SiMON FRASER UNIVERSITY
MEMORANDUM
To .....
.
.
enate
From
Senate Committee on Academic Planning
?
Senate Graduate Studies Committee
Date...
!Pr
.
i
.
1
.
18...4
.
The proposal for an M.Sc. Program in Statistics was considered and
approved by the Senate Committee on Academic Planning at its meeting of April 4, 1984
and by the Senate Graduate Studies Committee at its meeting of April 16, 1984.
In discussion, it was noted that initially the proposed program could be
mounted with existing faculty from the Department of Mathematics and
Statistics. In the long term, with growth in enrollment, additional
resources would be required.
MOTION:
"That Senate approve and recommend approval to the
Board of Governors, as set forth in S.84-34 , the
proposed M.Sc. Program in Statistics."
WJW/gg
0
SiMON 1iASE1{
MEMORANDUM
UNIVERSITY
?
-
?
s
?
' ?
Senate Ccninittee on Academic
John
?
Webster,Dean
Planii.g
Graduate Studies
Proposal for M.Sc. in Statistics ...
?
Date. ?
...............................
The proposed M.Sc. Prograrrfle in Statistics is attached for your
consideration. It was approved by the Assessrrent Caiittee for
new Graduate prcxjrarrrrs on March 26, 1984, and strongly endorsed,
particularly in light of the positive cczrnents of the external
reviewers.
The proposal is scheduled for consideration by the Senate Corrinittee
on Graduate Studies at its meeting April 16, 1984.
A, ?
a
IL'.Sii U\IvF:ll-;yy
tE!.OflANDU!'
:.
.
a
l .
..
t
:.
Wvk..tr
?
From ?
.
S
Pi&
t 9. ?
ri•t .
Acc"n.ic...
?
.Otin.,of..thitjç,
Subject. ?
M.Sc. PROGRAMME
?
Date ?
)j,ii
19fl4
In response to your memorandum of 29 March, I would like
to make the following comments:
1.
The program can be mounted initially without new faculty. For the
past several years we have offered, through our visitors, sufficiently
many courses at the graduate level to permit students to meet the
requirements of the new program. In addition, in the early stages,
while enrolments are low, we will be able to provide reading courses
where necessary. The consulting service has been manned voluntarily
in the past; we will continue to do this for the short term.
Eventually, however, and particularly if enrolments match our
expectations we will not be able to continue in this way. A reading
course for six to ten graduate students is impractical. Our visitors
must not be regarded as simply part of our teaching program. Our
NSERC funds may not always be sufficient to provide enough visitors.
Sabbaticals might place too great a strain on the system. Finally,
if the consulting service is as successful here as it is elsewhere,
the burden of administering it and supervising the students will
necessitate some teaching relief.
We should point out that a position in statistics was the top hiring
priority in the Faculty of Science this year, and that the Mathematics
Department is generally considered to be understaffed.
2.
We believe our graduates will have adequate experience in statistical
computing. Statistical computing will be an integral part of consulting
as well as the courses SThT 803, 804 and 805. In addition, we have
always encouraged our students to take numerical linear algebra courses,
which are very relevant to statistical computing. We will continue to
do so.
We believe the computing facilities at Simon Fraser are adequate to the
task. They are certainly not comparable to those at Berkeley but we
have a good supply of statistical packages.
In addition, we have easy access to computing facilities. We are
confident that Simon Fraser will continually update the computing
equipment.
.
/
-2-
?
3. ?
As all threereviewers indicate, the consulting service is crucial
to a useful applied Master's Degree in Statistics. The reviewers
make various suggestions concerning the organization of such a
service and we respond to these here.
Professor MacNeill suggests we charge for the service. Eventually,
we would like to do so, in cases where the consultee has research
funds available. Initially, however, we must establish the service in
the client's minds as a useful tool in their work. Furthermore
we must recognize that, at Western, the largest sources of revenue
of this type are grants given to medical researchers. We do not have
a Medical School here.
Professor Brillinger wonders about credit for the consulting courses
and payment to students participating. We believe payment to
graduate students is the exception rather than the rule; if Simon
Fraser wishes to emulate Berkeley, and have the Faculty of Graduate
Studies pay the students we could hardly object but it does not seem
necessary. As to credit for the course, we elected the present route
after discussions with the Registrar. Our suggested method is
administratively convenient. The actual work content of the course
can be evaluated by other institutions on the basis of the project
and letters of recommendation from our faculty.
.
?
Professor Brillinger also wonders about Ph.D. students and students
from other departments participating in the service. At Berkeley
there is a separate department of Biostatistics and it is primarily
these outside students who participate there. We do not expect many
students here from outside the Mathematics Department to be interested
or qualified but we will certainly accept any who are. We expect
our Ph.D. students will participate in the service. There will not
be many Ph.D. students and supervisors will have sufficient authority
to persuade them to participate. Nowadays, every Ph.D. student in
statistics seems to be anxious to do so.
Professors Kalbfleisch and MacNeill suggest application to NSERC for
support for the consulting service. We will do so.
?
4.
?
Replies to Professor MacNeill's further comments:
1) ?
Experimental design and regression methods are covered at SFU
in fourth year undergraduate courses. We will be using our
authority over course selection to ensure that students without
these subjects in their backgrounds make up the deficiencies.
.
?
/
-3-
Sample surveys form an area of statistical expertise rather
different from the usual areas. At present none of our faculty
have particular xpertise in this area. Ideally we would like
to mount a graduate course in sample surveys but our present
proposal already taxes our resources to the limit. If and
when further faculty are appointed we would mount a course in
this area.
Statistical computing will definitely be an important part of
participation in the consulting service.
2)
These courses are quite peripheral and we do not have the resources
to offer them. Occasionally, other departments (Economics for
example) may offer courses in some of these areas. We would
encourage our students to take them.
3)
Projects may involve software development if the opportunity
presents itself. Only time will tell.
4)
We will do our best to expose students to computer graphics and
database management. This will require consultation with computing
science and the computing centre.
5)
We have, in the past, accepted students from disciplines other
than mathematics and statistics. We always require such students
to make up the lacunae in their undergraduate backgrounds. Generally,
they do this in a qualifying semester.
?
0
G.A.C. Graham
GACG/bc 1
SiMON FRASER UNIVERSITY
MIMORAHDUM
0.A.C.
?
......................John M. Webster
ha
•
,4
?
......................
Subj.d. . ?
..c.
Des ......
I!
1
ch
.
?
)
.9
.
The Assessment Committee for the Statistics MSc Programme approved
the proposed programme at its meeting on Monday and recommended its
consideration by the Senate Committee.
The following points were raised by the Assessment Committee and
I suggest that you respond to these issues in a memorandum addressed to me:
1.
Clarify whether or not the additional faculty resources are essential
for the offering of the programme, and, if so, under what circumstances.
2.
The adequacy
of
the computing component in the programme, and of the
computing availability.
3.
All the reviewers were concerned about the consulting serv1ce,and the
format and centrality
of
consulting in the programme. You may wish to
address this issue.
4.
A response to each of the five items raised in Professor MacNeill's
review would be helpful.
I look forward to receiving your reply at your earliest convenience.
11. We
JMW / c b
0
cotfmrrs OF PROPOSAL
PAGE
1.
Introduction. ?
1
2.
Proposal in ricsc Format (with Appendix A).
?
2
3.
Answers to questions posed in draft minutes of meeting of
?
13
Wednesday, May 25, 1983 of Assessment Committee for New Graduate
Programs. Copy of Dr. Webster's memo to G.A.C. Graham of 8/6/83.
4.
Appendix I. Course Proposal Forms; memo from E.A. Weinstein.
?
20
5. Appendix II. New Calendar Entry for Mathematics. ?
43
6. Appendix III. Current C.V.'s for Eaves, Lockhart,
Routledge, ?
45
Stephens, Villegas and Weldon. (Not included for Senate distribution)
.
0
INTIDQCT ION
This program is designed to provide an alternative training plan for
Masters students in Statistics, working in the Mathematics. Department at SFU.
At present the students can take the degree by course work and thesis as in
?
is
other branches of Mathematics. The statisticians at SFU feel that a very
strong case can and should be made for training statisticians with more empha-
sis on course work and actual experience of data analysis through participa-
tion in a consulting service. This is in line with Statistics programs at
major universities throughout the world. At Simon Fraser University, the for-
mal machinery for introducing such a program is already in place; the program
explains how the suggested new M.Sc. program in Statistics will fit in to the
formal structure of SFU's Master degree and presses the need for such a pro-
gram.
0
MASTER OF SCIENCE IN
STATISTICS: PROJECT OPTION
I: General Information
1.
Title of Program: Master of Science in Statistics: Project Option
2.
Credential to be awarded to graduates: M.Sc.
3.
Department to offer program: Mathematics
'I
4.
Date of Senate Approval: ??
5.
Schedule for implementation: 1 September 1984
II: Program DescritIon
1.
Objectives:
a) To expose graduate students
in statistics to the practical nature
of
the discipline through participation In the departmental consulting service
and preparation of a project.
40
b) To expose students to a broader range of statistical techniques
by r.-
quiring (and providing) more coursework. This concentration on
breadth
of
experience rather than depth in one area will be enhanced by the
substitution
of a project for the thesis.
2.
Relationship of program to role and mission of
university:
This program will better serve graduate students in statistics
by provi-
ding them with a broader knowledge of statistics, more soundly based in prac-
tice.
?
Better education is, of course, a basic mission of Simon Fraser Uni-
versity.
?
In addition, we hope the consulting service will enhance contacts
between statisticians and researchers in other departments.
?
This inter-
?
disciplinary effort should increase the research effectiveness of both groups.
3.
Other similar programs in Statistics:
a) In British Columbia: none. ?
For comparison sake, as of Summer 1983:
University of Victoria: Offers a program with a requirement of 30 semes-
ter hours of credit and a thesis or a comprehensive examination program. The
thesis program requires 16 of the 30 semester hours be at the graduate
level
while the comprehensive examination program requires 24 of the 30 semester
hours at the graduate level. Neither program has the consulting
service
requirement of this program. Very few students take either program in Statis-
tics.
University of British Columbia: A thesis program with 24 semester
hours
of coursework of which at least 12 hours are at the graduate level.
?
Consul-
?
ting service experience is optional.
2.
b) Outside of British Columbia: ?
Programs requiring 25 to 30 scineter
?
hours of coursework and a final examination are available at Carleton,
Toronto,
Western Ontario, and York.
?
Programs requiring coursework and a
?
project or *major project* or paper" are available at McMaster, the Univer-
sity of
Alberta, and Waterloo. ?
At McMaster consulting service participation
is required. ?
The McMaster program is similar to our proposal.
4.
Relationship of
proposed program to those at the University of Victoria
and the University
of
British Columbia:
The program has two crucial differences from those at both other
schools
first, we require more graduate coursework than any of the thesis
programs - 20 semester hours as opposed to 16 or fewer; second, and most
important, we are alone in requiring participation in the Statistical Consul-
ting service. ?
The trade-off is that the project is less extensive than the
thesis. ?
This program is therefore more suited to providing broad exposure to?
statistical techniques
for students.
5. Curriculum:
We will be requiring 28 semester hours of coursework, at least 20 of
which are at the graduate level with the remainder at the 400-level or
higher.
?
In addition, students will be required to participate in the consul-
ting service
for two semesters. ?
We propose to administer this consulting
?
requirement by creating
two courses; Statistical Consulting I and Statistical
Consulting U.
?
Treating consulting as coursework has these advantages: the
students' transcripts will show that they have practical experience; the
serious commitment of
the statistics group to the importance of practical work
will be demonstrated
to the students, and formal supervision of the students'
participation will
be provided. Finally students will be required to submit
and defend a project. This would likely consist of the thorough analysis of a
data set, generally one encountered in the consulting service, with a written
report on the
analysis.
More detailed course
requirements are listed below:
a) Existing courses: We will be using
MATH 872-4 Probability I
MATH 873-4 Probability II
MATH 874-4 Probability: Selected Topics
as optional courses.
In addition we propose to rename:
MATH 875-4 Statistics I
?
as
?
STAT 801-4 Mathematical Statistics
MATH 876-4 Statistics II as ?
ST1T 802-4 Multivariate Analysis
and
?
MATH 877-4 Statistics -
Selected Topics as STAT 890-4 Selected Topics in
Statistics
The renaming serves two administrative purposes. ?
It allows all sta-
?
tistics graduate courses to be numbered rationally and it permits
4.
statistics courses to be listed toqether in the calendar.
We will be requiring STAT 801-4 and STAT 802-4 as compulsory courses.
In addition we will be using our existing undergraduate courses at the 400-
level as optional courses and, where thought necessary by a student's program
advisor, as required background courses for graduate courses. We will also be
using on occasion some graduate courses offered in other departments (for
example Time Series courses in Economics or Experimental Design courses in
Psychology).
b) New courses: (Course Proposal Forms attached in Appendix I)
j
)
Statistics Courses
STAT 803-4 Data Analysis
STAT 804-4 Time Series Analysis
STAT 805-4 Nonparametric Statistics and Discrete Data Analysis
MATH 871-4 Applied Probability Models
The courses STAT 803-4 and STAT 804-4 are computer oriented. In order to
ensure that our students have at least a minimum exposure to computer methods
we require students to take either STAT 803-4 or STAT 804-4 which are computer
oriented. A course in numerical linear algebra is in the approval process and
we intend to permit students to use this course, when it is approved as an
• ?
alternative route to satisfying our minimum computing requirements. We expect
our students will generally get much more exposure than this minimum through
the project, the consulting service and other coursework.
ii) Statistical Consulting
STAT 811-0 ?
Statistical Consulting I
STAT 812-0 ?
Statistical Consulting II
These courses would both be required.
?
They will be graded on a Satis-
factory/Unsatisfactory basis. ?
The use of two course numbers is intended to
?
make it easier for the Office of the Registrar to record two semesters of par-
ticipation on a student's transcript.
8. Details of consultation with non-university agencies
Drs. Weldon, Routledge and Stephens have considerable contacts with
government and industry; this comes about from consulting, from their
official
positions in Canadian Statistics and in Weldon's case because he has just
finished an NSERC Industrial fellowship under
McMillan-Bloedel.
Stephens is
president of the Statistical Society of Canada and as such has contact with
the Department of Energy, Mines and Resources, Statistics Canada, etc.
It is
very clear that such bodies recognize the ever present need to keep statistics
graduates up to date. Statistics Canada, for instance, have emphasized their
approval of the University of Waterloo's co-op program, because it means that
students get real experience with data, which is often of necessity
contaminated or partly
lost, etc. In industry also, the need is perceived for
statisticians
who can interact with others in the field and also with execu-
tivss and
managers. Nevertheless, much education is still needed in Canadian
industry to improve
awareness of what can be done with better methodology.
III. Need for Programi
Ii) The
need for this program arises because we feel that statisticians need
better training
at the Master's level. Because there are few Departments
of Statistics in Canada
(although this is changing), students entering
Masters programs in Statistics at most Canadian universities usually have
the B.Sc. in Mathematics. In their program they would have had at most
three or four courses in Statistics, compared with the full program which
a student in a Statistics Department would- have taken. Students some-
times enter Statistics programs from other disciplines, for example,
Psychology, Economics, Biological Sciences, etc; these are good back-
grounds, and students from other subjects should be encouraged to study
Statistics, since it leads to better applied scientists.
b) However, Statistics has now become a very vast field, and we want our
students with a Masters degree to leave with a broad knowledge of this
field. At present the course structure of the degree with the thesis is
too limited; students take few courses, and so have no course work in
wide areas of Statistics. They then write a thesis, which may or may not
contain a research component (in our Department it usually does), but of
necessity this is in a limited topic. The typical student spends a great
deal of time and energy on the thesis.
C)
It
cannot be over-emphasized that statisticians occupy a special role;
according to the position they hold they interact with specialists in
many other disciplines, and are looked to for practical guidance of how
to analyze difficult data. The Masters degree is the level at which they
should receive a
broad applied training. This is what this program is
intended to provide, by giving a wider range of course work, together
with good practice at interacting with other scientists (or other workers
with data) on their statistical problems. This will be done by having the
students take part in our consulting service. We expect our students to
learn a great deal from this part of their degree work. Such a Sta-
tistical consulting service has been set up by many universities in
Canada and the United States. It is recognized as a first class training
ground for applied statisticians. (Incidentally, it also has many
bene-
fits to the University community at large).
d) Other universities also recognize this as a preferred method of training
good applied statisticians, who will go out into industry, government,
and research science positions of all kinds. For example, course work
and no thesis is the accepted method of obtaining a Masters degree at the
University of
Toronto in Canada; Stanford University graduates at least
twenty-five master's students in Statistics each year without a thesis;
these students are
statisticians who will
not usually go on to do further
academic research
in Statistics. There are many other universities with
similar programs.
6.
• ?
It is also true that government (for example, Stats Canada), industry and
business recognize that wide training plus practical experience with data
is the necessity for a good statistician. The University of Waterloo in
Canada enjoys a high reputation in these circles because it gives this
type of training, even at the B.Sc. level.
e) Nowadays in Canada there is much stress laid, especially in high govern-
ment circles, both federal and provincial, on the need for the country to
develop high technology, and efficiency in its production. We cannot hope
to do this without producing first class applied mathematicians and sta-
tisticians. The great centers of high technology in the United States,
for example, draw extensively on the faculty and graduate students of
such Universities as Harvard, MIT, Stanford and Berkeley.
Provincially,
the government is encouraging Discovery Parks near the Universities and
Institutes of Technology. It is the Universities' job to rise to this
challenge and to lead the way in producing men and women with the trai-
ning to make them leaders in this field in Canada. (At present, when we
do produce first class people, there is a tendency for them to emi-
grate). The situation is circular; first class people will produce first
class technology, and first class technology will require people with
first class training. We are trying to look ahead and meet the need by
providing this training.
2) Enrolment
•
?
?
a) Evidence of student interest: our recent M.Sc. students have
?
expressed their feelings that they would have preferred a heavier course con-
tent, over as much of the field of statistics as possible, rather than fewer
courses with concentration in a thesis.
?
On a wider scale,
students (or gra-
?
duates, a little later on) often express similar sentiments. There is, in
fact, a continuing dialogue in the profession on how
statisticians should be
educated, with a broad consensus on the importance of applications, computer-
methods, and practical consulting.
b
& C)
Present enrolments in M.Sc. programs in Statistics amount to
roughly 4 new students per year.
?
Initially this is expected to continue
eventually rising to perhaps 10 per year.
?
Since the program requires ap-
?
proximately two years to complete the total increase in enrollment will be 12
FTE's.
Please refer also to page 12 below titled Enrollment in New Courses.
d) Growth limits: Growth is limited primarily by
availability of faculty
to supervise the consulting service. It seems that an upper limit of 12
students per year ( or roughly 24 enrolled at any one time)
is reasonable
here. ?
The limit is simply calculated as 4 students per faculty member at any
?
one time; if we obtain new faculty this limit would rise.
We project a minimum enrolment, of 10 students.
40 ?
3) Types of Jobs for which graduates of this program will
be suitable.
a) Statistics Canada, Department of Energy, Mines and Resources, and
other
government agencies.
b) Corresponding Provincial Agencies.
C)
Industrial
jobs
involving all levels of statistical analysis, e.g.
managerial (sales, forecasting, trends, etc.) or related to field work - envi-
ronmental, forestry, geostatistics, etc.
6) Medical Research Centres.
8)
Data gathering centres such as the Cancer Control Agency of B.C.
f)
Pharmaceutical companies (drug testing is a major source of employment
of Statisticians)
g)
The teaching profession at all levels.
4) Placements of our graduates over past three years.
a)
Spinelli - Cancer Control Agency
b)
Hutchinson - B.C. Tel., marketing
c)
Holguin - Not known, now in Denmark
d)
Byrd - Computing Centre at UBC; then in marketing in Ontario
0)
Janicot - Research division, French Army
f)
McLaren, Mab - to Ph.D. program
g)
Note that several students are doing an M.Sc. while working - e.g.
Chung in B.C. Systems Analysis.
IV. Present and Projected Resources
1)
Administrative Personnel: No new administrative personnel will be
required.
2)
Faculty:
(i)
To cover Stat 803, 804, 805 and Math 871 to be offered on a
rotating basis (each course once every two years) we need 2/3 of a faculty
member. In 83-84 we will be offering special topics versions of Math 871 and
Stat 804. Initially we would hope to be able to start the program without a
newfaculty member
by
offering some reading courses. We have written our
budget request on the basis of one faculty appointment in the Fall of 1984, at
the Assistant Professor level.
(ii)
The consulting courses can be mounted initially without any
new faculty. In the long run, however, they will probably require 1 faculty
member. We are not presently asking for faculty for this purpose. Our present
consulting service is manned voluntarily by Statistics faculty who get no
.
S
8.
formal credit; this situation is unlike that at most other
universities and
cannot be expected to continue indefinitely.
3)
Library resources: No new resources needed. See the memo from
E.A. Weinstein at the end of Appendix 1.
4) Capital Costs:
a) Office space: We will eventually need office space for 2 new
faculty members and 12 new graduate students.
b) Computing costs: There will be no direct need for new computer
services. However, the 2 new faculty members and 12 new graduate students
would use the computer facilities in their courses, research and the consul-
ting service. This probably represents only a small increase in total Mathema-
tics Department computer usage.
5)
Anticipated external funds:
a)
for teaching: none.
b)
for research: the 2 faculty members will be expected to have
NSERC grants as usual. See
?
below.
NSERC grants for present statistics faculty for 1984-85:
NAME ?
AMOUNT
n' ?
'r
Dr. D. Eaves
Dr. R. Lockhart
Dr. R. Routledge
Dr. M. Stephens (1983-84)
(1984-85 - in process)
Prof. C. Villegas
Dr. K.L. Weldon
$ 4,295.00
$ 4,295.00
$ 14,500.00
$ 19,381.00
$ 12,248.00
$ 5,777.00
•
Note: (Pages 9, 10, 11 on budget deleted).
SIMON FRASER UNIVERSITY
?
12.
0.
ENROLLMENT IN NEW COURSES
M-SC. IN
ST
ATISTICS: PJ)JECp OPTION
?
Annual FTE
?
Annual FTE
?
EnrollmentsProgram
?
?
in ?
New
Enrollment
courses
in
?
4 ?
2.3
?
8 ?
4.6
?
8 ?
4.6
Credit Units of?
?
Year ?
New Courses
?
1984-85 ?
8
?
1985-86 ?
8
?
1986-87
?
8
NOTE: The program normally lasts two years. A full time load is therefore 14
credit
therefore
hours
calculated
per year.
as
The
(credit
column
units
Annual
of new
FTE
courses
En
rollments
/ 14)
in
x Annual
New Courses
PTE
is
Enrollments in
P
rogram. This calculation is based on Sections A-i, A-2, B-i,
1-2 and 8-3 of New and Emergent Programs: Budget Submission Guidelines.
The figures presented here are based on a
continuation
of present
enrollment levels of roughly four new students per year. We hope
enrollments
will rise reasonably quickly to roughly 10 new students per year but have not -
included any such increase here.
S
0
Answers to Questions Posed in Draft Minutes of Meeting of Wednesday, May
25,
. ?
1983 of Assessment Committee for new Graduate Programs and Memo of Dr. Webster?
to Dr. Graham of 8/6/83.
NOTE: For reference a copy of Dr. Webster's memo follows these questions and
answers.
1 Questions:
(a) The proposed programme should be consistent in using the word 'pro-
ject' rather than 'non-thesis'.
(f) Identify more clearly the resources needed for offering the proposed
new courses; how frequently are these courses to be offered?
(g) The proposed programme should be rewritten and repackaged in a style
appropriate for presentation to Universities Council. In this form it will
also be forwarded for external assessment.
(i) The need (and supporting data) for a more developed rationale about
the demand for such a programme? What is the manpower projection for such a
programme?
(m) Are any additional computing resources needed and if so when might
they be needed, the cost and who is going to pay?
S
Also Points 2,3,4,5,9,10 of Dr. Webster's memo.
Reply:
?
See proposal attached in UCBC format.
2 Questions:
(b) It is not clearly stated whether or not the project is to be formally
defended and lodged in the Library.
(C)
Does the project carry any credit hour value?
Also point #6 of Dr. Webster's memo.
Reply: ?
Simon Fraser University general regulations (See 1983-1984 Calendar
page 213 #1.7.2.; ?
page 214 #1.9.1. and page 216 #1.10.6.) provide three ?
Acceptable options for Masters degrees: a thesis, two extended essays, or a
project.
?
Projects are less extensive than a thesis but are subject to the
same general regulations. They must be examined as a thesis would be, super-
vised as a thesis would be, and lodged in the library as a thesis would be.
We see our projects as not very different in style and scope from those for
which we have served as external examiners in the Department of Economics.
The general outline of our project Option is similar to theirs with the excep-
tion of consulting.
?
The project does not carry any credit hour value.
The project itself would generally consist of the thorough analysis of
some data set, generally one encountered through the consulting service,
I 1.
together with a written report on the analysis.
3 Questions:
(d)
Is the M.Sc. degree programme with project option designed
80
as to
allow direct entry into a Ph.D. degree?
(e)
If the student wishes to pursue a Ph.D. degree following the 'project
M.Sc.', is the student required to-do any further course work?
(o) How may the better than average student make the transition to a
broad-based research Ph.D. degree without completing a regular Master's re-
search degree?
Point #7 in Dr. Webster's memo.
Reply: ?
Students who have completed our M.Sc. degree will be well pre-
?
pared to begin a Ph.D. program in Statistics either at Simon Fraser or else-
where.
?
Those students seeking a theoretical Ph.D. would probably need to do?
coursework in Probability Theory and Real Analysis.
Students proceeding to a Ph.D. at Simon Fraser would certainly need to do
extra coursework in Analysis and Probability to prepare them for departmental
comprehensive examinations.
The reference in (o) above to "a regular Master's research degree" is not
entirely relevant to the Department of Mathematics since in some cases
Master's theses in this department presently contain little or no original
research.
?
is
4 Questions:
(h) How many students do they plan to admit? What is the impact of this
type of programme on the rest of the department and the
relationship
of the
new programme with existing programmes.
Point #8 of Dr. Webster's memo.
Reply: ?
Our projected enrolment initially is: S students per year.
?
Refer ?
to section III of the UCBC proposal.
The program will have no effect on the non-statistics part of the depart-
ment except in so far as it increases enrolment in numerical analysis
courses. ?
We expect virtually all of our new M.Sc. students in statistics
will pursue the project option.
?
All our present students have expressed
enthusiasm for the concept. The
high
enrolment in the very practically
oriented program at the Univrs1ty of Waterloo indicates the general popu-
larity of mch approaches.
5 Questions:
(j) Request additional information about the programmes at other univer-
sities and whether or nor these institutions are fulfilling current needs.
Reply: Please refer to section 11.3 and section 111.2 of the UCBC proposal.
14.
.
6 Questions:
(k) ?
Request additional information about the current statistical consul-
ting service the Department is presently doing.
?
Explain the relationship of
?
this service to the two new consulting courses.
(1)
Give
an enlarged course description for the consulting courses.
(s)
What is the difference between the two consulting courses?
(t)
Could the consulting courses be renamed 'internship' courses?
Point #11 in Dr. Webster'.s memo.
Reply: The Mathematics Department presently offers a free statistical consul-
ting service staffed by Statistics faculty. It is available to faculty and
graduate students throughout Simon Fraser wanting statistical advice concer-
ning their research.
At present the service operates as follows.
?
Statistics faculty make
?
themselves available for two hours per week - generally two faculty members at
a time. Clients drop in and get whatever advice they can.
?
They often come?
when it is too late, but the big problem is lack of follow up.
We think that mandatory participation by students in the consulting ser-
vice will provide some follow up. Statistical faculty would still make
them-
selves available
for 2 hours per week (or more if the service were in great
demand and the faculty could afford the time).
?
They would be accompanied by
our students. ?
When clients drop in the problems will be discussed as usual
but more sophisticated analyses would be considered.
?
The students would,?
under faculty supervision, seek out the literature and apply, with the client,
better techniques. We think this process benefits all three parties.
Clients, students and faculty, perhaps to different degrees, learn about the
difficulties of the data, about modern techniques and about the statistical
literature.
We intend to grade the courses only on a Satisfactory/Unsatisfactory
basis.
The consulting courses are named consulting courses rather than "intern-
ship" because "consulting" is the statistics "industry standard" term. We
have given two course numbers for consulting purely to enable the Office of
the Registrar to record easily two semesters of participation in the service
on a student's transcript.
7 Questions:
(n) It should be clearly stated that if a student has already taken MATH
875, 876, 877, the student may not take STAT 801, 802, or 890.
0
111..
• ?
(q)
(p)
Should
Greater
there
?
consistency
be a specificallyof
?
course
stated
outlinesCalendar ?
is ?
needed,
entry?
?
in style and
?
con-
0
tent.
Generally, the course outlines
are too brief.
Point #12 in Dr. Webster's memo.
Reply: ?
A revised calendar entry for Mathematics 9raduate programs is at-
tached in Appendix II.
?
Revised course outlines appear In Appendix I.
8 Questions:
(r) Request up-to-date CV's of faculty members.
Reply:
?
Current CV's for Eaves, Lockhart, Routledge, Stephens, Villegas and
Weldon are attach
e
d. (Appendix III)
9 Questions:
Point #13 of Dr. Webster's memo.
Reply: Applicants for this program will be processed by the Departmental Gra-
duate Studies Committee as usual.
0
%j%jI)\
JA (A
S
IA( I \ 1 \ L UHf
MEMORANDUM
Dr ... Graham, C...irma...
.
From
?
John ti. Webster
Mathematics Department
?
Dean of Graduate Stu!j:
tLstics..roo...
I write further to my memorandum of 31 May 1983 in which
I gave
you a list of comments on your proposed Statistics
p
rogramme in the form
.
of the unappL-oved minutes of the first
meeting of the Assessment Committee. As promised, I now write
and provide you with further reflections on this programme.
There will be sonic overlap of the comments made in this
memorandum with those made earlier. On the other hand, the
combination of the information in the two memoranda should 'jive
you a good indication as to the response of the Committee to
your proposal.
I will present my points in point form to
aid
clarity.
1.
My sense of the meeting was that the Committee members
. ?
were very much in favour of the programme but they were
somewhat concerned with the proposal as it now stands and
they wanted it to be revised so as to incorporate several
matters that will enhance the final product.
2.
The basic impression was that the programme, as
presented, was somewhat confused in its packaging in
that it was difficult to extract the dppropriate informa-
tion in order to be able to answer the various questions
that are likly to be raised by SCAP and Universities
Council. I urge you, once again, to refer to the
suggested outline for"new proposals"that I attached to my
previous memorandum.
3. I suggest that amongst other things
three clearly identifiable main areas.
philosophy of the programme, second the
curriculum design to resarch needs and,
calendar entry. Many of the comments I
automatically fall within one or other
your proposal include
First, the
nuts and bolts of
third, the
make below will
Df these three areas.
4. You should make a compelling case for the need for this
programme in view of current budgetary constraints in the
Province. The proposal should readily appear as a
"necessity" rather than "a nice to have" programme.
continued ........
-
?
2 ?
-
?
18.
more
the
5. The
p
clearly
roposal.
resources
stated
If new
r
equired
and
courses
not
to
C
are
onflict
offer
to be
this
in
offered
d
p
ifferent
rogramme
then
parts
must
where
be
of
are
even
course
in
the
though
offerings
resources
?
m
anpower,
it may
in
to
appear
other
or
give
are
these
as
areas?
we
a
doing
hidden
new
This
co
it
urses.
cost.
by
in itself
d
iminishing
I
Does
suggest
this
is
a
require
cost
the?
that you also consider computing costs as well as library,
a
perhaps
is
hardware
dditional
an
a
dditional
you
and
coat
should
personnel
in
pressure
that
consider
area
costs.
on
whether
too.
the
If you
D
epartmental
or
believe
not there
that
Office
is
there
an
then
thesis"
6.
Please
de
q
clarify
re, The
further
question
the
arose
meaning
as to
of
whether
the term
really
"
non-
it should be called a "project" degree, flow close is this
to an extended essay format; do you intend to type-up,
bind and lodge the projects in the Library the same as
Occurs with extended essays; how long will the project
take;
what is the nature of the
sup
ervision; why are there no
credits for the project; will the project be formally
defended like a thesis?
7.
I suggest that further
consideration
is given to
clarifying the possible flow from an M.Sc. in this area
entry
other
from
into
an
a
into
words
Ph.D.
M.Sc.
a
what
Ph.D.
programme.
in
would
Statistics
after
the
having
Is
admission
there
into
taken
a
a
Ph.D.
problem
this
requ
pr
irements
form
ogramme
with
of
tra
a
?
be
nsferring
for
In
master's degree.
area
no
8.
doubt
It
of
would
interest
would
be
form
worth-while
and
part
strengths
of the
relating
in
introduction
the
the
D
epartment.
pr
ogramme
and would
This
to the
clarify for outsiders and, indeed, for other members of the
F
University
u
rthermore,
the
it
thrusts
would indicate
and
d
irections
how this
within
new emphasis
your
D epartment.
on
and
the
strengths.'
Master' degree would integrate with other
pr
ogrammes
a
9.
these
master's
How
i)ro(Jrammcs
large
degree,
are
satisfy
the
if not
p
rogrames
the
why
needs
not?
in
for
In
other
other
stat
uni
isticians
words,
versities
the
with
?
Do
fact
justifi
that other
cation
uni
versjtj(
for us doing
.
s do
it
have
also.
this
On
form
the
of
contrary,
degree is
it
no
may
be a good Y:(.1,70n for u not putting forward this degree.
10.
A forecast of enrolments in association with the
with
comments
associated
made to
r
the
ationale,
item above
i fully
would
understand
be
h
elpful,
that
together
such
figures would be guesstimates but, nevertheless, they would
provide
a
nd
:
the line
the Committee
of thinking
with
that
samples
the Department
of data from
is Using
el
sewhere
in
pr.paration of the programme.
continued ..........
-3-
.
? 11.
Please
clarif y
the mode of operation
of
the Statistics
Consulting courses. Will these operate as internships
or are
credits offered for these courses, and if so why
two credits?
Do individuals get paid for a
consultation
as well as obtaining credits? This may well be a conflict
of interest. In what way does this consulting course
differ from the current problem-solving consultation
which already is offered by the Department?
12. There is a feeling in
the Committee that the course
description was rather inadequate and certainly inconsistent
from one course to another, especially as
regards
to the
description of the course content and the sample reading
and general material. Some statement should be made on
possible overlap of
credits between courses if this
occurs.
13. One final point that arose
organizational
procedures
to be
In other words, will applicants
processed by the regular Departi
just the same
as applicants for
within the Department.
was the question of the
used within the Department.
for this prorammne be
nental Graduate Committee
other dcqrce programmes
I shall
be very pleased to
convenience
to discuss aspects
made in
the earlier memorandum.
deliberations. The Assessment
forward to seeing the rewritten
meet with you at our mutual
?
of
the above points and those
Best of luck in your
initial
Committee 3 am sure will look
application of the proposal.
I,
/
?
\\\
I
?
C
0
SIMON FRASER U'NIVF.RSITY
U
New Graduate Course Proposal Form
.
CALENDAR INFORMATION:
Department:
MATHEMATICS
?
Course Number:
MATH 871-4?
Title:
Applied Probability Models
Description:
Applications of stochastic processes: Queues, inventories,
counters, etc.
?
Reliability and life testing. ?
Point proc'cnes.
?
Simulation.
Credit Hours: 4
?
Vector: 4-0-0
?
Prerequisite(s) if any:
Math
387 or
equivalent
ENROLIT AND SCHEDULING:
Estimated Enrolment: 5 ?
When will the course
first be offered: 84-3
How often will
the course be offered: Every second year
JUSTIFICATION: ?
(see attached)
RESOURCES:
Which Faculty member will normally teach the course: Lockhart, Weldon
and
others
What are the budgetary implications of mounting the
course: 1/6 of a faculty
member.
Are
there sufficient Library
resources (append details): Yes, see attached.
Appended: a)
Outline of the Course.
b) An indication of the competence of the Faculty member to give
the course.
C)
Library resources.
APPROVED: Dept. Graduate Studies Ctte:
?
Date:
Faculty Grad. Studies Ctte:
?
Date:
Faculty:
?
Date:
Senate Grad. Studies Ctte:
?
Date:
Senate:
?
Date:
C
COVERING MEMO CONCERNING MATH 871-4: APPLIED PROBABILITY MODELS
?
.
Justification:
This course is part of the proposed project option in Statistics. The
topics covered have broad application in such diverse areas as biology,
epidemiology, demography, traffic engineering, quality control, astronomy and
many others.
?
This breadth makes the course important and useful to prac-
?
tising statisticians.
Competence of faculty members:
Dr. Lockhart has done and is doing research in this area.
?
Dr. Weldon
has consulting and research experience in this area.
?
(Dr. Weldon has pre-
?
viously given a special topics course in this area.)
Library Resources:
The library is well supplied with texts in this area and subscribes to
all the major statistical journals. ?
See attached memo from E. A. Weinstein.
0
MAT4ATICS 871-4
APPLIED PROBABILITY MODELS?
Course Outline
Many processes which produce data are known to be probabilistic in nature -
such processes are called stochastic processes.
?
Examples are: the times at
which clients are served in a bank, or by a telephone exchange; the time in-
terval between arrivals of cosmic rays, or of events such as incidence of a
disease or of earthquakes; the genesis and spread of cells; the locations of
mineral deposits (spatial pràcesses).
?
This course covers the probabilistic
analysis of such processes which produce data to he analyzed by statistical
techniques. ?
Specific processes and topics are:
1.
Review of basic results in probability and stochastic processes: random
walks, Markov.chains, simple queuing systems, Poisson processes, branching
processes, birth and death processes.
2.
General queuing systems: inventories, storage, counters, traffic flow and
other applications.
3.
Renewal processes: reliability, life testing, survival analysis.
4.
Sampling models in stochastic processes, e.g. applications to
engineering
and medicine.
S. Point processes: in space in time; general theory and applications to
biology, mining, geostatistics, astronomy, demography.
6. Simulation techniques in exploring the properties of complicated models.
Suggested Text: Ross, Sheldon M., Introduction to Probability Models.
Academic Press, 1972.
This book will be supplemented by course notes in areas such
as 5 and 6 above.
S
0
24.
Name Change Only
SIMON FRASER UNIVERSITY ?
0
New Graduate Course Proposal Form
CALDAR INFORMATION:
Department:
MATHEMATICS ?
Course
Number: STAT 801-4
Title: Mathematical. Statistics
Description:
Advanced mathematical statistics.
?
A survey of basic concepts
in point e;timation, interval estimation and hypothesis testing. ?
Principles
of Inference.
C[editflours: 4
?
Vector:
4-0-0
?
Prerequisite(s) if any:
Students
with credit for Math 875-4 may not receive credit for Stat 801-4.
ENROLMENT AND SCHEDULING:
Estimated Enrolment: 5 When will the course first be offered: N/A
How often will the course be offered: Once a year
is
JUSTIFICATION:
This course Is replacing MATH 875-4 (Statistics I).
?
This is a name change
only.
RESOURO:
Which Faculty member will normally teach the course: Eaves, Lockhart,
Routledge, Stephens, Villegas & Weldon
What are the budgetary implications of mounting the course: None.
Are there sufficient Library resources (append details): Yes.
?
None needed.
APPROVED: Dept. Graduate Studies Ctte:
Faculty Grad. Studies Ctte:
Faculty:
Senate Grad. Studies Ctte:
Senate:
Date:
Date:
Date:
Date:
Date:
0
'_2!i!Y
SIMON FRASER UNIVERSITY
New Graduate Course
Proposal Form
CALENDAR INFORMATION:
Department:
MATHEMATICS
?
Course Number:
STAT
802-4
Title:
Multivariate Analysis
Description:
An advanced course in Multivariate Analysis.
?
Factor analysis,
discriminant analysis, principal components, canonical, correlations.
Multivariate regression and analysis of variance.
with
CreditHours:
credit for
4
?
Math
Vector:
876-4 may
4-0-0
not
?
receive
Prerequisite(s)
credit for Stat
if
802-4.
any: Students
?
EMLKENT AND SCHEDULING:
Estimated Enrolment: 5
?
When will the course first be offered: N/A
How often will the course be offered:
Once
a year
. ?
JUSTIFICATION:
only.
This course is replacing MATH 876-4 (Statistics II).
?
This is
a
name
change?
RESOURCES:
Which Faculty member will normally teach the course: Eaves, Lockhart,
Routledge, Stephens, Villegas & Weldon
What are the budgetary implications of mounting the course: None.
Are there sufficient Library resources (append details) : Yen.
?
None needed.
APPROVED: Dept. Graduate Studies Ctte:
?
Date:
Faculty Grad. Studies Ctte:
?
Date:
Faculty:
?
Date:
Senate Grad. Studies Ctte:
?
Date:
Senate: ?
Date:
0
SIMON FRASER
UNIVERSITY
New Graduate Course Proposal Form
CALENDAR INFORMATION:
Departments
MATHEMATICS
?
Course Number:
STAT 803-4
Titles
Data Analysis
Description: A
problem based course emphasizing the exploratory aspects of
statistical analysis
with emphasis on modern computer-oriented methods.
Credit Hours: 5
?
Vector: 4-0-0
?
Prerequisite(s) if
any: Math 472 or
equivalent or permission of the instructor
ENROIM&V1' AND SCHEDULING:
Estimated
Enrolment: 4
?
When will the course first be offered: 85-1
How often will the course be offered: Every second year
JUSTI?ICATIONZ
?
(see attached)
aE8ouES:
Which Faculty member will normally teach the course: Weldon, Routledge &
others.
What are the budgetary i
i
n
j
iiCatithlb vr
mounting the course: 1/b of a faculty
member.
re there
uELicieflt
Library resources (append
details):
Yes, see attached.
Appended: a) Outline of the Course.
b)
An indication of the competence of the Faculty member to give
the course.
c)
Library resources.
APPROVED: Dept. Graduate Studies Cttez -
?
Date:
Faculty Grad. Studies Ctte:
?
Date:
Faculty:
?
Date:
Senate Grad. Studies Ctte:
?
Date: -
Senate: ?
Date:
S
6-
;.7.
COVERING MEMO CONCERNING STAT 803-4:
DATA ANALYSIS
..
Justification:
This
course is part of
the proposed project option in Statistics.
?
The?
course deals
with
modern techniques widely used in the initial exploration of
large data sets.
?
Such
data sets are now very common in the social sciences,
medicine and other fields.
?
Exposure to the largely computer based techniques?
in this course is important to any practising statistician.
Competence of Faculty Members:
Dr. Weldon has had extensive experience through consulting and research
in this area. This material is standard in the conr.ultinj
e
xperience and
training of Dr. Routledge and other members of the statistics group.
Library Resources:
The library is well supplied with texts in this area and subscribes to
all the major statistical journals.
?
See attached memo from C. A. Weinstein.
[I
0
STATISTICS 803-4
DATA ANALYSIS
Course Outline
A striking revolution has occurred in Statistics with the arrival of the
computer. ?
Valuable methods of analysis were in the past not available be-
cause of lack of rapid computation. These methods, for example in analysing
Large quantities of data,
or data which is multivariate, are now available.
Also, new methods have been developed, which are computer intensive, for which
the property can be found out by extensive computer analysis.
?
Finally gra-
phical methods are
now developed for rapid assessment of data.
?
Computer-
oriented statistical analysis
is the subject of this course. The intention
is to bring the student to the forefront of modern methods of statistical
analysis.
Specific topics are as follows:
1.
Reduction of dimensionality. This is useful in simplifying the data set
when the information
content of several variables may be represented by
values of only a few variables.
2. Study
of multivariate dependencies. The main problem in this section is
that of determining whether (and to what degree) two sets of variables
carry any information in common.
?
0
3.
Multidimensional classification and clustering.
?
Classification is the
allocation into describable
categories of uncategorized items.
Cluaterinq is the determination of the existence and nature of categories
that
are
previously unknown or unexpected in the group of items measured.
4.
Assessment of statistical models. In this section the parametric models
(and related estimation and hypothesis tests) that have been proposed for
multivariate data are discussed, along with methods to check the
applicability of such models in particular instances, and to improve the
fit of such models when necessary.
S. Summarization and exposure of multivariate data.
?
The emphasis here is on?
graphical methods of exploring multivariate data sets.
Suggested Text:
?
Tukey, J. W., Exploratory Data Analysis. Addison-Wesley,
1977
This text will be supplemented by course notes and journal
articles.
2b.
.
0
SIMON F1AS}R UNIVERSITY
New Graduate Course Proposal Form
CALENDAR INFO14AT ION:
Department:
MATHEMATICS ?
Course Number:
STAT 004-4
Title:
Time Series Analysis
Description:
An introduction to time series models and their analysis.
?
Both
?
time-domain and frequency-domain techniques will be studied.
Credit Sours: 4
?
?
Vector: 4-0-0
?
Prerequisite(s) if any:
?
Math 472 or
?
equivalent or permission of the instructor.
ENROU4ENT AND SCHEDULING:
Estimated Enrolment: 5 When will the course first he offered: 85-3
How often will the course be offered: Once every two years
JUSTIFICATION:
See attached
• ?
memo.
Which Faculty member will normally teach the course:
Villegas, Routlecige &
others.
What are the budgetary implications of mounting the course: 1/6 of a faculty
member.
Are there sufficient Library resources (append details): Yes, see attached
memo.
Appended: a) Outline of the Course.
b) An indication of the competence of the Faculty member to give
the course.
C)
Library resources.
APPROVED: Dept. Graduate Studies Ctte:
?
Date:
Faculty Grad. Studies Ctte:
?
Date:
Faculty:
?
Date:
Senate Grad. Studies Ctte:
?
Date:
Senate:
?
Date:
30.
COVERING MEMO CONCERNING STAT 804-4: TIME SERIES ANALYSIS
Justification:
This is part of the proposed project option in Statistics.
?
This course?
studies techniques for analyzing observations on time-dependent processes.
Such techniques have wide applicability. Example of areas of application in-
clude the analysis of seismic data in geology, the search for patterns in the
fluctuations in
economic activity, and the study of periods in the vibrations
of aircraft speeding down a runway. ?
Such breadth of applicability makes this?
an important part of statistics.
Competence of Faculty Members:
Prof. Villegas is currently active in research in this area. Dr.
Routledge
has worked on both theoretical and applied problems in time series.
Time Series Analysis is an important part of the background of most members of
the Statistics Group in the Department of Mathematics.
Library Resources:
The library is well supplied with texts in this area and subscribes to
all the major statistical journals.
?
See attached memo from E. A. Weinstein.
?
0
0
STATISTICS 804-4
?
TIME SERIES ANALYSIS?
Course Outline
A time series is a sequence of values (or a vector of values) occurring in
time. Examples are the strengths of earthquake shocks at various times, the
daily stock price of a commodity, the sales of airline tickets per month, the
unemployment figures per quarter, etc.
?
In analyzing such series a model is
proposed, and the required parameters estimated.
?
Testing the model is also
very important. This course describes the historical approaches to finding
the correct model, together with methods of analysis, and of predicting fu-
ture values.
1.
The separation of trends, cycles and random perturbations in a time
series. ?
Moving averages and seasonal adjustment.
?
The autocorrelatjon?
function; its estimation and interpretation.
2.
Models for time series data.
?
Stationary processes, autoregressive-
moving-average processes, Gaussian processes.
3.
Statistical inference on
au
toregres3ive-rnov j
ngaverage processes.
Fitting a model to such a process.
?
Regression analysis with serially
• ?
correlated data.
4.
Introduction to spectral analysis and the search for cyclical fluctua-
tions.
5.
Prediction and Control.
Suggested Text:
?
?
Box, G.E.P. and Jenkins, G.M., Time Series Analysis:
forcasting and control. Holden-Day 1976.
0
SIMON FRASER UNIVERSITY
New Graduate
Course Proposal Form
CALIDAR
IEFO4ATION:
Deputaint:
4ATHF.MATICS ?
Course Number:
STAT
805-4
Title: Non-Parametric Statistics.
and Discrete Data Analysis
Description:
Order statistics, rank statistics, procedures based on the
empirical distribution function.
?
Asymptotic efficiencies. Goodness-of-fit.?
Contingency tables, log-linear models. Further topics.
Credit flours: 4
?
?
Vector:
4-0-0 ?
Prerequisite(s) if
any:
?
Math 372 &?
Math 473 or equivalent or permission of the instructor.
ENLMZNT AND SCHEDULING:
Estimated Enrolment: 4 When will the course first be offered: 85-3
110w often will the course be offered; Every second year
JUSTIFICATION:
See attached.
RHSOU*8i
Which Faculty member will normally teach the course: Stephens, Lockhart &
others.
What are the budgetary implications of mounting the course: 1/6 of a faculty
member.
Are there sufficient Library resources (append details): Yes, see attached.
Appended: a) Outline of the Course.
b) An indication of the competence of the Faculty member to give
the course.
C)
Library resources.
APPROVED: Dept. Graduate Studies Ctte:
?
Date;
Faculty Grad.
Studies Ctte: - ?
- Date:
Faculty:
?
Date:
Senate Grad. Studies Ctte: ?
- Date:
Senate: ?
Date:
.0
.
S
COVERING MEMO CONCERNING STAT 805-4: NON-PARAMETRIC STATISTICS AND DISCRETE
0
?
DATA ANALYSIS.
Justification:
This course is part of the proposed project option in Statistics.
?
Roth ?
of the fields covered by this course are Important to practising statis-
ticians.
?
Techniques in these areas are widely used -
particularly in the
?
social sciences, management science, biology and medical research.
Competence of faculty members:
Dr. Stephens has published numerous papers in this area and is an inter-
nationally recognized expert in the field of goodness-of-fit.
?
Dr. Lockhart
is currently working on problems in this area. The material covered in this
course is standard in the backgrounds of all the members of the Statistics
Group in the Department of Mathematics.
Library Resources:
The library is well supplied in this area and subscribes to all the major
statistical journals. ?
See attached memo from E. A. Weinstein.
40
1.
Ll
34.
STATISTICS 805-4?
Non-parametric Statistics and Discrete Data Analysis
?
COURSE OUTLINE
Non-parametric statistics is the area of statistical analysis especially
adapted
to situation whenever the assumptions of classical statatical analysis
are likely, or known, to be false; for example, the often-used assumption of
normality, exponentiality, or independence
may be relaxed. This field has
been
much
developed in recent years,
particularly with availability of com-
puters. The ranks of the observations are often used, and the problems of
distribution theory frequently become combinatoric.
?
Another important area
?
is model justification, that is, to decide if a model, classical or otherwise,
appears justified by
the data. ?
This is the area of goodness-of-fit, also
much developed in recent years.
?
Thirdly, discrete data, usually based on
counts, occurs often in statistics.
?
A major tool is contingency tables, and
there exist several techniques
for modelling and analyzing such tasks.
?
This
?
course covers these three major topics.
Content:
1.
Brief survey of undergraduate work in non-parametric statistics, the sign
test, the Mann Whitney test.
2.
Procedures based on ranks:
?
the Wilcoxon rank sum test, the Wilcoxon
signed rank test, Spearman
correlation.
3.
Distribution-free procedures and
goodness-of-fit.
?
Techniques based on
the empirical distribution.
4.
Other goodness-of-fit procedures: Chi-squared tests; regression tests.
5.
Properties of the above procedures: Asymptotic efficiency, etc.
6.
Contingency tables: exact tests and large sample approximations.
7.
The log-linear model.
S.
Loqit, probit and norinit analysis.
Suggested Texts:
Plackett, R.C., The Analysis of Categorical Data, 2nd ed. MacMillan, 1981.
Cox, D.R., The Analysis of Binary Data. Methuen, 1970.
Fienberg, Stephen E., The Analysis of Cross-classified categorical data. 2nd
ed.,
MIT press, 1980.
D'Agostino,
R.
and Stephens, M.A., Goodness-of-fit Techniques. (To appear,
Conover,
Marcel Dekker).W.J.,
?
Practical Non-parametric Statistics, Wiley, 1971.
0
.
Date:
Date:
Date:
Date:
Date: ?
_____
APPROVED: Dept. Graduate Studies Ctte:
Faculty Grad. Studies Ctte:
Faculty:
Senate Grad. Studies Ctte:
Senate:
SIPON FRASER UNIVERSITY ?
New Graduate Course Proposal Form
C
?
CALENDAR INFORMATION:
Department:
MATHEMATICS
?
Course Number:
STAT 811-0
Title:
Statistical Consulting I
Description:
Students will participate in the department statistical
consulting service under the direction of faculty members.
This course will be graded on a Satisfactory/Unsatisfactory basis.
Credit Hours: 0
?
Vector:
0-0-2 ?
Prerequisite(s) if any:
This coursil
is open to M.Sc. and Ph.D. students in Statistics.
ENROLMENT AND SCHEDULING:
Estimated Enrolment: 5
?
When will the course first be offered: 84-3, if
possible
How often will the course be offered: Every semester (3 times per year).
Initially possibly only Fall & Spring
JUSTIFICATION:
C
?
See covering memo.
Which Faculty member will normally teach the course: Eaves, Lockhart,
Routledge, Stephens, Villegas & Weldon
What are the budgetary implications of mounting the course: Initially none.
Eventually 1 faculty member will probably be required for STAT 811-0 and SPAT
812-0 and generally to run the consulting service.
Are there sufficient Library resources (append details): Yes, see attachment.
Appended: a) Outline of the Course.
b)
An indication of the competence of the Faculty member to give
the course.
c)
Library resources.
COURSE PROPOSAL
FOR
StAT 81-0: STATISTICAL CONSULTING I
Justification:
Participation
in a consulting service is an essential part of the trai-
ning of
a statistician.
?
This course is an integral part of the project
?
option in statistics.
Competence of Faculty
Members:
Statistical consulting is an important part of any academic statis-
tician's background.
?
All of the statisticians in the Mathematics Department
have substantial experience of this type.
?
The statistics group has been run-
?
ning a consulting service for the university for several years now.
Library Resources:
This course needs the same library resources academic statisticians need
for their work at all
times. ?
See
attached memo from E.
A. Weinstein.
.
0
.1.
STATISTICS 811-0
.
?
STATISTICAL CONSULTING I
OUTLINE OF COURSE
Students will be expected to participate in the analysis of problems
brought to the consulting service from start to finish. This means they will
be present when problems are first brought In; they will help in the formali-
zation of these problems and selection of appropriate techniques; they will
carry out analyses and they will participate in the interpretation of the re-
sults. ?
As they gain experience they will play larger and larger parts in the
process. ?
We expect that, in the second semester of consulting, they will
?
find a problem suitable for the basis of the M.Sc. project.
SIMON
FRASER
'UNIVERSITY
New Graduate Course Proposal Form
?
.
CAIJJDAR INFOIATION:
Departments
MATHEMATICS ?
Course
Number: STAT 812-0
Title: Statistical
Consulting II
Description:
Students will participate in the department statistical
consulting service under the direction of faculty members. This course will be
graded on a satisfactory/unsatisfactory basis.
Credit Hours:
0 ?
Vector:
0-0-2 ?
Prerequisite(s)
if any: This course
is open to M.Sc. and Ph.D. students in Statistics.
£NROLMENT AND SCHEDULING:
Estimated Enrolment: 5 ?
When will the course first be offered: 85-1
110w often will the course be offered: Every semester (3 times per year).
Initially possibly only Fall and Spring.
JUSTIFICATION:
?
.
See covering memo.
RESOURCES:
Which Faculty member will normally teach the course: Eaves, Lockhart,
Routledge, Stephens, Villegas & Weldon
What are the budgetary implications of mounting the course: Initially none.
Eventually 1 faculty member will probably be required for STAT 811-0 and STAT
812-0 and generally to run the Consulting Service.
Are there sufficient Library resources (append details): Yes, see attachment.
Appended: a) Outline of the Course.
b) An indication of the competence of the Faculty member to give
the course.
C)
Library resources.
APPROVED: Dept. Graduate Studies Ctte:
Faculty Grad. Studies Ctte:
Faculty:
Senate Grad. Studies Ctte:
Senate:
Date:
Date:
Date:
Date:
Date:
COURSE PROPOSAL FOR STAT 812-0: STATISTICAL CONSULTu,G II
.
??
-__ ?
Justification:
Participation in a consulting service is an essential part of the trai-
ning of a statistician. This course is an integral part of the project
option in statistics.
Competence of Faculty Members:
Statistical consulting is an important part of any academic statis-
tician'sbackground.
?
All of the statisticians in the Mathematics Department
have substantial experience of this type.
?
The statistics group has been run-
?
ning a consulting service for the university for
several
years now.
Library Resources:
This course needs the same library resources academic statisticians need
for their work at all times.
?
See attached memo from 9. A. Weinstein.
0
I;
STATISTICS 812-0
STATISTICAL CONSULTING II?
OUTLINE OF COURSE
Students will be expected to participate in the analysis of problems
brought to the consulting service from start to finish. This means they will
be present when problems are first brought in; they will help in the formali-
zation of these problems and selection of appropriate techniques; they will
carry out analyses and they will participate in the interpretation of the re-
sults. ?
As they gain experience they will play larger and larger parts in the
process. ?
We expect that, in the second semester of consulting, they will
?
find a problem suitable for the basis of the M.Sc. project.
S
10
-
?
.
?
L . ?
S... • ?
•
42.
Prof, J, F. Cochran
?
horn ?
E. A.
Weinstein
Dean of Science
?
Library
-.Sciences
Subled ?
Library Support for Applied
?
Db!C
?
82/10/22
Probability and Statistics
The prupocd expanded graduate program In applied probability
and statistics consists of twelve (12) new coure proposals,
I have examined these proposals and find no new 3rcas of
research or teaching therein, but rather cxtensio:s of areas
of long-time standing at S.F.U.
The
library collection In the areas of mathematical statistics,
probability
and
numerical analysis
are
currently complete and
have been for some years. The older literature in
support of
our
collection
is readily available from
U.B.C. and the
National
Science
Library.
Absolutely no additional funds will be needed
•
?
by the Library In support of this program.
all
c.c. Prof. C. Viliegas
Dept. of Mathematics
I/•
cL1
El
_:.(han..yOnx
SIMON FRASER UNIVERSITY
New Graduate Course Proposal Form
CALDAR INPOI*4ATION:
Department:
MATHEMATICS
?
Course Number: STAT 890-4
Title: Selected Topics in Statistics
Description:
Credit sours: 4
?
Vector: 4-0-0
?
Prerequisite(s) if any:
ENII' AND SREDTJLING:
Estimated Enrolment: 5
?
When will the course first he offered: N/A
How often will the course be offered: Once a year
•
JUSTIFICATION:
This course is replacing MATH 877-4 (Statistics: Selected Topics).
RESOURCES:
Which Faculty member will normally teach the course: Eaves, Lockhart,
Routledge, Stephens, Villegas & Weldon
What are the budgetary implications of mounting the course: None.
Are there sufficient Library resources (append details): Yes
Appended: a) Outline of the Course.
b)
An indication of the competence of the Faculty member to give
the course.
c)
Library resources.
APPROVED: Dept. Graduate Studies Ctte:
?
Date:
Faculty Grad. Studies Ctte:
?
Date: ?
-
Faculty:
?
Date: ?
-
Senate Grad. Studies Ctte:
?
Date:
Senate:
?
Date:
41.
APPENDIX II:
New Calendar Fntry
S
?
M.SC. PROGRAM
?
DEGREE REQUI R4XNTS
A candidate for the M.Sc. degree will normally be required to obtain a
total of at least 20 semester hours of credit for coursework beyond courses
taken for the Bachelor's degree. ?
Of these 20 hours, at least 12 are to be in
graduate courses or graduate seminars, and the remaining 8 may be chosen from
graduate courses or graduate seminars or 400-level undergraduate courses.
He/she will also be required to submit a satisfactory thesis and to attend an
oral examination based on that thesis and r
elated
topics.
oject
Option in Statistics
Students interested in pursuing an M.Sc. program in Statistics may follow
project option.
?
The project option is intended to give students Instruc-
?
Lon in a wide range of statistical techniques and also to provide experience
the practical application of statistics. ?
The program should be of inte-
?
st to students who wish to acquire statistical expertise.
Students in the program will be required:
(i) ?
To
complete at least 28 semester hours of credit for
coursework in
Statistics and related fields beyond courses taken for the Bache-
lor's degree. ?
Of these 28 hours, at least 20 are to be in gra-
duate courses or graduate seminars, and the remaining 8 may be
chosen from graduate courses or graduate seminars
or 400-level un-
dergraduate courses.
?
Normally these courses will include STAT
801-4 (Mathematical Statistics), STAT 802-4
(Multivariate Analysis)
and at least one of STAT 803-4 (Data Analysis), and STAT 804-4
(Time Series Analysis).
(ii)
To complete satisfactorily STAT 811-0 (Statistical Consulting
I)
and STAT 812-0 (Statistical Consulting II).
(iii) To
submit and defend successfully a project based on some problem
of statistical analysis (as outlined in the GENERAL REGULATIONS).
This problem will ordinarily arise out of the statistical consul-
ting service.
students with backgrounds in other disciplines, or with inadequate back-
ground in statistics, may be required to take certain undergraduate courses in
the Department in addition to the above requirements.
PH.D. PROGRAM
A candidate for the Ph.D. degree will generally be required to obtain at
least 28 semester hours of credit for coursework beyond courses taken for the
Bachelor's degree.
?
Of these 28 hours, at least 16 are to be in graduate
courses or graduate seminars or 400-level undergraduate courses.
?
The course-
.
44.
work
?
in
?
all ?
cas'
?
will ?
involve ?
study ?
in ?
at ?
least ?
four ?
different ?
areas
?
of
math
e
matics,
at toast one of
?
these areas being
?
in Analysis
?
(Applied, ?
Complex
or ?
Real).
?
Applied ?
Analysis ?
is ?
understood
?
to
?
include ?
Differential ?
and
Integral Equations and Methods
of Applied Mathematics.
Candidates for the Ph.D. degree will normally be required to pass a gene-
ral
examination. ?
At the option of the candidate's Committee, this shall con-
sist of either ?
written papers
?
in
?
four
?
areas of ?
study, ?
or
?
written papers
?
in
three areas of
?
study,
?
and graduate coursework
?
in an approved fourth area
?
in
which a grade of
?
'B' or higher is obtained.
?
In either case, one of the writ-
ten exam areas must be
?
in Analysis ?
(Applied, Complex or Real).
?
A candidate
ordinarily ?
will ?
not ?
be
?
allowed
?
to
?
take
?
the
?
general ?
examination ?
more ?
than
twice.
?
Students will be interviewed and advised regarding appropriate cour-
ses ?
and ?
examination
?
curricula. ?
Course ?
programs
?
and
?
examination ?
programs
require Departmental approval.
Students who have completed a Master's degree will normally be required
to attempt the general examination within one year of their initial registra-
tion in the Ph.!).
?
program.
candidate ?
for
?
the
?
Ph.D. ?
may ?
be
?
required by his/her ?
Supervisory Com-
mittee to acquire proficiency in reading mathematical papers in either French,
German or Russian.
Students ?
will
?
be
?
required ?
to
?
submit ?
and ?
successfully ?
defend ?
a ?
thesis
which will embody a significant contribution to mathematical knowledge.
Pot further information and regulations, refer to the General Regulations
section (1).
0
\\.
•Jsr'.4 ?
:
January 26, 1984
Professor David Brillinger
tk?T)artlmnt of StitiFtics
Uversity of California
Lorkt. 1
.ey, California
94720
Dear Professor Brillirer:
'hank you for
af j
reeir
y
3 to serve as an assessor of the proposed
Jc. St.atitics PrcJraJTrTe.
;u onorariwi
id by
this University for such services is
• ?
$200. I enclose a copy of the proposal, a copy of the University
C adui
to
Pxjula
t
i xs
gcJvernin1 all graduate students at Simon
Fraser, and a description of the proposed prcxrartr.
The Assesmnt Committee would appreciate your frank a..rnints
on the academic merit and suitability of the proposed program.
Please include in your report answers to the following questions:
1.
Is the ava i lable academic expertise (see attached curricula
vitae) sufficient to iiupluiunt the prograsmc?
2. Da
you think that graduates of the projrarrre will be of quality
('(A1rabie
to those producc.1 at the leairrj institutions in the
field?
3.
Hcjw large is the need for tJx graduates that this pr(xJrazTm
would procce and is it a continuing need?
4.
is the iarticular prcxjranne 'rojsed likely to rrcet the stated
objectives?
You sc&i1d note that these questions are not meant to limit the
range of your crncnts in any way.
/2
0
Profsor David I3ril1irjer
?
-2-
?
January 26, 1984
FurthcrTore, it would be most helpful if you could
make, in
addition to yir
other cximnts,
specific
rectxrmer1ations
on
either
the approval,
modification, delay, or disapproval of the
programre.
Your
report will be nwie available,
upon request, to nnhers
of the Ccxmiittees
and other
governing bodies both within and without
the University that must approve the progranr before it can be
izrplrcnted.
It would be appreciated if you could see your way to responding
within the next couple of.eks.
You sincerely,
all
John. ?ebster, Associate Vice-'..
President, Adcnic, and
Dean of
Graduate Stutlies.
'die
.\tcmnts ?
.
0
.
UNJ\'LflSITY OF CALIFORNIA. BEflKEI .EY
•
I,.v,s
•
IIll'E •
I 0
1
. A1.4.1 I
I. • III1.ItSIl)I
• 4A ?
11111.40 ?
%A ?
NI
•
I'4
0 ?
..,.
Au
rA (Pfl./.
TELEP)ION: (415)&42
- 27
091
?
1flI'AI4IMEJ1 OF STAIJ.rJCS
CABLE: UNIVCALB
?
ftEI(EI.EY, CALIFORNIA 94720
29 February 1984
John I. Weber
Dean of Graduate Studies
Simon Fraser University
Burnaby, B.C.
Canada
V5A
1S6
Dear Dean Weber,
Enclosed please find my assessment of the proposed M.Sc. Statistics
Programme. I think that it is a fine programme, one likely to contribute
measurably to the stock of statistical workers in your province and
country.
Yours sincerely
b0D
?
.
David R. Brilling'r
Professor
E7s41
OFFICE
l'
JA
0
uNlvEltsrrY OF CALIFORNIA. BEBKELEY
MM*Y*.IY flAYJ
?
JPWII4Y W'i
AtEI
YI
I4t%L*fl* SAN IflVA
0 S
AN
Y$A7.(
I'( ?
S1rA IIARIIAHJ. SANTA
CRLF
TILLUP110"51. (41S)6423751
?
or
STATISTICS
cAsI.E: WIVCAI.I3
?
I3KIIKFI.FY, CAIIFONIA 94720
29 February 1984
Assessment of Proposed .Sc. Statistics Programme
at Simon Fraser University
David R. Brillinger
The Proposal' Cut to bare details this programme
ir
seen to involve:
i). 28 semester hours of coursework with at least 20 hours at the graduate
evel,
ii)o
participation in
consulting
course for
two semesters,
iii).
submission and
defence of a
project,
to be later lodged in the
Library,
iv).
residence for 4 semesters,
v).
early hiring of 2/3
faculty member,
vi)* later hiring of a further faculty member,
vii)
.
initiation of various new courses'
Masters Degrees in Statistics' The M.A. and .Sc. degrees in Statistics
have two fundamental roles: a) as preliminaries to doctoral work and t) as
terminal preparation for applied work. The doctoral work
may
be in Statistics,
or
some
other field such as Economics or Education
.
When a thesis or project
is involved, it
is meant
to lay
the foundation for later independent work.
The students may be recent Bachelors or individuals with work experience
S
seeking further advanced training. Those obtaining the degree go on to
positions in education, industry, ar3culturc, government, health, business
and other professions.
Comments on the Present Proposal
The programme under consideration has all the characteristics just
listed for masters degrees and in addition is innovative in several repeote.
Its academic merit is high. It incorporates a close correlation between
theory and application as is vital for Statistics and recognises that good
work in applied statistics nearly always calls for more than mechanical
application of prepackaged methods. It recognises, further, that the amount
of statistics subject matter is now vast.
Answerc to some Questions:
1.
"Is the available academic expertise ... sufficient i..?' The calibre
of the current faculty is fine and their areas of interest are appropriate.
The new courses would give the programme character. As indicated in
the
proposal, these could be reading courses to begin; however early hiring
of a new faculty member (junior level) seems in order.
2. "... graduates
..e
quality comporable to •.. leading institutions •..V'
Yes, because of the substantial personal attention and supervision they
would experience; however I make one proviso. I do not know how contemporary
computing facilities are at SPU - these days the leading institutions
produce graduates with real experience of statistical computing.
30 "...
need for the graduates ...?" The demand for statisticians with
censulting and computing skills is dramatically high already and increasing
rapidly .
Some of the fields have been listed above. Individuals with ?'asters
or
- , -
decrees
are probably in greatest demand.
4 .
"...
programme ... to meet the ... objectives?" Yes, providing the
requested courses and staffing are approved.
Recomondation
I recommend that the indicated programme be approved in its present
form.
In
the attached Appendix, I indicate a few sugestione for canaideration.
Professor
.
0
-4--
Appendix
10
?
I • I wonder if STAT 802 is the best choice for the compulsory course
after STAT 801 . Perhaps, a somewhat modified, STAT 803 would be better.
II. Perhaps credit should be given for STAT 811, 312 • They will have
C.
?
grades and so would not count towards the 20 required hours. Giving
credit will allow other institutions to asses the amount of work involved,
Ph.D. students might chose to attend, so too might students from other
given to
Departments. Further, consideration mght be/'.aying the students
for
their
work here. (At Berkeley the Graduate Division does pay the students in a
comparable course about $150. per semester. Further, at Berkeley the faculty
member responsible for the course receives teaching credit for it.)
III.
Perhaps 1 nisued it, but I am not sure whether all the courses
are meant to be those of the Mathematics Department. If outside department
courses are to be allowed, I would
4uest
that it
is imperative thT.t any
student's programme be approved in advanc'
by ?
th'. Craduati. ?
Advisor.
lv. It would appear that the enrolirients in the ('very—other-year courses
would be higher, as they could contain a mix of first and seond year students.
V. I would expect the Consulting Service as well as providing educational
opportunities for the students, to provide research motivation for the
faculty. it will help students and faculty of many departments. it would
be in SFU'e interest to support it to the maximum extent posuible.
..-
University of Waterloo
V.i I
p r
fin), Ontario, (
.in,
N21
f(;I
Faculty of Mathematics
Department of Statistics
and Actuarial Science
519885 1211
March 1, 1984
Professor John H. Webster
Dean of Graduate Studies
Simon Fraser
University
Burnaby, B.C. V5A ]S6
Dear Professor Webster:
I have examined the proposal for an M.Sc.
programme in Statistics with project option at Simon Fraser
University. In my opinion, this is a strong proposal which
should
be approved. My report is enclosed.
Please let
me know if I can be of any further
assistance.
Sincerely yours,
mes G. Kalbfleisch
JGK:mk
?
Professor
End.
114R
61984
DEAN
OF
GRADUATZ
I
STUDIES OFFICE
.
Report on the proposed
M.Sc. in Statistics - Project Option
?
Department of Mathematics
Simon Fraser University
El
Prepared by
James G. Kalbfleisch, Professor
?
Department of Statistics and Actuarial Science
?
University of Waterloo
March 1, 1984
0
Report on M.Sc. in Statistics -
Project
Option
?
General Comments
A successful applied statistician must have
(].) basic mathematical skills so that theory can be
understood and modified as necessary;
(2)
a thorough grounding in statistical principles and
a wide variety of statistical methodologies;
(3)
familiarity with computers and their uses;
(4)
the ability to communicate effectively with clients,
so that problems are properly formulated and results
are communicated in the language of the client.
Beginning graduate students in Statistics are almost always
deficient in areas (2) and (4) , and depending upon their
backgrounds, they may also require additional work in (1) and
(3)
It is practically impossible to cover all of these areas in a
masters programme which involves a substantial thesis and a
limited amount of coursewórk. As a result, a coursework
masters with project, such as that proposed at Simon Fraser,
has become standard at many universities. it can be argued
that this sort of programme is preferable even for those who
will continue to the Ph.D. and an academic career, since they
will undoubtedly be involved in consulting and service
teaching to nob-mathematicians.
At the University of Waterloo, we have about 20 masters
students in Statistics. Almost without exception, they are
advised to choose a coursework masters with project. The
situation is similar at other Ontario universities with which
I am familiar.
Anwers. to Dean Webster's Questions
(1)
Is the available academic expertise sufficient to
implement the programme?
Yes. Simon Fraser University is fortunate in having an
active and well-qualified group of faculty members in
Statistics. My only concern is with the extra workload which
this programme will create. If this programme is approved,
and I believe that it should be, then high priority should be
given to hiring an additional faculty member in Statistics.
(2)
Do you think that graduates of the programme will be
of quality comparable to those produced at leading
institutions in the field?
Yes. Given the quality of the faculty, the provision of
hands-on consulting experience, and the variety of areas
covered in the proposed course offerings, graduates will be
well prepared for careers in government or industry, or for
Ph.D. work in Statistics.
(3)
How large is the need for graduates that this programme
would produce and is it a continuing need?
Our experience at Waterloo has been that there is a strong
demand for Masters graduates in Statistics. A recent
manpower survey in the U.S. identified Statistics and
Computer Science as areas in
which the projected number of
job openings at all levels greatly exceeds the expected
number of graduates. There is no reason to suppose that the
situation in Canada will differ markedly from that in the
U.S. Most of the job openings will be for graduates with
expertise in statistical applications and computing. This is
precisely
the sort of graduate which the proposed programme
aims to produce.
(4)
Is the particular programme proposed likely to meet
the stated objectives?
Yes. Consulting experience plus courses such as Stat 803-4
and 804-4 will expose students to statistical applications
and give them a greater breadth of training. Preparation and
defence of the project will help them to develop important
communication skills.
Recommendation
I strongly recommend that the proposed M.Sc. in Statistics
with project option be approved, and that high priority be
given to the hiring of an additional faculty member in
Statistics.
Additional Comments
It is commendable that Statistics faculty members provide a
consulting service to the university community. Undoubtedly
they feel that it is worthwhile for their own professional
development as well as for the assistance it provides to
others on campus.
If graduate students are to be involved, it will be necessary
to formalize and further develop the consulting service, and
the consulting service is likely to develop into a major
activity of the Statistics group. They should receive
recognition, perhaps in the form of reduced teaching loads,
for providing this service. In addition, they should seek
funding for support staff from the University, and from
external agencies such as N.S.E.R.C.
James G. Kalbfleisch,
University of Waterloo,
March 1, 1984.
The University of Western Ontario
Department of Statistical and Actuarial Sciences
Faculty of Science, Room 3005 EMSc
London, Canada
NSA 569
1984 02 23
Dr. John M. Webster
Associate Vice President, Academic
Office of the Dean of Graduate Studies
Simon Fraser University
Burnaby, B. C.
V5A lS6
Dear Dr. Webster:
I am enclosing my assessment of the M.Sc. Statistics Program: Statistics
Option proposed by the Department of Mathematics. My assessment is that
the plan is timely and that the faculty, particularly augmented as
suggested in the proposal, is entirely suitable and certainly capable of
implementing this program.
I would be most pleased to respond to a request if it was felt that I
could provide further information.
Cord tally yours,
Ian B. MacNeil]
Professor and Chairman
:j
FEB 27ftj
oT
0
.
A Critique Of The Proposal For
Master of Science in Statistics:
.
?
Project Option
Submitted by the
Department of Mathematics
Simon Fraser University
February 20, 1984
Ian 8.' MacNeill
Chairman and Professor
Department of Statistical
and Actuarial Sciences
and Director of Statlab
University of Western Ontario
0
I ?
I'.aJ'o:.td.
The ?
st.t i.t
icians
in ?
the ?
Department ?
of
?
Mat hernaric.s ?
propose ?
a ?
new ?
M.Sc.
program ?
In
Statistics
differing ?
From ?
the ?
tr;iditional
in ?
two ?
respects;
first, ?
more Lours
work
?
is ?
to ?
be
?
required; ?
and ?
second,
In place of ?
writing
a
?
thesis,
a ?
project
is
?
to ?
be carried
?
out
?
analysing ?
data
acquired under the
aegis ?
of the ?
statistical
consulting
?
service
?
(SCS). ?
It
Is ?
expected ?
that
graduates
from this
program will possess skills useful ?
In many ?
fields.
11 ASSESSMENT (Answers to questions posed In the letter accompanying the
appraisal documentation).
Q. is the available academic expertise sufficient to implement the program?
A. The statistical group at SF11 is very capable, and is recognized as such
by the NSERC Operating Grants Committee. They possess fine publication
records, and have demonstrated substantial interest and ability In the
application of statistical methodology. The group possesses expertise
in the area of statistical computing. Additional strength
in
the area
of sample survey design and analysis might be provided in future
additions to the faculty complement.
Q. Do you think the graduates of the program will be of quality comparable
to those produced at the leading institutions in the field?
A. Students graduating from this program will have been exposed to the
kind of training being required widely by employers of statisticians;
provided admission standards are suitably high, the quality of the
graduates should be more than competitive. Employers are asking for
demonstrated ability to apply statistics and to use the growing array
of statistical software. if such a program is not instituted, SF11 may
have difficulty attracting students to their MSc' programs in statistics.
Q
.
How large is the need for the graduates that this program would produce
and is it a continuing need?
A. It is my assessment that there is a large and growing need for masters-
level graduates with both interest and aptitude for applications and
with computing ability.
Q
.
is the program likely to meet the stated objectives?
A. The stated objective is, "to provide an alternative training plan
for Master's students in Statistics...". This objective certainly
ill be met. However, implicit in the proposal is the objective
or training students who will be effective applied statisticians
2.
capable of independent work in a variety of I ic'lds. It is my ;sessncnt
. that the program is highly likely to achieve this objective
,
provided
the quality of the students drafted into the program is high, and
provided the flow of projects into the SCS is sufficient to permit
each student to work on a meaningful project. The latter point may be
a major problem if the enrolment stated on page 6 materializes because
it Is likely that only a fraction of the projects coming to the SCS
will be suitable as MSc projects.
III DISCUSSiON
A. Rationale
The proposal makes the case that the approach to Masters-level graduate
education typically followed in Mathematics Departments may not always
be appropriate for training statisticians. This is so since statistics
Is a discipline with unique character among the mathematical sciences,
the uniqueness being determined by the role statistics can play in
scientific inference, in fact, many of the early contributors to
statistical methodology, including Fisher and Youdt
.
n, were active partic-
ipants in the process of scientific inference. The route to statistics
• through the world of mathematical deduction does not fit into this
tradition. Anyone who has taught science students knows that for t hem
concrete ex;implvs provide much clearer "proof" than ah;t rat t
argI'rru'nr
5.
This is simply a rc'l lcction of the fact that experimental verification
is the scientist's route to the truth. Thus, the science student is not
generally disturbed by the fact only one of a number of cases has been
settled, because that is how science often advances. The mathematician,
however, has less concern with the slow unfolding of imperfect knowledge
and is less interested in the untidiness of scientific reality than is
a statistician. Although this dichotomy is somewhat of an oversimplifi
-
cation, it illustrates a source of much missed communication between
the scientist and either the mathematician or the statistician who
emerges from a mathematical milieu. Students of statistics often iced
to be reminded of the role of statistics and statisticians in science;
they should think of themselves as statistical scientists rather than
as mathematicians. Indeed, statistics provides philosophical bases and
the technology of the scientific method, and has found application in
many new scientific fields, with the consequence that statistics has
become, within relatively recent times, the matrix in which are set the
biological, social and, to a certain extent, the physical and engineer-
ing sciences.
3.
Part of the reason for thi
,
proliferation f O
l
e "s' of st at 1st Ical
models has been the availability of increasingly sophist1<..,rd ((tnputlng
hardware and software. Without access to the appropriate computing
resources, certain statistical tools, such as spectral methods for time
series and graphical methods for multi-variate data, could be studied
only in the abstract. Modern computing facilities now make routine what
was once a curio.
B. Statistical Consulting Service
To be effective, the proposed M.Sc. program must have available a good
source of real statistical projects. I believe this will be accom-
plished by providing consistent and effective service in the areas of
statistical consulting and statistical computing. This requires exten-
sive facilities, operated by well-trained and experienced staff, and
effectively organized so that clients may be dealt with in a busncs:.-
like fashion. I have noted at Western that it is not sufficient to be
able only to provide to clients the experimental design at the beginning
of a study and the statistical analysis of the data at the end. If one
Is unable to provide data-management resources in the inter.iices
between design and analysis one may never see the client, either for
design or for analysis; instead he may go to the computing spc'ciallst
who can give advice on database management systems but who then may
lead the client up some rather exotic path which, often as not,
terminates in an expensive set of data which cannot be analysed to
answer the questions originally posed. Important roles that the
consulting service operated by the st.itisticlan at SFU
can
serve are
discussed below.
1. ?
Statistical Computing
The SCS can participa
t
e in the management of the interface
between statistics and computing. It -can' ensure that good
statistical software developed elsewhere will be purchased and
implemented on your systems. When need for a specific piece of
software arises, and none is available commercially, it can
press for local development. The SCS has, 1 believe, qualified
staff who can evaluate the algorithmic and mathematical perfor-
mance of statistical packages and subroutines. The SCS should
ensure that database management systems are available for the
bcnef it of research studies that produce large amounts of data.
] h. conput
ing
act
lvi t I es of
?
hc ?
vi ii
hcnc
l it
?
;raduat e
sen I or undergraduate s ude n
?
who •
?
h rough t he
I
r as %o(
i t
with SCS, will develop expertise in
.oniput
ing as
it
applies to
statistics. The statisticians have recognized that such exper-
tise will be required of all who graduate from their M.Sc.
program. Statisticians with such skills are presently In short
supply.
2. ?
Statistical Consulting
The major objective of SCS is the provision of a statistical
consulting service for researchers at SFIJ. Experience shows that
consultations can vary greatly in length and complexity, from a
one-hour discussion of basic statistical concepts, to a collabora-
tive research project. Furthermore, many kinds of expertise can
be called for in consulting work; there may be demand for
knowledgeable advice In such diverse areas as: design and analy-
sis of experiments, non-parametric methods, regre5sion analysis,
sample survey design and analysis, time series analysis and
multivariate analysis. Also, associated problems of data manage-
ment may arise. It should be the ultimate aim of the SC and the
SCS to accommodate this entire range, to the extent that resour-
0 ?
cc's permit.
Coping with such a range of problems is beyond the scope of any
single individual. A. client should have available to him the
services of a well-balanced team; it is Important that a sample
survey expert should design surveys, and that an expert in the
design and analysis of experiments should give advice in such
matters. It is also important that a consultant be backed-up by
other statisticians to whom he can turn for advice when difficult
matters arise. Another requirement is that good statistical
software and advice on the computer-processing of large data
sets be available to the consultant. In addition, it is helpful
if good support staff is available to handle the routine aspects
of consultations.
A SCS consultant can 'be backed-up by the faculty members whose
interests and abilities (over the main areas of special i7.;it ion
• ?
in scat 1st ics and statistical coniput ing.
?
In
add it Ion,
?
t
lie se
?
statisticians have the depth of training and experience c.o
enable them to fill ?
lacunae in their coverage. Also, chose
?
statisticians are aware of developments on the national and
5.
i rite rn.t )On.,l
?
',t it 1st ics .nd
?
stat 1st ic;il c cn,put iru
?
LflC5. ?
Each
has connections with the out side ;t at i f.t ic,i1 curnuniry which
enables him to obtain second opinions on almost any matter
concerning statistics and statistical computing. It would be a
difficult and expensive matter to duplicate this expertise and
these connections.
According to the proposal, statistical consulting and statistical
computing will be considered part of the training of M.Sc.
students in Statistics, which suggests each graduate student
will be asked to participate in the consulting and computing
aspects of SCS. This can serve a dual role. First, students can
be placed in an excellent learning environment. They can observe
experienced consultants at work, thus gaining, firsthand, know-
ledge that is difficult to come by in any other way. Also they
can learn to use the computer to analyze large data sets by
working on problems requiring this knowledge and by having
specialists close at hand who can give cogent advice when
required. The second benefit to accrue from having graduate
students exposed to the consulting program is that they them-
selves can contribute to the program. In the beginning, before
much learning has taken place, they can do necessary but routine
jobs that otherwise would take up the time of experienced
consultants. Later, as the graduate students gain confidence and
experience, they can be asked to assume greater responsibilities
in consultations. By graduation they should be competent consul-
tants in their own right.
C. Financial Considerations
The
discussion above implies that as the SCS evolves it may acquire
more resources. 1 have several suggestions as to how these acquisitions
may be financed. First, the SCS might apply for a NSERC Infrastructure
Grant. Given the high quality of the statistical faculty and their
applied interests, it seems entirely reasonable that such an applica-
tion should be successful. Also, the SCS might, consider applying for
NSERC
equipment grants to purchase such computing equipment as may not
likel4
be
available in a mathematics department; In particular I have
in rrvind graphics equipment. Another source of funds is the clients
themselves. At Western we try to charge all clients for consulting
0
6.
time of faculty, staff and graduate students. Our experience is that
. most clients have research funds that allow for such costs. Further-
more, graduate students who come for assistance from other departments
usually have supervisors with financial resources sufficient to pay
for the required assistance. However, we do not charge for initial
consultations, and anyone who cannot afford our rates is provided
nevertheless with service to the best of our abilities.
We
find that
advice and work that is paid for often is better appreciated than that
which is obtained free. it is my belief that statisticians in the SCS
should not be modest about charging for their consulting services;
what they have to offer is valuable, and the funds generated can be
helpful in embellishing their programs.
V Further Comments and questions
1.
The proposed new courses are important. However, it is not clear to me
where students will obtain instruction in sample surveys, experimental
design, and regression methods. I assume statistical computing will be
a significant part of Stat 811-0 and Stat 812-0.
2.
Other courses which could be considered for addition to the program
are: quality assurance sciences; survival analysis; forecasting; and
operational research methods.
3.
Will projects involve software development and/or implementation? Such
projects can have residual benefits for the SCS.
4.
Will students be exposed to database management systems and to statis-
tical computer graphics methods? Such exposure will be valuable for
job-seeking purposes.
5.
Taking students into the MSc program from such fields as Economics and
Biology is commendable (see page 5). It might prove to be necessary to
• ?
have them take a make-up semester to avoid unnecessary attrition in the
probability and mathematical statistics courses.
