are blind. Each of these papers
presents issues and solutions for the introductory course.

Unfortunately, solutions in the first statistics course do not necessarily translate to higher
level courses. Consider data visualization, mathematical writing, and software. In the first
statistics course, graphical representation of data often involve scatterplots, histograms,
boxplots, and so forth. But in a second course, one needs to make residual plots with an
eye towards checking for non-linearity, or auto-correlation of residuals. As the number of
explanatory variables grows, so too does the number of possible scatterplots and residual
plots. The first statistics course can be taught without requiring any computer programming
(the author’s university avoids all programming in the first course). However, in the second
course, one cannot avoid programming to fit logistic regression, use best subsets model
selection, and so forth. Finally, the most complicated mathematical writing in a first course
may only involve a single equation with few mathematical symbols; in the second course,
one computes quantities like Mallow’s Cp (shown later as equation 3).

1.2 Common Assistive Technologies for Students who are Blind

Any product or piece of equipment that is used to improve access to course materials for
students with disabilities is considered assistive technology. Many postsecondary institu-
tions maintain assistive technology for use by students with disabilities. Providing text in
an alternative format is a common accommodation request for students, particularly those
with print-related disabilities, which include visual impairments and reading based learning
disabilities. With text in alternative format, or alt-text, students can access reading materials
(e.g., textbooks, journal articles, workbooks) digitally through the use of assistive technol-
ogy. Examples of assistive technology include text-to-speech and screen reading software,
where a student’s personal computer will “read” the information aloud using a digitized
voice. At Wake Forest University, when students with disabilities request text in alternative
format, they have several options: (a) LAC staff can request digital copies directly from the
publisher and then further edit the digital files, using optical character recognition (OCR)
software, to be used with the student’s preferred assistive technology; (b) students can ac-
cess digital copies of the text from groups like Bookshare (https://www.bookshare.org/) or

4
 Journal of Statistics Education, Volume 23, Number 2, (2015)

Learning Ally (https://www.learningally.org/), which provide files that are ready to be used
with assistive technology; or (c) digital text can be created within the LAC by removing
the binding of the book, scanning the individual pages using a high speed scanner, and
converting the text to be used with the assistive technology.

Although Braille can provide access to written information for students who are visually
impaired or blind, it is generally considered inefficient and cost prohibitive due to the
volume of Braille pages that are produced. For example, one textbook in Braille might
take up several shelves on a bookcase once printed, so alternative accommodations are
preferred. Some blind students use assistive technology known as Braille notetakers. These
devices allow students to type notes and read in Braille, and they operate similarly to a
laptop computer. However, with Braille readers, Braille dots representing ASCII characters
are refreshed at the fingertips of the user, instead of appearing on a computer screen. The
student at Wake Forest University used the Braillenote R
Braille Reader (Humanware n.d.)
and the JAWS screen reader (Freedom Scientific n.d.).

2. Statistics 2: Statistical Models

The course MTH 256 Statistical Models is a 3 credit hour semester long course offered
at Wake Forest University. The pre-requisite is effectively “any previous first course in
statistics” which may include AP statistics, the introductory statistics course offered in the
Mathematics department, or a course from another department, so long as the content is
primarily in statistics. We assume no previous background with calculus or computer pro-
gramming. The course currently uses the textbook Stat2: Building Models for a World of
Data (Cannon et. al. 2013). In spring 2014 the course covered the following topics: simple
and multiple regression, along with diagnostics, testing and model selection (most of chap-
ters 0-4); categorical data and interactions (loosely based on chapters 5 and 6, but taught
as a special case of regression); and simple and multiple logistic regression (chapters 9-
11). Sighted students use R through the R Commander package (Fox 2005). Our student
worked with R in a different manner, which is described in detail in section 3.23.2. Course-
work included weekly assignments, and contained a blend of applied data analysis, report
writing, and theoretical questions. The course also had three in-class, closed-book exams.
For the final project, each student found a suitable dataset, fully analyzed it, and wrote
up the findings in a full statistical report. Most reports ranged in the 10–15 page range,
including figures. Students could work alone or with a partner on the final project.

In the remainder of this section we demonstrate the visual, computational, and mathemat-
ical techniques students need in the second statistics course. The reader is encouraged to
consider the challenges that a blind student might face.

5
 Journal of Statistics Education, Volume 23, Number 2, (2015)

2.1 Example: High Peaks Data Set

An example of a typical dataset for this course is the HighPeaks data, contained in the
Stat2Data R package (Lock 2012). These data contain measurements on the 46 “High
Peaks” mountains in the Adirondack Park of northern New York State. The High Peaks
were originally believed to be all mountains taller than 4000 feet above sea level in the
Adirondack Park, though later measurements would confirm that four are slightly below
4000 feet, and one other mountain was missed. Measurements include Elevation (feet),
Difficulty of hike (on a scale of 1–7), Ascent (vertical gain in feet from the trailhead to
the summit), Length (round trip distance in miles) and Time (average number of hours
to complete the round trip hike). We emphasize data visualization prior to model fitting.
It is obvious to the students that a longer hike or a higher vertical ascent would make
a mountain more difficult. One natural question to ask is if the difficulty is only some
combination of Length and Ascent, or if it considers other factors. Students might suggest
steepness, the presence or absence of a marked trail, and trail conditions as other possible
factors. A favorite teaching method of the instructor was to encourage speculation on other
possible variables or explanations for observed effects, then ask if it is possible to use the
data on hand to answer firmly on the issue. In this case, it is possible. By plotting Length
vs. Ascent and using different plotting symbols for Difficulty, a student can identify pairs
of mountains in which the hike with both a higher Ascent and longer Distance is classified
as easier, disproving the claim that difficulty relates only to Length and Ascent. This is
shown in Figure 1, and serves as a nice example of the power of proper displays of data
without any model fitting required.

After continuing with additional exploratory data analysis and producing histograms, box-
plots, tables, and other visual aids, students next explore what makes a hike take longer.
Students learn to fit a simple linear model,

Time = β0 + β1 · Length + ε , ε ∼ N(0, σ ), (1)

where β0 is the intercept, β1 is the slope, and the errors ε are assumed to be independent
and follow a zero-mean normal distribution with some constant standard deviation σ . The
fitted model is represented as

[ = βb0 + βb1 · Length = 2.04817 + 0.68427 · Length.

Time (2)

6
 Journal of Statistics Education, Volume 23, Number 2, (2015)

Difficulty
2 Is Difficulty Only Ascent and Length?
3
4 Disproving Pair
5
6
7
4000
Ascent (feet)
3000
2000

6 8 10 12 14 16 18

Length (miles)

Figure 1. This scatterplot can be used to disprove the claim that the variable Difficulty is
only some combination of the ascent and length of hike. The pair of mountains highlighted
shows a case where a mountain with both a larger ascent and longer hike is classified as a
less difficult hike.

The R code needed to fit this model and its associated output are both shown below. Stu-
dents learn to interpret each piece of this output, and scan for the elements they need for
specific tests.

> library(Stat2Data)
> data(HighPeaks)
> attach(HighPeaks)
> lm1=lm(Time~Length)
> summary(lm1)

Call:
lm(formula = Time ~ Length)

Residuals:
Min 1Q Median 3Q Max
-2.4491 -0.6687 -0.0122 0.5590 4.0034

7
 Journal of Statistics Education, Volume 23, Number 2, (2015)

18

4
Residuals (hours)
3
Time (hours)
14

2
1
10

0
8

-2
6

6 8 10 12 14 16 18 6 8 10 12 14

Length (miles) Fitted Values (hours)

Figure 2. Scatterplot and residual plot for Time (hours) vs. Length (miles). Students will
use these figures to assess assumptions of linearity and constant variance.

Coefficients:
Estimate Std. Error t value Pr(>|t|)
(Intercept) 2.04817 0.80371 2.548 0.0144 *
Length 0.68427 0.06162 11.105 2.39e-14 ***
---
Signif. codes: 0 ’***’ 0.001 ’**’ 0.01 ’*’ 0.05 ’.’ 0.1 ’ ’ 1

Residual standard error: 1.449 on 44 degrees of freedom

Multiple R-squared: 0.737, Adjusted R-squared: 0.7311
F-statistic: 123.3 on 1 and 44 DF, p-value: 2.39e-14

Figure 2 shows a scatterplot of Time vs. Length, along with a best-fit line and residual plot.
Important lessons from this figure include being able to identify signs of nonlinearity, non-
constant variance, and possible outliers or influential points. These figures might remain
on the projector screen in class for several minutes as students learn to look for important
features. Figure 3 shows a QQ plot and histogram of the residuals, both of which are used
to assess the assumption of normality of errors. The QQ plot here shows some evidence of
non-normality in the tails.

As more complicated models are fit with differing numbers of explanatory variables, stu-

8
 Journal of Statistics Education, Volume 23, Number 2, (2015)

Normal Q-Q Plot

10 12 14
4
Sample Quantiles
3

Frequency
2

8
1

6
0

4
-1

2
-2

0
-2 -1 0 1 2 -2 0 2 4

Theoretical Quantiles Residuals

Figure 3. Quantile comparison and histogram of residuals from Equation 2.

dents turn to model selection criteria such as Mallow’s Cp, computed as

SSEm
Cp = + 2(m + 1) − n, (3)
SSEk /(n − k − 1)

where m is the number of predictors in the model under consideration, k is the number of
possible predictors in the full model, SSEm and SSEk are the sum of squared errors from a
model with only m and k predictors, respectively, and n is the number of data points.

This example demonstrates what we define as three sets of materials essential to the course:
mathematical writing, computer programming, and visual displays. Equations (1)–(3) show
just a bit of the mathematical complexity required; Figures (1)–(3) show examples of im-
ages from which we extract information about data or models; and the R commands and
output give a flavor for the computer programming needed. For each of these, students
must master both interpretation and production. We found it useful to organize these six
skills into a table, shown in Table 1. Assisting the student in the course necessarily meant
enabling her to master each of these six skills. The next section will discuss the specific
technologies and approaches used to help her with each skill.

9
 Journal of Statistics Education, Volume 23, Number 2, (2015)

Table 1. The six required skills a student must master in Statistical Models, labeled A through F.
Interpreting Producing
Mathematical Writing A B
Computer Code C D
Images and Graphical Displays E F

3. Assistive Technologies Employed

3.1 Reading and Writing Mathematics

Grade 2 English Braille involves six dots, placed in two columns of three, which are either
raised or flattened. Ignoring the blank space, there are 26 − 1 = 63 unique characters which
can be produced with only six dots. With this limitation, a number of conventions have
been developed to give this 6-bit system the greatest reach. For example: the numbers 1
through 9 are identical to the letters a through i, and 0 is also j; a special single character
dictates whether or not a number or letter is intended; a single character indicates whether a
letter is capital or lower case; ligatures such as th, sh, and ch are represented as single char-
acters; and an elaborate system of contractions and short-hand has been developed, with
afn representing “afternoon”, for example. The ASCII symbols 6-bit Braille can replicate
are known as Braille ASCII (http://www.dotlessbraille.org/displays.htm).

It should be immediately obvious that it is a struggle to use 6-bit Braille to write higher
mathematics. Merely think of all of the symbols from differential calculus, linear algebra,
logic, and so forth, and imagine trying to write all of these with a mere 63 characters.
Thus, alternative forms of Braille have been developed for special purposes. In mathemat-
ics, Nemeth Braille (Nemeth 1972) was first developed in 1952 by Abraham Nemeth to
better represent mathematical equations and writing. Physicist John Gardner and Math-
ematician Norberto Salinas developed Gardner-Salinas Braille (Jackson 2002), an 8-bit
Braille which extended Grade 2 English Braille to better capture symbols commonly used
in mathematics, such as operators, Greek letters, and various punctuation marks.

An alternative is to use LATEX for writing mathematical and statistical material. LATEX is
a markup language used in the statistics community for typesetting statistical writing. All
mathematical writing can be reduced to ASCII characters used to represent various mathe-
matical symbols. For those less familiar with LATEX, this manuscript was written in LATEX.
Equations (1)–(3) in this paper were typed as:

Time = \beta_0 + \beta_1 \cdot Length + \epsilon,

\hspace{4mm} \epsilon \sim N(0, \sigma)

10
 Journal of Statistics Education, Volume 23, Number 2, (2015)

\widehat{Time} = \widehat{\beta}_0 + \widehat{\beta}_1 \cdot Length

= 2.04817 + 0.68427 \cdot Length

C_p = \frac{SSE_m}{SSE_k/(n-k-1)} + 2(m+1) - n

The crucial value of LATEX for the blind student is that it can represent all information
of a mathematical expression purely in terms of ASCII characters, which can themselves
be easily read by either screen readers or Braille displays. Since the student would very
likely continue to take mathematics courses, we decided the time investment to learn LATEX
was worthwhile. Most mathematics faculty at Wake Forest University are already fluent in
LATEX, and can provide materials relatively easily. Due to the mathematical symbols and
equations required, many textbooks in mathematics and statistics are often first written in
LATEX, and we suspected it might be easier to get accessible materials from publishers. Fi-
nally, since the student could type on a regular keyboard, typing LATEX source code would
be a viable means of her writing her own mathematics and statistics. LATEX is simply a con-
vention for typing mathematical symbols using only ASCII characters, it is not a software
program itself. Various proprietary compilers and editors have been developed, but these
are not expensive. For example, the widely used WinEdt (http://www.winedt.com/) has a
single educational license for $60, and 5 educational licenses can be purchased for only
$200.

The student found it easiest to open LATEX files both on her laptop and also in her Brail-
lenote refreshable display. She would listen to the code being read by JAWS on the laptop,
while feeling the Braille version of these equations on her Braillenote. It wasn’t until after
the course ended that the student discovered the possibility of syncing her screen reader
JAWS and the Braillenote so that she could simultaneously hear and feel the LATEX code.
She described that screen readers are designed to read either word-by-word, or character-
by-character. A document written in LATEX is best read by combining these two approaches.
For example, paragraphs of text are read word by word; equations are read as a combina-
tion of words and characters. Syncing (through bluetooth) the screen reader and Braillenote
would help a student learn LATEX.

When it came time for her to type mathematics for assignments or exams, the student would
simply type the code into a .txt file on her laptop. As it was not possible for the student to
compile the code to see if it was all correct, very minor coding errors such as missing dollar
signs, unclosed braces, missing slashes, or an incorrect use of the % symbol would need
to be quickly corrected before her work could be graded. Over time, the faculty member
stopped spending time to correct minor LATEX mistakes and simply graded the LATEX source
code without compiling. This focus on the LATEX source code (and not the compiled text)

11
 Journal of Statistics Education, Volume 23, Number 2, (2015)

is an important consideration when teaching blind students. Consider the formatting of the
“four steps” of modeling, as written in the notation common in the Cannon et. al. (2013)
book:

1. Choose the model Y = β0 + β1 · X + ε , ε ∼ N(0, σ ).

2. Fit the model to obtain Yb = βb0 + βb1 · X.

3. Assess the model.

4. Use the model to predict at X = 100.

The precise code used to write the above steps is:

\begin{enumerate}
\item \textbf{Choose} the model $Y = \beta_0 + \beta_1 \cdot X + \epsilon,
\hspace{4mm} \epsilon \sim N(0, \sigma).$
\item \textbf{Fit} the model to obtain $\widehat{Y} = \widehat{\beta}_0
+ \widehat{\beta}_1 \cdot X$.
\item \textbf{Assess} the model.
\item \textbf{Use} the model to predict at $X = 100$.
\end{enumerate}

The compiled, formatted writing looks quite elegant. The enumerate environment kept
track of counting for us and we never needed to write out step numbers, the widehat
looks a bit prettier than the smaller hat, the hspace added a bit of blank space in line
(1) between the ε and its distribution, and so forth. However, the blind student reading the
source code directly would find some of this unnecessarily distracting. A more efficient
means of conveying the same mathematical information in LATEX would be

1. Choose the model $Y = \beta_0 + \beta_1 \cdot X + \epsilon,

\epsilon \sim N(0, \sigma).$
2. Fit the model to obtain $\hat{Y} = \hat{\beta}_0 + \hat{\beta}_1 \cdot X$.
3. Assess the model.
4. Use the model to predict at $X = 100$.

In particular, the changes are: the enumerate environment has been scrapped in favor of
writing “1,” “2,” . . . instead; all instances of widehat have been replaced by the simpler
hat; the command hspace was removed; bold was removed, and the double slash used to
indicate a line break was also removed. If this simpler code is compiled, the result looks

12
 Journal of Statistics Education, Volume 23, Number 2, (2015)

quite a bit worse to a sighted user, as shown below:

1. Choose the model Y = β0 + β1 · X + ε , ε ∼ N(0, σ ). 2. Fit the model to obtain Ŷ =

β̂0 + β̂1 · X. 3. Assess the model. 4. Use the model to predict at X = 100.

However, the scaled back source code shown above is clearly preferable for someone read-
ing the code directly. Thus, items A and B on Table 1 were handled through the use of
LaTeX, and a quick summary of these accommodations is shown in Table 2. Next, we
move on to items C and D.

Table 2. A summary of the major accommodations made for mathematical reading and writing, and re-
maining challenges

Challenge Accommodation and Assistive Technology Remaining Challenges

We obtained a high quality, searchable pdf file
Obtaining an Some organization of multiple files
from the publisher, and a graduate student copied
accessible was required. Images were stored
this text into a text editor and re-typed equations
textbook separately.
in LATEX.
The instructor would re-write some notes in Additional merging of files was
Course notes “minimal” LATEX and send to the student in required if the student wished to
advance. add notes during class time.
A more timid student who is blind
The instructor practiced a clear, verbal
In-class notes might not wish to repeatedly ask
description of all equations and images displayed.
for clarification.
Submitted LATEX files often did not
compile (minor errors, missing $
Writing The student typed all work in LATEX on a regular
symbols, etc.) so the instructor
mathematics keyboard.
either fixed and compiled, or
graded the LATEX source directly.

3.2 Computer Programming

In “Statistical Software from a Blind Person’s Perspective,” Godfrey (2013) discusses in

great detail how blind users interact with different statistical software programs. The pa-
per’s subtitle says “R is the best, but we can make it better,” a statement we agree with
in full. R can deal exclusively in text for data input, analysis, and output; it requires no
graphical interface; help pages are written as text files; and functions and commands can
be modified to suit the needs of particular users. In these respects, R offers the blind user
many advantages. However, two essential modifications for the blind user are described

13
 Journal of Statistics Education, Volume 23, Number 2, (2015)

by Godfrey in clear detail. These are 1. how to run R in a terminal window, and 2. how
to utilize the package BrailleR (Godfrey 2012) to get full compatibility with the screen
reader.

We developed this approach using R version 2.15.1. Most students using R early in their
statistics careers will run the program inside a window, with several nested windows in-
side. One is the R console window which shows directly typed commands and R output.
Another shows the script, and additional windows show images. These nested windows
interfere with the ordinary use of the screen reader, and thus navigating was not possible
for the blind user. To our knowledge, there is no easy way to toggle between these nested
windows in a manner that allows the screen reader to work properly. Following Godfrey
(2012), we instead set up R to run in the terminal window. By running R in the terminal
window, it ensures that the screen reader will stay focused on the window with both input
and output. By spending a bit of time creating a desktop shortcut and modifying the PATH
environment variable, it is possible for the student who is blind to open a working R ses-
sion with minimal effort, and be totally independent with computer input and output. The
specific steps to achieve this for the Windows 7 operating system are as follows:

1. On the desktop, create a new shortcut to the Command Line terminal cmd.exe. This
is the link the blind user will open and simply type the letter “R” and hit enter to
open up a session. However, this requires that one modify the PATH environment
variable.

2. Modify the Path variable:

(a) Go to Control Panel, System, Advanced System Settings,Environment Variables

(b) Find Path under System Variables, and click Edit
(c) Add the directory containing R.exe to the Path variable. On the author’s com-
puter with Windows 7, the directory was

C:\Program Files\R\R-2.15.1\bin\i386;

3. Open cmd.exe on the desktop, and simply type R and press enter. R should open.

The end result of this is that one can find the cmd.exe icon on the desktop, open it, and
simply type the letter R to run R in the terminal window. Packages can be installed in
the terminal by typing install.packages("BrailleR"), for example. After waiting a
few seconds, the blind user can simply press the ENTER key which will select the de-
fault “0-Cloud” as the mirror for the installation; only in rare cases would a package be

14
 Journal of Statistics Education, Volume 23, Number 2, (2015)

unavailable from this mirror. Unless there is an error of some sort, the rest will happen au-
tomatically. Once a package has been installed correctly, on subsequent sessions the user
can simply type library(BrailleR) to load it. The working directory is identified by
typing getwd().

The second essential modification is to create a file which captures all user keystrokes
along with R output which can be easily read with a screen reader. Here we will describe
what happens in a trivial example of using R, in which the user only wants to add 1 and 2:

> 1+2
[1] 3

Once the user has typed “1+2” the screen reader recited this entry once, and then recites
the output line “[1] 3” exactly once. The screen reader will read the bracketed [1], which
is unfortunate but not too bothersome after a while. However, suppose that the user needed
to hear things again. . . there is no way to get the screen reader to repeat what has been run.
Pressing the up arrow will bring back previously typed commands which are read by the
screen reader, but there is no simple way to have the reader re-read the R output in response
to those commands. Hence, it is useful to have a file which captures all input commands
and output results into a log, and the user can repeatedly refer to this log. This file can be
created by typing

library(BrailleR)
txtStart(file="output.txt").

This creates a file called “output.txt’ inside the working directory, and this file will capture
all input and output commands. This file is best opened in a web browser such as Internet
Explorer (we used version 9.0). By pressing Alt-Tab, the user can toggle between the ter-
minal running R and the file output.txt. When in Internet Explorer, the user can press
F5 and the browser will refresh the file, showing all typed commands and R output in re-
sponse. This file works quite well with screen readers, allowing the student to listen line by
line or character by character. This file also serves as a useful learning tool for program-
ming. Indeed, sighted students in the class expressed an interest in this file as it serves as a
log of all successful (and unsuccessful!) commands typed, archiving all output.

The student commented that the BraillerR package was one of the most important assis-
tive technologies in the course. The VI() command (described in the next section) allowed
her to interpret all histograms and assess normality independently; the alternative would

15
 Journal of Statistics Education, Volume 23, Number 2, (2015)

Table 3. A summary of the major accommodations made for computer programming and interpretation,
and remaining challenges
Accommodation and Assistive
Challenge Remaining Challenges
Technology
Getting R to This change would need to be
We set up R to run in the terminal window
work with the made to every computer the student
by modifying the path variable.
screen reader uses.
Losing the cursor
in the terminal
Press the Alt key. None.
window in
Windows 7
The student used the BrailleR package
Recording and the txtStart() command to make a The log file records everything, so
submitted code log. This file was opened in an internet it can become quite long with
and output browser which allowed for refreshing the accumulated errors.
file.
Tabulated output (from the
The student used the VI() command summary() of a linear model, for
Interpreting R
when applicable, but otherwise simply example) must be listened to row
output
practiced listening for relevant output. by row instead of column by
column.

have been to print all histograms as tactile images, which is more time consuming and re-
quires an assistant. The txtStart() command essentially made learning R programming
accessible. Most students refine programming skills through trial and error, and the log file
saving all commands and output allowed her to do this as well.

Two final comments are needed. First, there is a bizarre tendency for the cursor to get
“lost” when toggling between the terminal window and Internet Explorer. This is a well-
documented issue in which the keyboard becomes unresponsive when toggling back to the
terminal window from a different window. Fortunately, quite by accident Godfrey and Er-
hardt (2014) discovered a solution; pressing the Alt key after toggling back to the terminal
window remedies the lost cursor. Second, we did not consider alternative statistical soft-
ware programs and so cannot comment on how well they would have performed for our
needs. Our university teaches exclusively in R, and given the available packages to make
R more accessible to blind students, we simply chose this option.

Table 3 shows a summary of these accommodations for computer programming.

16
 Journal of Statistics Education, Volume 23, Number 2, (2015)

3.3 Producing and Interpreting Images

The question we asked ourselves for each image in class was this: How does a visual
representation function for students who are sighted, and what would serve as a comparable
function for a student who is blind? Sometimes, an image is not at all needed; it is only a
convenient way for sighted people to store and present the information. Within this spirit,
the BrailleR package (Godfrey 2012) contains a useful command VI() which can wrap
other commands to make the R output more accessible to the blind user. Consider the
histogram shown on the right panel of Figure 3. If the R command hist is placed inside
the wrapper function VI(), the output is:

>VI(hist(lm1$resid))
This is a histogram, with lm1$resid marked on the x-axis, unless you explicitly
used the xlab argument.
There are a total of 46 elements for this variable.
It has 8 bins with equal widths, starting at -3 and ending at 5.
The mids and counts for the bins are...
mid = -2.5 count = 4
mid = -1.5 count = 6
mid = -0.5 count = 14
mid = 0.5 count = 13
mid = 1.5 count = 5
mid = 2.5 count = 1
mid = 3.5 count = 2
mid = 4.5 count = 1

Thus, the command has extracted the needed information from the histogram, and de-
scribed it in a format which is accessible to the blind user. The student made extensive use
of this command when checking for normality of variables or residuals. A histogram may
not be as valuable for detecting non-normality as a quantile quantile plot such as the one
shown in Figure 3, but it is often enough. Information lost in the translation from a QQ plot
to a histogram include things like: the values of the theoretical quantiles at which the de-
parture from normality begins (seen as where the points depart from the line on a QQ plot);
a sense of whether a few points are merely outliers, or truly indicative of a non-normal tail;
and so forth. Given that histograms could be made fully accessible, whereas something
like a QQ plot poses much more of a challenge, we felt it reasonable to allow the student to
solely use histograms to assess normality of errors for homework and projects. If the VI()
command is used with a function for which it has no special modification, it will simply
“print” the object, as shown here:

> VI(boxplot(Elevation))

17
 Journal of Statistics Education, Volume 23, Number 2, (2015)

No VI method has yet been written for this type of object

so it has been printed for you in its entirety.
$stats
[,1]
[1,] 3820.0
[2,] 4100.0
[3,] 4380.5
[4,] 4627.0
[5,] 5344.0
attr(,"class")
...

This is still quite useful, once the user has learned the structure of these R objects. For
the boxplot, the five numbers are always the lower whisker, first quartile, median, third
quartile, and upper whisker.

For images which cannot be easily converted or described in text, we purchased a Desk-
top Braille Embosser from Viewplus (http://www.viewplus.com/products/braille-printers/
desktop-braille-printers/). An embosser prints dots in high durability paper. It can be used
to print straight text which is simply Braille, or images which are known as tactile images
in the blind community. At the time of purchase the embosser cost $4,995, and was by far
the greatest expense associated with accommodating the student. Our justification was that
the student was a first year student at Wake Forest University, interested in majoring in a
quantitative field, and likely to take numerous additional courses with images, graphs, and
other visual information.

Figure 4 shows an example of a tactile image (it is the same scatterplot as in Figure 2).
Darker dots are taller, and lighter dots are shorter, so the student can feel around 5 differ-
ent “shades.” All images in our course were first created in R. The only modifications at
this stage are to ensure that any plotting symbols or lines are heavy enough to print well.
Plotting symbols should be solid (pch=15 produces solid squares, which work well with
the Braille Embosser) and large (cex=1.4 works well). Lines should have a line width
of lwd=2 at least. Colors, different plotting symbols, different line types (dashed, dotted,
solid) and figure legends did not print well on the Embosser and we made no use of them.
Returning to Figure 1, it is not possible to convert this to a tactile image on a regularly
sized sheet of paper. This message would need to be conveyed in some other fashion.

Once an image has been produced, it is copied as a bitmap. The Tiger Software Suite (http://
www.viewplus.com/products/software/braille-translator/) that comes with the Braille Em-

18
 Journal of Statistics Education, Volume 23, Number 2, (2015)

Figure 4. Scatterplot of Time vs. Length of the 46 High Peaks when pasted into Tiger
Designer. Plotting symbols should be dark, large, but overlapping to a minimal degree.
The line must be wide enough to be felt. It is also important that the entire plotting area
be boxed, so the student can understand the range of possible points. Braille labels take up
much more space than text (think of writing in size 19 font), so space is at a premium. Only
every other tick mark on the axes was labeled, and the axis labels appear much larger.

bosser had a program called Tiger Designer, which facilitated the conversion of an image.
The simplest way we found to use it was to paste the bitmap from R, adjust the contrast
as needed, and retype over all axis labels using Braille. The software allows one to type
in English, but Braille is displayed. These were then printed on Braille paper, three hole
punched, and stored in a three ring binder. Over the course of the semester, the number of
images grew quite rapidly so organization became more of a challenge. We found it useful
to write information explaining each image to a sighted person (titles, page numbers, etc.)
on the back of each, to remedy the inevitable—when the binder rings fail to close, and all
pages fall to the floor.

19
 Journal of Statistics Education, Volume 23, Number 2, (2015)

We should also mention another more affordable solution known as a Picture In a Flash, or
PIAF (Humanware n.d.). This prints an image using special ink on paper, and the sheet is
then heated which causes the ink to blister, and physically deform the paper. We originally
thought a PIAF would work best as it can print a straight line, but found that the Braille
embosser offers greater flexibility, and longer lasting images.

What has been described above only deals with item E on Table 1. It may come as a sur-
prise to learn that Braille embossers are not really designed to be operated independently
by the blind user. So, images in lectures or notes could be prepared by the faculty member
in advance and provided to the student, but the greatest challenge with images were those
situations where the student needed to produce an image. Think of the process of choos-
ing an appropriate model for data. Sighted students might make several dozen images of
scatterplots, residual plots, and so forth when deciding how to choose a best model for a
given dataset. Each image might be viewed for only a few seconds. The student could not
use the Braille embosser independently, and thus could not produce these images herself.
For homework assignments and projects where producing and interpreting intermediate
images was essential to the task at hand, the student either worked directly with the faculty
member or with a graduate student trained in making tactile images.

Table 4 shows a summary of these accommodations for images.

Table 4. A summary of the major accommodations made for images, and remaining challenges

Accommodation and Assistive

Challenge Remaining Challenges
Technology
The VI() command in the BrailleR
Histograms (for There are minor issues (bin sizes)
package converts these to a more
residuals) that are overcome with practice.
accessible format.
The instructor developed a convention for One-on-one meetings with the
Images which can be
describing those images used in class for student were needed to confirm
easily described
30 seconds or less. that this convention was effective.
The number of images posed some
We printed tactile images on a Braille storage and access issues as the
Images which cannot
embosser, labeled them in both Braille and semester progressed. Also,
be easily described
English, and stored them in binders. essential images for class needed to
be prepared in advance.

20
 Journal of Statistics Education, Volume 23, Number 2, (2015)

4. Other Changes Required

The textbook for the course was not available in any accessible format. We are deeply
grateful to W. H. Freeman for providing us with a very high quality, searchable PDF ver-
sion of the entire Stat2 textbook, though obtaining this was not a simple matter. When we
contacted W. H. Freeman asking for a digital version of the text, we were initially told they
could only provide a free hardcopy of the textbook or free access to an e-copy. We pressed,
insisting that neither were accessible to the student. W. H. Freeman eventually agreed to
send a high quality, searchable PDF version of the text, but their reluctance is quite under-
standable. One could imagine such a file being posted online, which would greatly impact
book sales. However, trust must be maintained between all those involved in education for
the blind.

The searchable function of this file was essential. What we mean by searchable is that the
PDF is not merely an image of text, but the text itself can be highlighted and copied. This
was necessary to convert the PDF into an accessible format. In principle, the program In-
ftyReader (http://www.inftyreader.org/) can be used to convert PDF back into LATEX source
code, and InftyReader has been successfully utilized to enhance accessibility in mathemat-
ics for the blind (Yamaguchi et. al. 2008). However, in our particular course we found
that the formatting of the textbook PDF file prevented InftyReader from working well. We
therefore handled the conversion ourselves. For all text, we were able to highlight and copy
the text directly, and paste it into an ordinary text file. Equations were re-typed in LATEX
source code. Tables were re-typed using the tabular environment in LATEX. In place of
images, we would write “(Image 1.1 goes here: Scatterplot of Time vs. Length for the 46
Adirondack High Peaks)” into the LATEX file, then re-make the image ourselves using the
Braille embosser. A graduate student in Mathematics at Wake Forest University worked on
this chapter-by-chapter and would send the completed files and tactile images throughout
the semester. The faculty member wrote out all classroom notes in LATEX in advance and
e-mailed them to the student. The student was able to take modest notes during class using
her Braillenote, though she primarily just listened during class.

During class, all notes written on the white board were simply dictated by the faculty
member when written. For example, when writing the equation for Mallow’s Cp (shown
earlier as Equation (3)), the faculty member would say “Cp equals SSEm divided by the
quantity SSEk over n-minus-k-minus-1, and then to this entire quantity is added 2-times-m-
plus-one, minus n.” On rare occasions, the student would ask for a brief clarification (“is
SSEk over n-minus-k-minus-1 all in the denominator?”), but this happened very rarely,
possibly because all class notes were provided in advance so the student had it written
in LATEX correctly already. Sighted students did not seem bothered by this dictation, and
indeed a few seemed to prefer it since it helped remove confusion with symbols, notation,

21
 Journal of Statistics Education, Volume 23, Number 2, (2015)

and faculty penmanship.

Images that were essential to a lecture—involving, say, at least 30 seconds of class time—
were all printed by the faculty member in advance using the Braille embosser. When the
relevant image was displayed on the projector, the faculty member would simply direct the
student to the correct tactile image being discussed. We found that many images in Stat2
are displayed and discussed for very little time. For these, we would simply describe them,
or ask sighted students to describe them. This turned out to have unexpected educational
benefits for the sighted students as well. For example, the residual plot shown on the right
in Figure 2 might be displayed, and the professor would ask “are the assumptions of simple
linear regression met here?” One sighted student might say “No, the variance is not con-
stant” to be quickly rebutted by another sighted student saying that “yes, it’s close enough.”
The faculty member would ask each to elaborate. This begins by stating what type of plot
is being shown, and what the two axes represent. The first student might describe a general
funnel shape extending to the right, indicating the residuals were larger for larger fitted
values; the second might simply point out that it’s only perhaps 3 data points in the up-
per right corner giving this illusion, and we shouldn’t over-interpret outliers. In any event,
these verbal descriptions were originally used to benefit the student who is blind, but in fact
they helped sighted students see different features of these images. They also forced all stu-
dents to spend a moment describing an image before they jumped into drawing conclusions
about it—surely this is a habit we could all stand to benefit from.

The faculty member learned to avoid the use of ambiguous pronouns such as this, these,
it, and so forth, and instead refer to all objects by their proper names. At first, the faculty
member also deliberately put his hands in his pockets to avoid the temptation to gesture
or point during class discussions. Hand motions and gesturing are of course a positive
teaching technique for sighted students, but the danger is that they can be used to substitute
for verbal descriptions, and therefore the blind student could not follow the lecture. Over
the course of the semester, these negative habits abated.

A final piece of the accommodations was a standing, weekly one-on-one meeting between
the faculty member and the student. These meetings allowed both to clear up any errors,
misunderstandings, or issues that arose in the previous week, and set clear goals for the
upcoming week. Sometimes these meetings would last only 30 minutes, and deal with a
few homework questions. Other times these meetings used substantial time for working
with R, teaching LATEXcommands for new material, revisiting class notes, or just generally
providing support and advice to the student in all her academic pursuits. An honest and
open relationship—in which the student feels free to express any and all frustrations or
questions about the course—is absolutely essential to success in the classroom.

22
 Journal of Statistics Education, Volume 23, Number 2, (2015)

5. Discussion

Despite postponing her first statistics course until the spring semester, allowing the univer-
sity more time to determine how to best provide accessible math and statistics coursework,
the staff at Wake Forest University were still overwhelmed with the number of potential
and novel solutions. It had been much easier to provide accessible material to courses in the
Humanities and even some of the Sciences, where the coursework was more text-rich or
the text in alternative format already existed. Our efforts at researching best practices pro-
vided some helpful starting points, but we were never certain that our proposed approach
using varied assistive technologies would be appropriate.

Providing the technology and the alternative course material for this statistics course re-
quired the efforts of several faculty, staff, and students. In addition to the time and energy
both the faculty member and the LAC staff member spent researching and proposing al-
ternative methods for presenting the course information, many hours were spent by the
mathematics graduate student: (a) retyping all of the equations into LATEX; (b) printing all
needed images using the embosser; and (c) collating all of the classroom examples to pro-
vide a course packet of examples for the student. Furthermore, the graduate student and
faculty member were required to become familiar with Braille and with the assistive tech-
nology essential to providing the alternative text and using the Braille embosser. It would
have been valuable to have a full-time staff member at the university who is dedicated to
making course material accessible and who has the knowledge and skills to provide ongo-
ing support for assistive technology. However, we did not have such a staff member, and
were extremely grateful to our graduate student for filling this need.

Even though the student will likely benefit in her academic future and her anticipated vo-
cation from having learned LATEX, she faced a steep learning curve and had to become
fluent in a short period of time. Her incoming knowledge of assistive technology was ex-
tremely helpful, and it is important to note that knowledge and experience will vary among
students, depending on the degree to which they used assistive technology in their primary
and secondary education. Here we were fortunate that the student was well versed in screen
readers, Braille notetakers, and the use of a regular keyboard for typing. This allowed us
to focus heavily on learning LATEX and R right away. The student recommends that others
who are blind and wish to study statistics should learn LATEX as soon as possible, preferably
even before the second statistics course. This would allow them to focus more on learning
the material.

Due to the nature of how we provided access to this course, the student did not actually
register for the same course number as the other students enrolled in the course. Instead, the
faculty member, in consultation with the department, chose to follow an independent study

23
 Journal of Statistics Education, Volume 23, Number 2, (2015)

model. In addition to attending the class meetings and completing the class requirements
for the original course, the student and faculty member supplemented instruction through
individual meetings. At the end of the semester, she met (if not exceeded) the same learning
objectives as the other students enrolled in the actual course, and she is prepared to advance
through her major at Wake Forest University. However, the amount of time the faculty
member spent researching and meeting individually with the student is more accurately
reflected as an independent study.

The faculty member who taught this course is currently the only one in the department
with the knowledge and experience to provide an accessible experience for courses that
rely heavily on the visual representation of data. Ideally, his knowledge and experience
can be easily transferable to other mathematics, statistics, or business courses, because this
student intends to complete a major in one of these areas. We have found that the upfront
and ongoing costs of providing these accommodations are likely to be highly variable from
university to university. Much of the cost estimates would depend on the particular student,
background of the instructor, depth and resources of the Learning Assistance Center at the
university, background of student assistants, and so forth.

Providing an accessible course in this fashion highlighted the importance of collaboration

between faculty and student services on college campuses. Faculty members should be
encouraged to consult and problem solve with other campus resources, particularly the
Disability Support Services offices. The responsibility of providing accessible material is
shared among the faculty, student, and staff members at the institution.

It has already been noted that teaching a statistics course to a student who is blind can
enhance not only the instructors pedagogical approach, but the student-instructor relation-
ship, too. The individual attention required at this level of instruction provided helpful
guidance and feedback for both the instructor and the student. It seems unlikely that a tra-
ditional classroom environment, even with a high level of participation and engagement,
could achieve the same level of understanding and learning. We recommend to mathemat-
ics and statistics faculty that they find ways to work with their administrators to protect the
weekly one-on-one hours which are surely necessary.

We should mention that in situations where the student is completing graded work with
an assistant, there is a danger that the “assistance” will grow beyond what it should be
(Marson et. al. (2013) call this countertransference). This can be difficult to avoid, but
must be avoided if the student is to have a fair and equitable learning environment. In
addition, research has shown that students feel resentment towards teachers who give too
much assistance (Landau et. al. 2003). For graded exams, the Learning Assistance Cen-
ter administered all graded exams for the student, overseen by a trained specialist rather

24
 Journal of Statistics Education, Volume 23, Number 2, (2015)

than the faculty member who would grade the exam. The student was asked to sign an
exam contract with the Learning Assistance Center. This contract reminds the student of
the Honor Code and the list of impermissible resources. Homework problems posed less
of a concern, because all students in the class are permitted to discuss problems with one
another so long as each student fully writes and hands in his or her own solutions. Many
of the conversations between the blind student and her assistant would mirror those among
sighted students, and thus posed no real problem. As for interactions with the faculty mem-
ber regarding grading (discussing how partial credit was assigned, for example) the faculty
member simply had to listen carefully and use professional judgement.

Another point that warrants discussion is the relationship between what we did in this sec-
ond course and what experience the student had in the first course. Returning to the table
of six challenges (shown in Table 1), we consider them for a hypothetical student who is
blind. For items A and C, we assume no previous background in either case. We assume
only that the student has a means of reading LATEX and computer programming output (a
screen reader, refreshable Braille display, etc). However, prior exposure to LATEX in the
intro course would have been valuable. Item A further requires some basic familiarity with
algebra and commonly used equations in statistics (mean, standard deviation, standard-
izing a random variable, manipulating square root, square, logarithm, etc.). For items B
and D, we do assume that the student is able to use a regular keyboard for typing LATEX
and computer code. However, there is nothing from the introductory course that would be
needed in this regard. For item E, we assume that in the student’s first course they have
learned the basics of a scatterplot, fitting a straight line to a scatterplot, histograms, the
normal distribution, etc. It is extremely difficult for a student who is blind to master item
F unassisted, so we make no assumptions on the background obtained in the introductory
course.

The student at Wake Forest University had done well in the introductory AP statistics
course, and seemed to have roughly the same background skills and knowledge as any
other sighted student taking a second statistics course. She felt that her preparation was
excellent coming into the second course. Since she was planning to major in a quantitative
business field, the investment of time in LATEX was worthwhile. We suspect that there may
be situations in which a student who isn’t planning to take a large number of math and
statistics courses may hesitate to make such a time investment learning LATEX; however
we still recommend its use. The student also remarked that the greatest remaining chal-
lenge was time. Reading LATEX source code character by character, getting R code to work
properly with the txtStart(), and working with an assistant on homework and projects
all take time. The student also would sometimes first write her answers in Braille (using a
shorthand convention that she understood), and then translate this into LaTeX for submis-
sion. This additional step adds time. Instructors must be aware of the inherent additional

25
 Journal of Statistics Education, Volume 23, Number 2, (2015)

time demands placed on the student who is blind.

We firmly hope that this description of our experience will help promote the discipline
of statistics to the visually impaired and blind, and that other faculty will find a frame-
work to build upon. There is no doubt that additional technologies and approaches can
be incorporated to enhance the learning experience. Indeed, one of the greatest sources of
satisfaction for us at Wake Forest University is that the student has registered for the next
courses in sequence, and is proceeding with her chosen major. We would gladly discuss
our work in further detail to any interested person, and hope to continue to be of service to
the community of blind and visually impaired students, and their universities.

6. Acknowledgments

The authors thank Dr. Jonathan Godfrey at Massey University for his expertise in teaching
statistics to the blind. The authors also wish to thank Dr. Van D. Westervelt at the Learn-
ing Assistance Center at Wake Forest University for his enthusiasm and support. Finally,
the authors wish to thank Spencer Ashley, Sara Reinke and Rebecca Kotsonis for their
assistance in LaTeX programming and tactile image preparation.

REFERENCES
Americans with Disabilities Act of 1990, Pub. L. No. 101-336, §2, 104 Stat 328 (1991).

Americans with Disabilities Act Amendment Act Pub. L. No. 110-336, §3406 (2008).

Cannon, A., Cobb, G., Hartlaub, B., Legler, J., Lock, R., Moore, T., Rossman, A., and Wit-
mer, J. (2013), Stat2: Building Models for a World of Data, New York: W.H. Freeman.

Fox, J. (2005), “Getting Started with the R Commander: A Basic-Statistics Graphical User
Interface to R,” Journal of Statistical Software, 14(9), 1–42.

Freedom Scientific (n.d.), retrieved August 19, 2014 from

http://www.freedomscientific.com/Products/Blindness/Jaws

Gardner, J. A., Lundquist, R., and Sahyun, S. (1998, March), TRIANGLE: a tri-modal ac-
cess program for reading, writing and doing math. In Proceedings of the CSUN Interna-
tional Conference on Technology and Persons with Disabilities, Los Angles. Converging
Technologies for Improving Human Performance (pre-publication on-line version) (Vol.
123).

26
 Journal of Statistics Education, Volume 23, Number 2, (2015)

Gibson, W. E. and Darron, C. (1999), “Teaching Statistics to a Student Who is Blind,”

Teaching of Psychology, 26(2), 130–131.

Godfrey, A. J. R. (2013), “Statistical Software from a Blind Person’s Perspective,” R Jour-

nal, 5(1), 73–79.

Godfrey, A. J. R., and Erhardt R. J. (2014), “Addendum to Statistical Software from a Blind
Person’s Perspective,” R Journal, 6(1), 182.

Godfrey, A. J. R. (2012), “BrailleR: Improved Access for Blind UseRs,” R package version
0.4. [p77]

Humanware (n.d.), retrieved August 19, 2014, from

http://www.humanware.com/en-usa/products/blindness/braillenotes

Jackson, A. (2002), “The World of Blind Mathematicians,” Notices of the American Math-
ematical Society, 49(10), 1246–1251.

Landau, S., Russell, M., Gourgey, K., Erin, J. N., and Cowan, J. (2003), “Use of the Talking
Tactile Tablet in Mathematics Testing,” Journal of Visual Impairment and Blindness,
97(2), 85–96.

Lock, R. (2012), Stat2Data. R package version 1.6.

Marson, S. M., Harrington, C. F., and Walls, A. (2013), “Teaching Introductory Statistics
to Blind Students,” Teaching Statistics, 35(1), 21–25.

Mechan, A., Hoffert, D., and Hoffert, L. (1993), “Strategies and Resources for Teaching
Statistics to Visually Impaired Students,” Teaching of Psychology, 20, 242–244.

Nemeth, A. (1972), The Nemeth Braille Code for Mathematics and Science Notation, (1972
rev.), Produced in braille for the Library of Congress, National Library Service for the
Blind and Physically Handicapped by the American Printing House for the Blind.

Raue, K., and Lewis, L. (2011), “Students With Disabilities at Degree-Granting Postsec-
ondary Institutions (NCES 2011-018),” U.S. Department of Education, National Center
for Educational Statistics. Washington, DC: U.S. Government Printing Office.

Section 504 of the Rehabilitation Act of 1973, Pub. L. No. 93-112, 87 Stat. 394

Vita, A. C., and Kataoka, V. Y. (2014), “Blind Students’ Learning of Probability Through
the use of a Tactile Model,” Statistics Education Research Journal, 13(2).

Yamaguchi, K., Komada, T., Kawane, F., and Suzuki, M. (2008), “New Features in Math
Accessibility with Infty Software,” Computers Helping People with Special Needs,
Berlin: Springer, pp. 892–899.

27
 Journal of Statistics Education, Volume 23, Number 2, (2015)

Robert J. Erhardt
Department of Mathematics and Statistics
Wake Forest University
PO Box 7388
127 Manchester Hall
Winston-Salem, NC 27109
Email: [email protected]

Michael P. Shuman
Learning Assistance Center
Wake Forest University
PO Box 7283
Winston-Salem, NC 27109

Volume 23 (2015) | Archive | Index | Data Archive | Resources | Editorial Board |

Guidelines for Authors | Guidelines for Data Contributors | Guidelines for Readers/Data Users |
Home Page |
Contact JSE | ASA Publications|

28