Chapter I: Methods--Qualitative Techniques Aided by
Survey Research

     Summary: This thesis uses multiple methods.  Surveys
     of all 1993 and 1994 szu graduating seniors use a
     quantitative approach.  The rest of the study's data
     come from various documents and journal articles as
     well as extensive interviews with job-seekers and
     recent graduates.  Since qualitative research is so
     personally tied to the researcher, space is given
     for introducing the researcher.  Then, after
     presenting a general introduction to the qualitative
     method, the remainder of the chapter describes in
     detail the other research techniques and strategies
     used for this thesis.
     
Introducing the Researcher

In selecting a method, the researcher is influenced by
various factors: including his/her own academic
background as well as personal experience and preference,
the characteristics of the task, and the methods followed
by his/her supervisor(s).  Before explaining in detail
the methods used in this thesis, this chapter focuses on
why and how the method was chosen.  It begins by
presenting the researcher's personal experiences and
preferences and suggesting how these have affected his
method.
     At the University of Hong Kong (HKU) all post-
graduate research students in Education, at the
completion of their first year of enrolment, are required
to present a seminar before fellow students and
interested faculty.  The main purposes of this endeavour
are three-fold: to determine if the research design is
sufficiently workable and sound for the student to be
advanced to candidacy; to provide feedback to the student
who, up to that time, had probably received little input
from members of the faculty other than his/her immediate
supervisor(s); and to give students experience in
presenting ideas in public, a process which usually
clarifies their own thinking.  This seminar is extremely
important because, unlike in the American system where
students complete coursework, higher degree candidates at
HKU and other tertiary institutions following the British
model have for the most part only their theses to show
for their study.  Consequently, research post-graduates
such as myself are not exposed to a methods survey which
presents a cookbook, from which one may sample and
ultimately choose a favourite.  Instead, students rely on
their own education backgrounds (in many cases a Masters
in Education).  They develop their own methods, usually
heavily influenced by the techniques and strategies
commonly used by their supervisor(s).  The seminar
provides the one (and often only) opportunity for the
student to receive comments on his/her methodology from
academics who themselves may use different methods.
     A few days after my seminar, one of the faculty who
had attended my session confessed to me he was still
trying to understand the method I was using.  Admitting
he was himself a "numbers man," he said he was still
finding the area of qualitative research difficult to
define.  My response was, "So am I."
     I remarked at my seminar that I had been heavily
influenced (when I took course work for a Masters in City
Planning) by planners and economists who encouraged me to
view the world as a big regression equation, with
independent variables affecting changes in the dependent
variable.  Each profession has a point-of-view: lawyers
may see the world as a bundle of rights, medical doctors
may define it in terms of cures and diseases, physicists
as a world of natural laws, language teachers according
to style and structure.  Planners like myself see a world
run along a rational planning model, one which includes:
goals, data collection, evaluation and change.  I am most
comfortable talking about goals, about whether they are
achieved as evidenced by data, and about what changes
would be necessary in order to achieve one's stated
objectives.  As a resource often employed by planners,
econometric tools appeal to me.
     This study, however, does not involve complicated
mathematics; nor would econometricians recognize it as
one done by one of their peers (which, by the way, I am
not).  One's academic training and personal preference
for choosing one particular method over another plays, as
I have suggested, a role in one's choice of method.  But
an even more important consideration is whether the
method suits the data.  In the early stages of my
research I discovered that sophisticated statistical
methods do not suit the data of this study.  But the
story starts earlier.
     At the time I began contemplating post-graduate
study, I had been teaching at Shenzhen University for
several years.  Initially, I was hired in 1988 because
the school desired a certain number of Caucasian
expatriate staff for their "foreign presence."  I was
assigned to teach an oral English course.  Also known as
"conversation," this course was a core part of the
curriculum for English majors.  Expatriate teachers,
however, were given no materials and no direction from
school leaders.  After the first year, I convinced the
head of the Foreign Language Department that my services
would be better utilized in teaching a more rigorous
writing course which was then absent from the English
major sequence.  I also asked to be given the "Western
culture" course, which then existed but lacked academic
rigor and was little more than a foreigner talking about
life in the U.S.  Just as with the conversation course,
little was expected from me.  Not only was I given no
guidance and my performance never assessed, but I was
permitted to do whatever I wanted.  This freedom helped
me develop a personalized teaching style as I imported
methods and materials from the U.S.1  Thus, before I had
even contemplated writing a thesis, I was a participant
in education at Shenzhen University.
     In 1991 I took a six-month sabbatical from szu,
during which time I talked with educators and researchers
in the field of China studies and education.  During my
meetings and readings over this period, I found myself
more and more enticed by the idea of research, a field I
had abandoned a decade earlier, when I stopped working as
a policy analyst for the U.S. Government.  In my first
years at szu I had written several conference papers and
found certain issues of nagging concern--issues relating
to quality of education, student culture, teaching
methodology, and cultural influences in education.  As I
started to explore these issues, I was becoming a
"participant as observer."2  When I returned from
sabbatical, I continued a literature review in the
general areas of Chinese history, culture and education,
eventually narrowing my interests into a thesis proposal.
At this point, my modus operanti also changed.  I was now
moving toward becoming more observer than  participant.
Although I was still teaching and working on curriculum
reform, my eyes and ears had become more widely open.  I
was paying more attention, cultivating informants who
gave me information that I, a foreign teacher, would not
be privileged to in the normal course of events.  In
getting more information, I was also becoming aware of
many serious problems (academic, administrative,
political) facing the school.  During my last year, I
delivered papers at two conferences in China.3  The
audiences warmly received my comments, which were
somewhat critical of aspects of education at szu.  This
criticism reflected faculty-wide concerns about quality
of academics and educational management, and my
vocalizing them was, I suppose, the main reason the
school leaders decided not to assign me classes to teach,
effectively terminating my six years' employment.  In
sum, during my first several years of research for this
thesis, I was an "active participant."4  Then, when I
left szu, I became a "privileged observer"5 at least in
terms of my informants, the subjects of my study.  My
former host institution viewed me as an observer non
grata.
     In developing my research proposal, I knew I should
take advantage of my participation at szu and the
availability of data.  Generally, researchers based in
the West who come to China must either obtain a co-author
or else attach themselves to an institution which serves
as official sponsor.6  With the open-door policy of the
last decade, researchers from abroad have been able to
find (usually for payment) official sponsors and co-
authors without too much difficulty, depending of course
on the sensitivity of their research.  My already being
on staff at szu was the first step through this open
door.  In the jargon of those who write on qualitative
methods, I had "gained access."7  Since there had been
little research on szu, I suspected I was sitting on a
valuable reservoir of raw data. The question was how to
fish out the data, in other words what methods to employ.

General Introduction to Qualitative Method

The method employed in this thesis goes under several
names: naturalistic inquiry,8 educational ethnography,9
and qualitative design.10  Qualitative research strongly
influenced by ethnography has produced some valuable
studies in education,11 but only recently have
educational ethnographers studied Chinese schools.12  The
study at hand was made possible by the author's
participant-observation while serving in the role of
"foreign teacher" for six years at Shenzhen University,
an undergraduate institution in the prc.  Being in the
field over an extended period of time may allow the China
researcher to see life more as it really is, not just as
it is presented for "foreign guests."  It gives the
scholar sufficient time to develop personal connections
necessary for cultivating informants.  And it permits
critical analysis of the type of documents which too
often are taken by visiting academics at face value.
     The general descriptor for the type of research
undertaken in this thesis is qualitative.  In deference
to anthropologists, I will avoid using their term
"ethnographic."13  Unfortunately, qualitative is often a
responsive term, one used defensively to specifically
suggest non-quantitative.  For those who insist on
juxtaposing qualitative with quantitative, this term is
perhaps not altogether suitable.  Some qualitative
research might be referred to as "analysis by anecdote,"
where incidents from cases provide most of the
researcher's data.  The term anecdote itself is
pejorative among scholars, although eminent contributors
to knowledge--such as Plato, Freud and Piaget--used
personal stories as tools in their theory building
workshops.  Qualitative research methods in general meet
with scepticism on several fronts; indeed, the
qualitative/quantitative distinction may be a false
dichotomy.14  To many educators, data equate with
quantifiable configurations.  Case studies are permitted,
but they serve best in the company of similar cases which
together might allow for statements of generalization.
Qualitative studies that stand alone (in other words, are
not comparative in purpose) generally leave the issue of
generalizability up to the reader.  They present very
descriptive, original data which in themselves serve to
enlighten scholars.  Often, the qualitative approach is
inductive in the sense that data are collected (step 1)
and from these data, theories evolve (step 2).
Qualitative and quantitative are not mutually
exclusive.15  These methods are often, in fact, mutually
supportive.  This opinion seems not to be held by those I
call diehard qualitativists, who divide the world into
paradigms and have little positive to say about the
positivists.16  The warning "There are lies, damned lies
and statistics" suggests a public perception that numbers
alone do not tell the whole truth.  It seems to me that
the main differences between the two -tatives concern the
matter of dominance.  In quantitative studies,
observations, interviews and similar "soft" data are used
for support.  They do not dominate the research.  In
contrast, in qualitative studies, "hard" data (something
that computers and statistical software can generate)
serve the opposite function.  They usually support one's
observations but are not themselves the heart and soul of
the research.  Another role of quantitative data in
qualitative studies is to establish the environment.
These data, in the form of frequency distributions, are
descriptive.  In the current research, the data collected
by survey helped paint a picture of graduate employment.
Since little research had been done on job-seekers in
China, at the start I had no clear idea what the
environment looked like.  Moreover, fundamental
descriptive data on the students were lacking, such as
the socio-economic status of parents.  Few studies in
China illuminated the Shenzhen picture.  And, comparative
studies did not shed much light, either.
     The method employed in this study is greatly
influenced by ethnography.  This term is slightly more
definable than qualitative as it grows out of a rich
discipline, the anthropological tradition.17  Certainly,
my study is cast more comfortably in the shadow of the
works of Margaret Mead or Fei Xiaotong than the studies
of economists whose findings grow out of manipulations of
national income and employment data.  Ethnography is one
of the major methods of educational research.  Its main
emphasis is on participant-observation of the small scale
as a method, with an attempt to understand the culture
and symbolic lie of the actors involved (Masemann, 1986,
23).  A classic ethnographic study in education involved
an anthropologist studying an American high school
executive and becoming so much part of the environment
that he was known simply as the man in the principal's
office.18  In China, whether serving on a team of
visiting experts evaluating rural education19 or on the
teaching staff and environment of Shanghai high
schools,20 the educational ethnographer assesses the
environment of his/her study by living in that
environment.  In pointing out why he chooses the term
"ethnographic research" Cheng (1993) explains that it
"clearly indicates the data-theory relationship, the role
of the researcher, [and] the characteristics of research
design."  D.L. Lancy21says that "...ethnography is
characterized first and foremost by prolonged engagement
and participant observation."  The length of time I
served in China (I use the term served more in the
missionary than prisoner sense) cannot alone qualify me
as an ethnographer, but it does negate the accusation of
my being an imposter by using "blitzkrieg ethnography."22
     Data-theory relationships can be stratified along
the inductive/deductive lines.  Ethnographic studies
follow the former line with observation preceding theory.
"[E]ffective theory construction depends on both
inductive and deductive procedures," as P.J. Pelto & G.H.
Pelto23 point out, but ethnographers tend to collect
their data before theories are built, rather than vice-
versa.  This is partly due to the fact that theories are
not developed in vacuums; without previously acquired
data to describe the world which the researcher examines,
theory construction even in the natural sciences is
nearly impossible.  In the human sciences, it is
impossible.  The anthropologist, for example, lives in a
small community (a village, urban neighborhood, or
institution) and collects data, and a theory grows from
his/her experiences.  Many ethnographers go into the
field without a theory they are trying to prove.  In this
vein, my study did not commence with a theory.  The
subject I wanted to examine was data-scarce.  I had to
collect the most rudimentary data--such as on the types
of jobs graduates found--before I could begin to define
the role of relationships in the job procurement process.
Indeed, the hallowed and mandated order of the written
thesis: theory, method, findings, conclusion is at odds
with the thesis process for qualitative studies, which
might be more accurately chronicalized as: method,
findings, theory, conclusion.  One is asked "What is
theorized from the study" rather than "How did the
hypothesis test?"
     A significant concern of ethnographers is validity
and reliability.24  To this end, ethnographers are
advised to present not only their observations but also
the procedures they used.  That seems fair enough,
although some theorists point out that qualitative
research by its very nature places limits on reliability,
since most studies cannot be duplicated,25 whereas many
quantitative studies (in medicine, for example) can be
duplicated.  The value of the ethnographic study is not
in its replicability; but whether it yields insights.
The question of validity, however, cannot be discarded so
easily.  Indeed, it is important that researchers measure
what they think they are measuring.26  Yvonna S. Lincoln
& E.G. Guba27 discuss how researchers can establish
trustworthiness so that their findings and
interpretations will be found credible.  One of the most
obvious is to use multiple methods (often called
triangulation) for collecting data: interviews, review of
artifacts and documents, surveys or censuses.  Prolonged
engagement in the field doing persistent observation is
important.  Also, the researcher can search for negative
cases--ones that are exceptional and which shed light on
the inquiry.  Another way to increase trustworthiness is
to check back with the very people one studied.  Once I
had moved full time to Hong Kong University, my frequent
trips over the border to Shenzhen allowed me ready access
to my informants.  I often checked my findings with them
and these feedback conversations gave me a deeper
understanding of my findings.  Another strategy is peer
debriefing.  Unfortunately, given the small number of
full-time students enroled in the post-graduate program I
attended, the atomization of the intellectual community
and the paucity of peers engaging in my kind of study, I
experienced virtually no peer debriefing at HKU.  I have,
however, read about it,28 and it seems to provide an
excellent opportunity not only to check ones findings but
also to experience the collegiality that is supposed to
characterize the academy.

The Method as Interpreted for This Study

This section provides a detailed description of my
method.  It first looks at the data available before the
study.  Next, it deals with data collection through a
survey instrument.  The final part discusses my
interviews with students and graduates.

Data Available before Study

When I initiated my study, data on szu students and
graduates were rather scarce.  Admission records list the
names of students, their sex, the name and location of
their high school and their college entrance examination
scores.  The school maintains dossiers (dang'an) on
individual students, which presumably includes the
parents' addresses.  This information, as far as I can
determine, is not computerized in a way that lends itself
to database manipulation or statistical analysis.  Szu
does not have mass mailings to students.  Most contact
with students is done through their academic departments,
which handle such activities as fees collection.  Parents
of students who fail courses are notified by letter over
spring or summer vacation, but this is about the only
external communication the school has with parents of
students.29  All other correspondence is internal and
directed specifically to students, usually through class
monitors or in department or school-wide assemblies.
     Even less reliable information is available on
alumni.  During the school's first five years, a graduate-
run alumni association was active.  After several years
of data collection, in May 1990, it published an alumni
directory that included for each graduate: sex, hometown
(i.e., parents' home at admission), work address, work
phone, domicile, job title, and a memo regarding whether
the graduate was living abroad.  The directory's
publication was, perhaps, the terminal act of the
association.  After June 1989, the group, which was
legally incorporated and independent from the university,
virtually severed ties with the school and eventually
stopped functioning altogether.  The data in this
directory are substantially out-of-date.  In an attempt
to verify the information published on the graduates of
the Foreign Language Department, for examples, my
students discovered that fewer than one-fifth of the
addresses remained valid.
     After I began research, I met with the university
president on two occasions, in October 1992,  shortly
after he formally assumed office, and then a year later.
At both meetings, averaging two and a half hours each, I
apprised him of my research interests, advised him of the
importance of alumni relations to American universities
and offered my assistance in developing a viable alumni
organization and database.  I was referred to a committee
which he had appointed to encourage alumni participation
in the school's upcoming tenth anniversary celebration.
This group consisted of the heads of two academic
departments and two senior bureaucrats.  As one committee
member explained to me, the group was unable to
communicate with the alumni association "for political
reasons."  The committee sent out form letters to alumni
(based on the earlier directory's addresses) in an
attempt to pre-sell and market a �228 pictorial.30  In
the letter, which in fact was the first and only
correspondence between the school and its alumni,
graduates were offered the opportunity to buy a full-
page, colour advertisement to display their
entrepreneurial successes.  The cost was �8,000.  In the
published volume, ten graduates have half-page spreads in
a section entitled "Cradle of Entrepreneurs."  Rather
than collect and update data on alumni in a new
directory, this group decided to entrust alumni relations
to the individual academic departments.  As of Fall 1994,
no departments had prepared alumni directories.
     The school, through its vice-president in charge of
student affairs, assigned the freshman class the task of
surveying alumni during their 1994 summer break.  Each
department gave its freshman the names of several alumni
and asked them to conduct interviews and complete a
questionnaire.  This was part of the freshmen's "social
investigation," an activity outside course work which
comes under the rubric of moral/political education, what
elsewhere might be classified as civics.  Students were
paid �15 per interview.  The questionnaire used to
"investigate society" consisted of 25 closed response
questions (multiple choice) and 5 short-essay type
allowing open responses.  The form itself more resembled
a test than a survey instrument.  Many of the questions
solicit alumni opinion on either their alma mater or
their employment.  The wording of the questions on
employment cover many of the areas in my questionnaire
discussed below.  Questions 14-23 cover employment
process and preferences.
     I have chosen not to examine any data collected
through this instrument (a point mooted by the prospect
that I might not have been given access to the data
anyway).  This questionnaire appeared a year after my
first survey, too late to provide the type of descriptive
data helpful for the ethnographer at the early stages of
research.  More important, however, are the shortcoming
of the survey instrument itself, the distribution
procedure and resulting sampling bias.  Many of the
methodological problems with survey research in the prc
described by Stanley Rosen31 apply to this survey.  In
addition, the fact that students were given a financial
incentive (although a slight one) to turn in a completed
form raises doubts about the integrity of the data.  No
sampling procedure was employed.  Any alumnus/alumna who
could be located became eligible for the sample, leading
the responses to favour recent graduates and those who
had not changed jobs.  The questions themselves force the
subject to choose mutually exclusive categories which, in
reality, do not cover the whole range of possibilities
(e.g., questions 15, 25) and are not exclusive (e.g.,
questions 7, 10, 20).  Finally, the data were not coded
on the questionnaire, which leads me to suspect that the
returned instruments could not be analyzed with
statistical software.
     In sum, szu was unable to provide me information on
the mass of graduates.  The directory published in 1990
is outdated.  Any data collected by freshmen in the
summer of 1994 are suspect.  Despite my having two
interviews with the szu president on the advantages to
the school's establishing relations with alumni and
producing an alumni data base, no coordinated attempt to
collect data on alumni currently existed until Summer
1994, too late in the course of my research.  Any data I
needed I had to collect myself.

Survey

A general survey can help the ethnographer picture the
environment.  Without using much more than univariate
frequencies, survey data helps profile the subjects of
the study and enables the researcher to define areas
which can be used later to characterize subjects.  Using
attitudinal questions, it can give views on preferences
and perceptions.
     Each year after returning to school from spring
festival break,  szu seniors complete their graduation
theses and immediately hunt for jobs.  Most students are
working full time by the date of graduation, which is the
third week in June.  Over the two- day period before
graduation, seniors are processed out of school.  They
are given a form that must be certified (by red chop) by
more than a dozen departments.  For example, the library
verifies that students have no outstanding books or
fines; the housing office certifies as acceptable the
condition of dormitory furniture.  The academic
departments document that the student has completed all
requirements for graduation; other school authorities
verify that identification cards and railway discount
passes have been returned.
     After making several requests to the university
president and vice-presidents, I was given permission to
survey graduating seniors in order to obtain data on how
they found jobs.  I suggested that seniors complete a
questionnaire as part of the exit procedure described
above.  This request was not disapproved, although one
vice-president, however, objected to my asking
sociological and political questions.  Most of these,
which were not germane to this thesis's research
question, were removed.  He further disapproved of a
question concerning parents' CCP affiliation.  I had once
done a favour for one of the school's mid-level managers
and thus had established good relations with him.
Fortunately, when the disapproving vice-president was
hospitalized with a digestion ailment, my former
colleague secured the other vice-president's approval.  I
had the questionnaire privately printed, as none of the
officials I was dealing with wanted his office to pay the
�900 cost, and I was reluctant to pursue this issue as it
offered more opportunities for others to examine (and
delay) the survey.
     The Student Affairs Office director, whom I had met
only once in an unsuccessful attempt to get his office to
set up an alumni data base, agreed to have his staff
distribute the instruments to each academic department
with instructions that students complete and return them
during the exit procedure.  His favourable decision was
crucial and possibly influenced by three factors.  I
presented him with a signed note (kai tiaozi) from my
supportive vice-president requesting the lower official's
assistance.  Second, distributing the survey incurred
little time or effort of his staff.  Third, his approval
was a sort of compensation for his refusing my earlier
requests that his office get more involved in alumni
affairs.
     During the two days of exit procedure, I sat at the
academic affairs table and chopped students' forms after
they submitted a completed questionnaire.  I refused to
chop forms from those who had not completed the survey; I
accepted, however, all completed forms, even those in
which respondents had "straight lined" (e.g., "55555" or
"44444") attitude responses. (Straight line or other
patterned responses (e.g., "12341234") responses were
later recoded as missing data.)
     The instrument was originally composed in English
(Appendix A) and translated into Chinese (Appendix B),
the teaching medium of all departments in Shenzhen
University, except for Foreign Language.  It was "back-
translated" into English to ensure that the original
meanings survived translation.  The questionnaire was
then pre-tested on several graduating seniors and on six
alumni working in Shenzhen.  It became apparent that few
students had ever been exposed to a survey instrument.
Consequently, a revision included a more fully worded
clause promising confidentiality, as well as additional
instructions on how to respond to a Likert-type five
point scale for attitude questions, and specific
instructions to "mark only one choice" for several
questions.
     Over ninety percent of graduating seniors completed
the questionnaire, making the "survey" more of a census.
Although seniors have until several weeks after
graduation to submit the chopped form, most seem eager to
conclude their obligations with the university as
immediately as possible.  The completed responses
included 549 in the baccalaureate track and 239 in junior
college.  No bias appears to characterize the five
percent who failed to complete their forms before
graduation day.  Some may have returned to their
hometowns, in which case they are not working in Shenzhen
and not part of my study.
     The instrument is composed of 129 variables, of
which 35 are attitude questions.  The responses had to be
coded, then readied for the computer.  Six students did
these tasks.  Like the freshmen discussed earlier,
Shenzhen University's sophomores and juniors are required
to undertake two weeks of "social investigation/practice"
(shihui shijian) during summer vacation.  They must give
the university a form chopped and certified by a work-
unit.  Many work-units in Shenzhen do not want to have
szu students hanging around their offices, and many
students cannot get their forms chopped.  Thus, this
requirement has become subject to broad interpretation.
My department head permitted me to use sophomore students
for coding if I could find a Shenzhen employer who would
chop their forms (thus relieving the department of such
an obligation).  This task was handled by several of my
former students who had access to chops or were able to
get their fathers to chop the forms.  In total, my six
sophomore students took about 300 hours to code the data
and ready it for "keypunching."  I input the data myself,
and thus was able to correct some coding errors.
     The following graduation, in June 1994, I composed a
greatly shortened questionnaire (Appendix C) for
graduating seniors.  Having gone through the process
described above the year before, again I had the
questionnaires printed privately.  I did not seek
approval, partly because an administrative reshuffle had
relocated all the staff who had been so helpful the year
before.  Thus, out of fear that the newly appointed vice-
president in charge of student affairs and others would
not approve, I simply deposited the questionnaires with
the worker in the Student Affairs Office who had been
involved in the previous effort and hoped for the best.
During the exit procedure, I occasionally visited but
turned the entire process to the staff from academic
affairs.  Again, I feared that my open involvement might
prompt a veto among newly appointed officials.  At that
time in the midst of contract negotiations, I did not
want the completed instruments to be held hostage by
school officials.  After the exit procedure, during the
time when all university officials were attending the
commencement ceremony, I went to the office where the
questionnaires were deposited and claimed them, with no
protest offered by the lower ranked cadres who, excluded
from the commencement ceremonies, were on break in the
office.

Measuring Guanxi Use

From the 1993 survey data the variable guanxi.use was
constructed by summing the scores of three statements:
          Attitude 3: I can't easily change jobs because
          I must respect others who helped me get the
          job.
     Attitude 12: Guanxi was important in my getting a
job.
     Attitude 15: I used the backdoor to get my job.
Responses were made according to a 5 choice Likert-type
scale:  false, mostly false, sometimes false/sometimes
true, mostly true and true, coded 0 to 4 respectively.
The possible range of raw scores was thus 0 to 12, with a
mean of 5.1 (Table 1.1). A variety of more complex
indices with weighted variables were constructed but
eventually discarded in favour of this simple component
variable.  These statements were direct and straight
forward and thus appeared the best indicator of guanxi.
The summing of variables allowed for a larger range which
permitted greater statistical manipulation.  The variable
was also collapsed into four categories to facilitate
bivariate analysis (See Table 1.2).
     Some researchers in China have chosen not to use the
specific term guanxi for fear that respondents would not
be honest due to the political sensitivity of the
concept.32  My pre-tests and preparatory interviews
showed quite the contrary.  Students were not only
willing to respond.  They were quite unconcerned about
anonymity and some even boasted about their connections.

Interviews with Graduates and Alumni

The survey data collected by questionnaire has been
supplemented with discussions I had with szu graduates
who work in Shenzhen.  About two dozen students have been
interviewed, many for several hours each.  In addition,
students in a class I taught wrote essays on the job
search techniques of their elder schoolmates.33  A sketch
of the key interviewees appears in Appendix E.  Some of
the conversations took the form of one-on-one interviews
while others took place during "focused group
discussions."  These latter events often occurred during
meals, which I occasionally was not allowed to pay for.
(In many cases alumni earned more than their former
teacher; for the teacher to pick up the tab might be
interpreted as face-losing.)  The selection of alumni
depended to a large extent on contacting alumni who were
my former students and asking them to contact their
classmates.  The majority of students interviewed came
from the Foreign Language, Foreign Trade and Finance, and
Management departments.  These graduates worked in the
economic sectors covered by most alumni.
     Not included in the above interview count are the
numerous contacts I had with students during my six years
of teaching at szu.  From my first week on campus, I
encouraged students to "drop by" (later, to "phone
first") and chat.  Over my tenure I probably spent
hundreds of hours talking with students.  In some cases I
served as a counsellor (the guidance office was closed in
1989) and advisor to the lovelorn.  Only rarely were
students driven by ulterior motives (e.g., sponsorship to
the U.S., participation in black market activities).
These conversations were not formally recorded and were
never focused on the topic of my thesis.  They do,
nonetheless, provide illuminating data, especially for
one who was being educated on a daily basis about the
intricacies of an alien culture, "understanding the
native perspective" in the jargon of ethnographers.
     In addition, over the period of my research I had
numerous discussions with Chinese educators and policy
makers who were able to shed light on the workings of the
job allocation system.  A ten-day stay in Shanghai during
March 1994 allowed me to meet with job placement officers
at four tertiary institutions: Fudan University, Shanghai
University, Jiaotong University and Shanghai
International Studies University, as well as with
researchers at the Shanghai Institute of Human Resource
Development.  In addition, I visited the labour exchange
market for university students and potential employers.
     Quantitative theorists are usually concerned with
the problems of sampling34.  In these types of studies,
academics generally agree on statistical procedures.  But
qualitative studies more resemble a loose collection of
case studies.  Rather than employing the rules of
probability and emphasizing randomness, the qualitative
researcher commonly attempts to get as diverse a set of
respondents as possible.  Given the general difficulty in
getting informants in the first place, it is certainly
unlikely that "all bases will be covered."  That is a
built-in limitation to the qualitative approach.  The
best one can do is to ensure that the audience is aware
of possible sampling flaws, especially what kinds of
people are omitted.  For this thesis, I talked with
several hundred students and alumni about their jobs, but
I would not call these conversations "interviews."  I
took interviews with about twenty-five students and
graduates, the key ones of which are identified in
Appendix E.  They cover a range of disciplines but most
come from the Foreign Language Department where I taught.
The interviews lasted anywhere from 5 minutes to 5 hours.
They were generally conducted in English, the only
language in which both researcher and subject were
fluent.  In some cases where the informant's English was
of below average standards, I included one of my students
who could provide interpretation if needed.
     Despite my general inclination towards technology, I
did not use a computer to analyze my data.35  Instead, I
typed up my field notes and usually used coloured markers
to code and categorize data.  My data logs consisted of
both interview notes and my own comments during
interviews, which I always bracketed so as not to get the
two confused.
     Qualitative research is often a deeply personal
experience.36  Researchers often get emotionally involved
with their subjects.  In addition, each study is unique,
more so than statistical studies.  Qualitative studies
proceed in unexpected ways, not according to pre-arranged
designs.  As M.D. LeCompte & J. Preissle37 point out:

     Some investigators stumble upon an interesting group
     or event and formulate their designs to match this
     fortuitous occurrence.  Other researchers discover
     an intriguing instrument or technique for data
     collection and compose a problem its use would
     illuminate.  The order in which decisions are made,
     as a consequence, may be somewhat variable and
     idiosyncratic.

Thus, it is not unusual for research questions to be
changed; such reformulation is to be expected.  Although
my basic question remained the spine of the research, the
tissue around it changed.  My views on guanxi also
changed.  I began to see it less as a mechanism of
unfairness and more as a tool in the hands of skilled
artists.  I also began to realize that it was perceived
by students as a more important tool than it actually
was.
     Ethnography finds its roots in the Greek words
ethnos (people) and graphia (writing).38  It is: writing
about people.  The researcher is not detached from the
people whom s/he observes.  Because I was not Chinese, I
was an alien among the Han Chinese; I was allowed to be
an "observer as acceptable incompetent."39  J. Ogbu, a
Nigerian by birth, makes this point in the introduction
to his published thesis.40  The Afro-Americans whom he
interviewed in Stockton, California, accepted his most
basic questions about discrimination because they figured
he was ignorant.  A black American could not have done
Ogbu's study; a Chinese could not have done mine.
Foreigners in China often complain about being treated as
foreigners; but for this researcher this aspect of
ethnocentrism and xenophobia was a blessing in disguise.
_______________________________
1.See my three essays, (1993a), "Computers and Academic
Writing;" (1993b), "Transferability of Curriculum
Innovations;" (1991), "Roleplaying Helps Inactive
Students."

2.Wolcott (1984), The Man in the Principal's Office.

3.Agelasto (1994), "Shenzhen University's Search for Cultural
Identity."

4.Wolcott (1988), "Ethnographic Research in Education," 194.

5.Ibid.

6.These relationships are often mutually exploitive.  The
Western party needs the Chinese academics for gathering
data.  The Chinese collaborator needs his/her Western
peers to analyze the data and write them up in a form
acceptable for publication in an international journal.
Neither party could do the research independently.  Many
Western scholars treat their Chinese partners as low
level flunkies and necessary evils: someone with the
right connections who contributes little more than
collecting data.  The Chinese parties view their "foreign
friends" as walking dollar signs, providing an open check
book for travel abroad.  Despite this worse case
scenario, some legitimate collaborations occur.  See,
e.g., Lewin, et al. (1994), Educational Innovation in
China.

7.Bogdan & Biklen (1982), Qualitative Research for Education,
120.  For interesting experience about gaining access,
see: Ball (1984), "Beachside Reconsidered."

8.Lincoln & Guba (1985), Naturalistic Inquiry.

9.LeCompte & Preissle (1993), Ethnography and Qualitative
Design in Educational Research.

10.E.g., Lofland and Lofland (1984), Analyzing Social
Settings.

11.Ball (1981), Beachside Comprehensive; Wolcott (1984), The
Man in the Principal's Office; Ogbu (1974), The Next
Generation.

12.Ross (1993), China Learns English; Schoenhals (1993), The
Paradox of Power in a People's Republic of China Middle
School; Trueba & Zou (1994), Power in Education: The Case
of Miao University Students and Its Significance for
American Culture.

13.All ethnographic research is qualitative, but not all
qualitative research is ethnographic, a term that implies
that the observer becomes one with the community s/he
examines.  Taking interviews, a technique widely used by
sociologists and political scientists, may be part of
qualitative design and is associated with ethnography.
But interviews alone do not suffice.  Living among those
being researched is essential for the ethnographer; it
gives one the appearance of a spy, which is exactly what
the ethnographer is.

14.Morrow (1994), Critical Theory and Methodology, ch. 8.

15.LeCompte & Preissle (1993), Ethnography and Qualitative
Design.

16.Lincoln & Guba (1985), Naturalistic Inquiry; Guba & Lincoln
(1989), Fourth Generation Evaluation.

17.Lancy (1993), Qualitative Research in Education, ch. 2.

18.Wolcott (1984), The Man in the Principal's Office.

19.Cheng (1991), Planning of Basic Education in China.

20.Ross (1993) China Learns English; Schoenhals (1993), The
Paradox of Power.

21.Lancy (1993), Qualitative Research in Education, 66.

22.Rist (1980), "Blitzkrieg Ethnography."

23.Pelto & Pelto (1978), Anthropological Research.

24.Ibid., 34ff.

25.LeCompte & Preissle (1993), Ethnography and Qualitative
Design, 332-341.

26.See, ibid., 341-354.

27.Lincoln & Guba (1985), Naturalistic Inquiry, ch. 11.

28.Ely et al. (1991), Doing Qualitative Research.

29.I knew of several cases where students intercepted these
notices and their parents never knew they failed courses.

30.Zhang & Chen (1993),Shenzhen University: Its First Decade.

31.Rosen (1987), "Survey Research in the People's Republic of
China."

32.E.g., Bian (1994), Work and Inequality in Urban China.  See
discussion in Chapter III of this thesis.

33.This is discussed at the beginning of Chapter IX.

34.See, LeCompte & Preissle (1993), Ethnography and
Qualitative Design in Educational Research ch. 3.

35.A description of software available at the time of research
appears in Tesch (1990), Qualitative Research, 147-298.

36.Ely (1991), Doing Qualitative Research.

37.LeCompte & Preissle (1993), Ethnography and Qualitative
Design, 156.

38.Ibid., 1.

39.Lofland (1971), Analyzing Social Settings, 100.

40.Ogbu (1974), The Next Generation.