Content uploaded by Christian Goebel
Author content
All content in this area was uploaded by Christian Goebel on Dec 18, 2014
Content may be subject to copyright.
Journal of
Current Chinese Affairs
China aktuell
The Entanglement between Science and Politics
Göbel, Christian (2014), Let’s Not Go There: Coping with (Pre-) Selection Bias in
Collaborative Field Research, in:
Journal of Current Chinese Affairs
, 43, 2, 87–106.
URN: http://nbn-resolving.org/urn/resolver.pl?urn:nbn:de:gbv:18-4-7489
ISSN: 1868-4874 (online), ISSN: 1868-1026 (print)
The online version of this article and the other articles can be found at:
<www.CurrentChineseAffairs.org>
Published by
GIGA German Institute of Global and Area Studies, Institute of Asian Studies
in cooperation with the National Institute of Chinese Studies, White Rose East Asia
Centre at the Universities of Leeds and Sheffield and Hamburg University Press.
The
Journal of Current Chinese Affairs
is an Open Access publication.
It may be read, copied and distributed free of charge according to the conditions of the
Creative Commons Attribution-No Derivative Works 3.0 License.
To subscribe to the print edition: <ias@giga-hamburg.de>
For an e-mail alert please register at: <www.CurrentChineseAffairs.org>
The
Journal of Current Chinese Affairs
is part of the GIGA Journal Family which includes:
Africa Spectrum ● Journal of Current Chinese Affairs ● Journal of Current Southeast
Asian Affairs ● Journal of Politics in Latin America ● <www.giga-journal-family.org>
Journal of Current Chinese Affairs 2/2014: 87–106
Let’s Not Go There: Coping with (Pre-)
Selection Bias in Collaborative Field
Research
Christian GÖBEL
Abstract: Field research in China often requires the researcher to
cooperate with two kinds of actors: research collaborators, such as
those at universities or official think tanks, and local officials. These
actors facilitate or enhance field access, but such access comes at the
price of a potential “pre-selection bias” in data collection. Some scholars
have argued that dependence on these “gatekeepers” introduces a
significant bias into research outcomes. I argue, however, that the
constraints faced by China scholars in their field studies are not abso-
lute, but function by degree. The CCP is monolithic neither in its
organization nor in the thoughts of its agents, and close collaboration
with local partners can help remove normative bias rather than neces-
sarily introducing it. Most importantly, an argument built exclusively
on the power of structural constraints discounts China scholars’ most
crucial abilities: to learn, to think critically and to research holistically.
Manuscript received 20 January 2014; accepted 12 March 2014
Keywords: China, field research, case selection, interviews, data col-
lection
Dr. Christian Göbel is a professor of Modern China Studies at Vi-
enna University. His current research projects examine the role of
change agents in policy innovation in China and the impact of infor-
mation technology on the operation of non-democratic regimes. He
was trained in Political Science and China Studies in Erlangen, Taibei,
Heidelberg and Duisburg. Previous to his appointment in Vienna, he
held positions in Lund and Heidelberg. He is the author of The Politics
of Rural Reform in China (Routledge 2010) and The Politics of Community
Building in Urban China (Routledge 2011, with Thomas Heberer) and
has published widely on topics related to state–society relations and
political reform in China and Taiwan.
E-mail: <Christian.Goebel@univie.ac.at>
88Christian Göbel
Introduction
When conducting fieldwork, we form impressions, collect data and
conduct interviews. Needless to say, it is an essential part of many
research projects. Fieldwork serves China scholars not only in acquir-
ing the information that is needed to test, refine or even formulate
our theories, but also on an individual level in accumulating the experi-
ence that is needed to intuitively pre-judge the “probability of a given
explanation being correct in a certain kind of setting” (McKeown
2010: 159). The more experience a researcher has, the better he or
she will be able to assign that probability, and therefore to formulate
theories that are both relevant and robust. A good example of this
“folk Bayesian” approach (McKeown 2010: 158) is Kevin O’Brien’s
(2006) “staying open to unforeseen ideas (and even new topics of
enquiry)”, described in the volume Doing Fieldwork in China edited by
Maria Heimer and Stig Thøgersen (2006a). Experience is the key to
judging if a particular event or insight merits a change in research
design or not – for inexperienced researchers, most ideas will be un-
foreseen and most topics new. Hence, fieldwork plays an essential
role not only in how we explain a particular phenomenon, but also in
forming what I like to call our “research personality”. This encom-
passes the topics we consider as relevant (O’Brien 2006), our inter-
view strategy (Solinger 2006) and even how we speak (Thøgersen
2006).
It is therefore relevant to consider the potential impact of the
particular nature of the Chinese one-party regime on our findings and
perceptions. The “guidance” (ሬੁ, daoxiang) of public opinion is an
integral part of the ruling strategy of the Chinese Communist Party
(CCP), and the production of information is tightly regulated. Foreign
researchers who conduct “independent” – that is, not officially sanc-
tioned – research are breaking the law (Heimer and Thøgersen
2006b). Such breaches of law are common and often tolerated (see,
for example, Hansen 2006), and while “guerrilla interviewing” (Gold
1989) might work well with social actors, it quickly reaches its limits
when the focus is on politics. Because of this, many researchers, es-
pecially those who study Chinese politics, seek the partnership of an
official local collaborator such as a university or a government think
tank (Smith 2013). An official partnership enhances access to political
circles but comes at the price of a potential pre-selection bias that is
introduced into the data collection process. Official organizations
Coping in Collaborative Field Research 89
function as gatekeepers: They ultimately decide where foreign re-
searchers can travel and to whom they can speak, and, by threatening
to withdraw their cooperation in the future, they even influence what
the foreign researcher can and cannot publish. For those who study
Chinese local politics, such as the author of this research note, the
danger of making inferences based on biased information is even
greater, as not only the collaborators but also the local politicians are
constituent parts of an authoritarian regime that is very sensitive as
regards domestic and foreign public opinion.
The presence of these potential biases can easily lead to the as-
sumption that, perhaps without knowing it, field researchers are
“walking in the footsteps of the Chinese Communist Party” (Hansen
2006). Carsten Holz even suggests that “China scholars [have] all
been bought” (Holz 2007). In light of the constraints outlined above,
the assertion that publications based on field research are biased to-
wards CCP norms is logically compelling, as is, by extension, the
assertion that our accumulated experiences have made us favourably
disposed towards the CCP and its rule. Because of the huge implica-
tions of these assumptions for what we know about China, how we
study China and how we evaluate political, social and economic de-
velopments in China, these assumptions deserve to be closely scrutin-
ized. While confirming that there are indeed structural constraints
particular to field research in China, and that these constraints can
influence what is published and how, I contend that this relationship
is far from deterministic, that the “pre-selection bias” induced by our
research collaboration is less of a problem than these statements sug-
gest, and that the quality of our data can in fact be improved through
these connections.
The article1 develops as follows: First, I will argue that the asser-
tion that field research inevitably draws China scholars toward pre-
senting an overly positive picture of China is built on flawed prem-
ises. It presupposes a monolithic CCP, CCP agents who inevitably
espouse CCP narratives, collaborators who willingly introduce bias
1 Most of the field research described in this contribution was conducted in the
context of a project on agency in local policy innovation funded by the Swedish
Science Council (Vetenskapsrådet Project No. 2011-1495). I am grateful to
Björn Alpermann, Kevin O’Brien, Chen Xuelian, Rüdiger Frank, Sarah Hanisch,
He Zengke and one of the anonymous reviewers for comments to a previous
version of this article.
90Christian Göbel
into our research, and field researchers who are unable to change
their predispositions when faced with new evidence. While I do not
deny that the circumstances under which we conduct research can
influence what we discover, this influence is not absolute, but a mat-
ter of degree. In fact, there is an important “intervening variable” that
can prevent aforementioned structural constraints from translating
into scholarly publications with a CCP bias: the researcher. An argu-
ment built exclusively on the power of structural constraints neglects
the power of methodological standards that can help us both identify
and address ideological and selection biases. In addition, the person-
ality, experience and skills of a researcher influence the data collection
process, as the various contributions in Heimer and Thøgersen’s
edited volume lucidly illustrate. The present contribution draws on
examples from the author’s research on local politics. Hence, when
referring to “cases”, I usually mean territorial units such as different
counties, townships or villages.
Flawed Premises
At first sight, the argument that various constraints inevitably intro-
duce a pro-CCP bias into scholarship based on field research in Chi-
na is convincing: First, official gatekeepers ensure that scholars gain
access only to cases that reflect positively on the CCP; second, inter-
viewees praise the party and its policies; third, because future field
access is blocked if the CCP does not deem research outcomes ac-
ceptable, China scholars “habitually please the Chinese Communist
Party” (Holz 2007: 36). Four implicit premises underlie this argu-
ment:
First, CCP agents always behave in conformance with the stand-
ards set by the CCP.
Second, CCP agents always repeat the authoritative narratives set
by the CCP.
Third, research collaborators introduce a normative bias into
researchers’ data.
Fourth, the field researchers integrate this bias into their output.
In the following sections, I will show that these premises are flawed.
The first premise is that our research partners or the cadres we
interview always do what the party wants. Indeed, sometimes it is in
Coping in Collaborative Field Research 91
their best interest to do what the party (whoever it may be that repre-
sents it) wants them to do. Yet, as research on central–local relations
in China amply demonstrates, party policies have been a bad predic-
tor of the behaviour of local officials. This also applies to dealing
with foreign researchers. During the ten years I have now conducted
fieldwork on and off, I did experience encounters like those de-
scribed by Holz (2007: 38). I was picked up by an official car, wined
and dined, and the questions I asked at interviews were answered
with quotes from official policies. However, these were individual
occurrences early in my career, and less the result of conscious design
than of my insistence on visiting certain locations where officials
trusted neither my collaborators nor me. Far more often, I obtained
information that was very critical of central government policies.
Eventually this information led me to publish a book that illustrates
how
the central government skilfully obfuscated its own role in gener-
ating the [peasant] burden problem and instead played peasants
and local cadres […] against each other (Göbel 2010: 11).
Being frustrated with what they perceived as unrealistic expectations
from the central government, these officials seemed to derive satis-
faction from complaining about their superiors. Critical accounts that
portray the central government as manipulative, driven by short-term
interests and not very efficient could not have been what the party
wanted, yet many such accounts exist. One reviewer of this article put
it very nicely:
Virtually all research on local politics belies the fact that the re-
gime is a single, monolithic entity that speaks with one voice. Lo-
cal officials literally complain all the time about their superiors,
and subordinates and researchers, foreign or otherwise, merely
have to listen and write down what they say.
The second premise is that cadres always assert the party line. Al-
though this premise has already been refuted by the examples above,
the sections that will follow necessitate a discussion of this issue in
greater detail. In my experience, being fed propaganda is less of a
problem than is the fact that certain topics are not being discussed at
all, which to a large degree falls into the responsibility of the re-
searcher. The plausibility of what one is told can be probed by means
of comparison and additional evidence, but the failure to ask the
92Christian Göbel
“right” questions leaves us with “unknown unknowns”. By contrast,
asking the “right” question but not receiving an answer yields
“known unknowns” that can be addressed at a later stage or by look-
ing at additional cases (Silver 2012). Very often, such “known un-
knowns” pertain to issues that interviewees do not wish to discuss for
one reason or another, which should motivate us to look for different
ways to fill that space on the map. As will be subsequently outlined,
written sources, further interviews (with different persons or the
same individual at a later point in time) or reasoned deductions with
encouragement for further research can serve this purpose.
The third premise pertains to Chinese collaborators in the field-
work process. Because they can pre-select research sites and might
participate in the field research, it is possible that they introduce a bias
into our research. However, collaborators can also be an asset and help
reduce bias: They can suggest cases where the information sought is
likely to be found, point out the peculiarities of a research location,
cooperate in interviews and identify unexpected findings. In addition,
they can help address the foreign scholar’s normative bias by pointing
out where facts are stretched to accommodate the researcher’s ideo-
logical position. Whether a collaborator proves to be an asset or not
very likely depends on the terms of collaboration. If the collaborators
have a stake in the project – for instance, if they have the chance to co-
author articles that further their own political or academic careers or
the opportunity to cooperate in future projects, or if the threat of a
project’s cancellation would mean losing research funds – they are
more likely to contribute to its success. In contrast, treating collabora-
tors as data slaves and denying them co-ownership of the results incen-
tivizes them to minimize their risks, reduce their commitment and
exploit the project for short-term benefits. That we cannot completely
overcome the problem of working with a truncated sample is often not
the fault of our collaborators, who need to rely on personal contacts
themselves to gain access to research locations. Every single Chinese
researcher whom I asked about sampling procedures in topics where
access to or cooperation with local governments is required told me
that personal connections were indispensable in gaining access to field
sites. In such cases, truncated samples are not the result of a pre-
selection bias introduced by our collaborators, but rather come about
because the government in China is at once authoritarian and frag-
mented (Lieberthal and Oksenberg 1988).
Coping in Collaborative Field Research 93
The fourth premise is that field researchers cannot influence the
situation they are in. In fact, the notions of being “bought” or being a
“tool” (Holz 2007) deny them the most important facility that a sci-
entist should have: critical thinking. It is true that if data is collected
in cooperation with an official partner, researchers often walk “in the
footsteps” of the party: Especially if their own funds are not signifi-
cant enough to pay salaries, foreign researcher might become add-ons
to existing research projects. These are frequently opaque internal
games that involve Chinese politicians, scientists and researchers and
that are played for money, influence and reputation. These processes
are doubtlessly relevant, interesting and deserve to be studied, but
they should not distract researchers from the purpose of their field
visit: to gather the information needed to answer a particular research
question. Instead of brooding too much on the implications of the
political nature of research in China, researchers should focus on only
two issues: the representativeness of the cases being explored, and
the availability and quality of the information obtained. It is extremely
important to be aware of how the circumstances of access to the field
might bias the results, but the primary concern should not be how
this bias is produced but, rather, where it points and how it can be
reduced.
Representativeness
Although “pre-selection bias” denotes that local collaborators impose
limits on the selection of field research locations, the term is, strictly
speaking, incorrect in the context of small-N research (Gerring 2007).
As small samples can never be representative for a very large popula-
tion of cases, the problem of selection bias does not exist there. In
case-study research, the danger instead lies in misjudging the signifi-
cance of a particular case. What is perceived by the researcher as a
“typical” performer in a particular policy field, such as environmental
regulation, political reform or policy implementation, might indeed be
a case where a policy is working extremely well. Another example of
misjudgement is the assumption that one’s sample contains bad, in-
termediate and good performers, when indeed all cases are “models”
that perform extremely well when compared to the rest of China.
However, not being able to tell how a few cases perform when com-
pared to the whole population is not a problem of selection bias but,
94Christian Göbel
rather, of ignorance about how the totality of cases is distributed along
a particular feature. This might be due to a lack of information or an
underspecified research design. If a research design is not well speci-
fied – that is, if it is unclear what factors will be singled out for an
explanation – then it is difficult to assess the performance of a case
when compared with other cases. Most China scholars are probably
aware that small-N research drawn from a truncated sample is useful
for proving that a certain phenomenon can exist in China and for
generating hypotheses that can later be tested in large-N studies.
However, it does not tell us anything about how common an observed
phenomenon really is, and what the alternative outcomes may be.
Having a clearly specified research design does not mean being
inflexible: When an assumption proves incorrect, or when a more
interesting or relevant puzzle presents itself during fieldwork, the
research design can, and perhaps should, be adjusted or changed.
However, being flexible is different from being unprepared. As stated
above, it is the researcher’s experience and prior assumptions about a
phenomenon that enables her/ him to make informed judgements
about the relevance of a particular phenomenon.
Innovative research topics often present themselves when initial
assumptions about how certain processes work are proven wrong in
the field. In my experience, confronting interviewees with theoretical
assumptions almost always leads to animated discussions about their
validity in a particular context. The alternative explanations that sur-
face in such discussions are extremely useful for refining one’s hy-
potheses. The better that one is prepared, and the clearer that one’s
research aims are, the more substantial those discussions will be. For
example, when conducting research on the rural tax and fee reform, I
hypothesized that the abolition of the agricultural tax would render
village elections meaningless. With county- and township-level gov-
ernments basically deciding on and financing village-level projects,
the elected villager’s committee would become nearly obsolete. The
official I interviewed retorted that elections would become, in fact,
more important. He foresaw the selling of village land and other
communal property by the committee and argued that the election
and recall rights specified in the Organic Law of Village Committees
might prove formidable weapons in the resistance of such measures.
The Wukan incident proved him right.
Coping in Collaborative Field Research 95
It follows from all of the above that the probability of getting
meaningful results is higher if field research is thoroughly prepared.
Especially if research funds are scarce, it often pays to spend more
time on refining the research design before setting out to conduct
field research. If hypotheses are clearly specified, and if it is clear
what the purpose of the project is, then it is also easier to identify the
localities that will be most useful to visit. Ideally, the value distribu-
tion of the phenomenon that is to be explained should be visualized
and/ or juxtaposed with likely explanatory factors on a scatterplot.
When conducting research on the rural tax and fee reform, I drew
GIS (geographic information system) maps to visualize the effect of
the reform on county-level expenditures in Anhui Province (Göbel
2010: Chapter 7). This enabled me to identify clusters of counties
where the impact of the reform was more or less severe, and thus to
try to get access to one county in each cluster.
Not only was case selection made easier through the techniques
described above, but this and descriptive statistics also made it pos-
sible for me to specify whether the cases I had already studied were
typical, deviant or diverse with respect to the impacts of the tax re-
form on local finances in China. A wealth of data on various subjects
for provinces, cities and counties is easily obtainable from statistical
yearbooks. Although the reliability of this data is doubtful, it at least
allows the researcher to roughly grasp the differences between locali-
ties. For example, it is likely that the actual revenue of a particular
county is considerably higher or lower than the official statistics indi-
cate, but it is unlikely that a poor county ends up in the middle of the
revenue distribution. Statistical data is only one way to fathom distri-
butions. Another way is to mine pre-existing case studies. Chinese
colleagues have already explored many of the topics non-Chinese
researchers are working on, and often case studies for several loca-
tions are already available. By the time articles on China’s rural tax
and fee reform started to be published in Western languages, a lot of
research had already been published in Chinese. Newspapers and the
Internet are other valuable sources of information about develop-
ments in China. Such data also allows researchers to assess, for ex-
ample, how the implementation of a certain policy varies across Chi-
na. On this basis, what kind of case is needed can be specified, and
the location that is eventually chosen can be compared to other local-
ities that existing research already covers (Hurst 2010). The local
96Christian Göbel
partners are usually better informed about developments in the Chi-
nese scientific community than the foreign researchers, which is an-
other reason why the need to collaborate should be seen as an oppor-
tunity instead of a burden (Alpermann 2009).
Availability and Quality of Data
Two main concerns of field researchers are the availability and quality
of data. While the quality of the quantitative and qualitative data one
collects is often problematic, the “unknown unknowns” are the big-
ger challenge. As for quality, numerical data is indicative of a locality’s
place in relation to other localities, but is nevertheless seldom accu-
rate. Statistics serve political purposes, so there is an incentive to
“correct” data either upwards or downwards, depending on the spe-
cifics. However, the collection of statistical data, useful as it may be,
is usually not the main purpose of field research. Much of the infor-
mation sought is non-numerical – for example, regarding decision-
making structures and processes, communication networks and the
actions of relevant actors. Arguably, judging and improving the quali-
ty of this kind of information is more feasible than getting correct
numbers, because it is easier to forge data than to make up a narrative
that survives close scrutiny. When probing for details, cross-refer-
encing different narratives, probing counterfactuals and asking for
evidence, contradictions often surface in a made-up story, especially if
the interviewee does not know the questions in advance, and if sev-
eral people involved in the same story are questioned separately from
each other. In my experience, officials do say things like “the situa-
tion has vastly improved since we implemented policy X”, but when
they are probed for how they define “improvement” and asked for
evidence of tangible changes, doubtful assertions reveal themselves
very quickly, and these doubts are usually confirmed when other
officials or members of the general public are interviewed (for excel-
lent detective work in this respect, see Smith 2009).
Once more, this kind of scrutiny is possible only when one is
well prepared. The better informed the researcher is about a location
and the phenomena s/he is interested in, the more targeted her/ his
questions can be and the easier it is for her/ him to evaluate the
truthfulness of the narrative. I should add at this point that I have
never had to hand in my list of questions in advance. Also, despite
Coping in Collaborative Field Research 97
the fact that we might be “walking in the footsteps of the Chinese
Communist Party”, routines can be changed. Officials might prefer
group interviews, but my collaborator and I usually ask to speak to
our interviewees one after another. So far, we have never been re-
fused this request.
It is my impression that acquiring accurate information is a big-
ger problem than the danger of being told a carefully crafted fairy
tale. Rather than fabricating a story, officials will simply refuse to
yield information. They might remain silent, smile knowingly, excuse
themselves for not being able to talk about a sensitive issue, claim to
not remember, promise to tell later, or change the subject. On the
one hand, this is unfortunate, but on the other hand, answers like
these turn an “unknown unknown” into a “known unknown”. Con-
ducting field research is like drawing maps: When testing an assump-
tion by tracing the chain of events that unfolded in a particular place,
it is important to first establish the boundaries of a particular story.
By means of thorough preparation, careful observation, and many
interviews and conversations, the number of “unknown unknowns”
within that map can be reduced. Needless to say, one can never be
sure of knowing everything – this is the nature of “unknown un-
knowns”. However, the broader and deeper the information about a
phenomenon or process, the fewer “unknown unknowns” will re-
main. With each answer and observation, the picture becomes clearer.
Information that is not revealed corresponds to blank spots on the
map that often mark sensitive territory which interviewees try to
avoid. These blank spots need to be clearly identified as such, and
additional probing should be undertaken in an attempt to subse-
quently “fill them in”. The better researchers are prepared, and the
more obstinate they are, the better their map will be. The following
strategies should help to reduce uncertainty.
Strategies
First, know what you want. As elaborated above, it is important to
have a carefully specified research design. It should be clear what the
research question is, if and how others have already answered it, and
how the project will contribute to a better understanding of a certain
phenomenon or process. Specifying working hypotheses helps to
clarify not only the purpose of the project but also what kind of in-
98Christian Göbel
formation or data is needed to answer the research question and what
kind of cases should be examined (George and Bennett 2005: 67–88).
Setting out with very clear objectives is a constraint only if ideas are
held onto even after they prove incorrect, uninteresting or unfeasible.
As explained above, a set of clear hypotheses allows us to not only
use our precious time in the field more effectively but also elicit re-
sponses from our interviewees. If being turned into a propaganda
tool is indeed such a danger, then we should come well prepared.
When conducting research on online complaints in China, I was
shown GIS-based visualizations of complaint frequencies in one lo-
cality. My immediate assumption was that such maps are used to
identify places where certain problems are especially pertinent in
order to implement targeted measures to prevent social unrest. In a
second visit to that location, I brought up this assumption, but was
told that the maps’ mere purpose was to impress official visitors.
When I insisted that these maps could be used to maintain stability, it
was explained to me that this was impractical. In the course of a long
discussion, I came to realize that the employees knew the places they
were assigned to supervise so well and communicated with one an-
other so often that they did not need any maps to highlight problem
areas. This discussion yielded the new hypotheses about the relation-
ship between technology and agency in China’s stability maintenance
policies that inform my present research project.
Second, be informed. Unwittingly becoming a “victim of propa-
ganda” is a danger not only if we do not know what we want but also
if we are uninformed. This pertains, first, to existing research. What
we discover in the places we visit should be evaluated in light of what
we already know. Is it likely that, for example, county X is performing
so much better with respect to environmental protection than other
counties? Are the institutional or ideal constraints that scholars found
in other locations not at play here? Why not? Are the explanations
that officials give when confronted with these questions convincing?
Related to that, being well informed about the research location that
is being visited and aware of the backgrounds of the political leaders
in question enables a researcher to ask pointed questions, and makes
interviews more interesting for both parties. If the interviewer is pre-
pared and interested, interviewees engage more deeply in the conver-
sation and sometimes even reveal things they were previously unwill-
ing to. In a recent interview conducted by myself and a collaborator,
Coping in Collaborative Field Research 99
an official was visibly annoyed at our visit. He decided to use the
opportunity to practise his English with me. We chatted a bit about
China and the US, where he had worked prior to becoming an offi-
cial. At one point, I asked him why he came back, claiming that he
was obviously overqualified for this job. As soon as I said this, he
jerked upright and asked me how I got that impression. This led to a
heated discussion between the three of us about the requirements and
perks of being an official and about the need for personal initiative in
“getting things done”, which had been our topic at the time. As both
my collaborator and I were intimately familiar with the design, subject
matter and current progress of the project, we were able to comple-
ment each other’s questions and comments. When we parted ways,
our interviewee even commented that we had been able to get him
interested in the interview, which would not have been possible if we
had not been very familiar with the subject matter.
Third, be intimately familiar with the topic under study. It has
seldom been the case that a discussion in which we were well in-
formed about a particular policy, the specifics of a research location
and the rank and duties of our interviewee did not lead to a good
discussion. Having such knowledge allowed us to ask, for example,
rather detailed questions about how a specific target was to be
achieved in a given locality. Very often, this was the very puzzle the
official also had to solve, and the discussion revealed the enabling and
constraining factors we were also interested in. One example was the
case of tax and fee reduction and the financing of public services;
another, more recent one was the question of how the powerful pub-
lic security bureaus’ resistance to being included in a public perfor-
mance evaluation programme could be overcome. In this case, we
learned much about persuasion, lobbying and alliance formation
within the government. Familiarity also entails the ability to translate
academic concepts into the ideological categories or “codes” Chinese
officials use to speak about sensitive topics. For example, questions
about social unrest might be introduced by first referring to challeng-
es to achieving a “harmonious society” or the implementation of
“social management” policies and then asking for details.
The reason I am making these points, which are self-evident for
seasoned researchers, is that inexperienced researchers are often un-
aware of the “personal” factors that influence research outcomes and
insecure when first conducting field research (see Sæther 2006). As I
100Christian Göbel
pointed out above, the purpose of field research is not only to collect
data but also to build up research experience. I can be so adamant
regarding the “rules” I am formulating here because I violated them
myself when I conducted my first field study a decade ago. An ex-
perience I am unlikely to forget is a conversation with an official of a
provincial tax department: After I had asked two questions, he gave
me a stack of regulations and an official work report and asked me if
I had read this material. When I admitted I had not, he showed me
the door and asked me to come back the next day – prepared! Since
then, I have made it a point to be as well informed as I can be on the
issue I am examining, which makes interviews so much more pleasant
for both sides (a similar point is made by Solinger 2006). I have
found that officials enjoy talking to a foreign researcher who has at
least a rough understanding of the regulatory environment they are
operating under. I do not always live up to my own high standards,
and these are usually the field trips that do not go well.
Fourth, be inquisitive. Anyone who has ever read a government
work report knows the phrases that claim that a particular service has
been “greatly” improved, another “perfected”, and that the relation-
ship between cadres and the masses is now so much better. As men-
tioned above, officials are conditioned to use certain codes when
discussing sensitive issues or trying to conceal that nothing at all has
been done, and they tend to use these phrases extensively in inter-
views. Hence, it is important to follow up and ask for evidence.
When, for example, a township official claimed that her visits to
peasant households had improved the local political climate, my col-
laborator and I not only asked for the location and names of recently
visited households but also enquired about the exact nature of such
visits. Where did the conversation take place, what kind of snacks or
drinks was she offered, and what were the subjects of the conversa-
tion? Were all visits similar, or were there differences? What kind of
differences? What stories did she recall, and who told them? When
tracing decision-making processes, we follow the same strategy. We
ask very detailed questions such as who came up when and where
with a certain idea, what the immediate reactions were, and who was
involved in formalizing a policy based on it. Where possible, we ask
the same questions of all people who were involved in this process.
We express our doubt openly if we think that things do not match
up, and the explanations that follow sometimes corroborate a story
Coping in Collaborative Field Research 101
and sometimes reveal that we did not fully understand a particular
process. Where possible, we ask for figures, numbers and names. In
one Anhui township, I asked so many questions about revenues and
expenditures that the accountant fetched a large file that contained all
of the previous year’s revenues and expenditures, put it on the table,
and told me that he and his colleagues would now leave to attend to
important matters. I could “rest for a while” in his office if I wanted.
More often than not, being inquisitive helps in reducing the blank
spots on the map. As a rule of thumb, however, the more illegal or
irregular behaviour involved, the more difficult it is to fill the blank
spots.
Fifth, be patient. Some pieces of information are easier to come
by than others. However, being unable to obtain information during
one research visit does not mean that the information is unobtain-
able. Sometimes it takes several visits, or even a long-term stay in one
location, to arrive at a clear picture about a process or an event. Time
can work in favour of a researcher in at least two ways: First, infor-
mation that is deemed sensitive at one time might not be considered
sensitive at a later date; second, repeated contacts between interview-
er and interviewee can serve to build up trust. There is often consid-
erable risk involved for interviewees who talk to foreign researchers,
especially if the interviewees are officials, the topics sensitive, and the
interviews conducted unofficially. Getting to know each other better
helps to alleviate worries that the information that is volunteered will
fall into the wrong hands and ultimately hurt the interviewee. On the
one hand, the researcher can take the time to assure the interviewee
that her/his information will be stored safely, and that her/ his iden-
tity will not be revealed in publications resulting from the project. On
the other hand, repeated interactions in which the researcher reveals
some information about the research project and her/ his underlying
motivations go far towards assuring the interviewee that the research-
er has no ulterior motives. Fortunately, communication by e-mail or
voicemail now allows us to exchange information or follow up on
unanswered questions. Where the interviewee is generally willing to
help the researcher, but does not dare to reveal information to a for-
eign person, the collaborator might conduct a second interview alone.
Here, a trusted local collaborator can once more prove to be an asset
in a research project that involves fieldwork.
102Christian Göbel
Sixth, be open. In my experience, being clear about my research
objectives and sharing my insights with interviewees goes a long way
towards building the trust needed for them to volunteer information.
Collaborators, and more so our interviewees, take risks in cooperating
with us, and for that they deserve to be respected. While respect
should define the interaction between interviewer and interviewee
from the beginning, trust needs to be built up. Researchers might
suspect that an interviewee has not told them everything they want to
know, and interviewees might suspect that researchers will misrepre-
sent what they are being told or accidentally reveal sensitive infor-
mation. Thomas Heberer and Gunter Schubert regard local officials
as a “strategic group” (Heberer and Schubert 2012), which in my
experience entails that they are careful about discussing sensitive
issues and sharing data with any person who is not “one of them” (㠚
ᐡӪ, zijiren). Getting to know each other often helps to ensure the
interviewee that the researcher can be trusted to keep information
secret, and the interviewee might feel safe in revealing more infor-
mation than was initially planned. Besides building trust, credibly
assuring an interviewee that information will be treated confidentially
and knowing some of the codes and phrases used to speak about
certain topics, creating an artificial divide between the official and
private capacities of the interviewee sometimes helps interviewees
feel at ease. For example, it is important to stress that the meeting’s
purpose is not to “conduct an interview” (䇯䰞, fangwen), which is an
official act, but rather to “chat” (㙺ཙ, liaotian), which is a private act.
Sometimes, issues are only revealed when we explicitly ask about a
respondent’s “personal opinion” (њӪⲴ㿱, geren de yijian). Hence,
part of the trust-building seems to entail demonstrating that the in-
terviewer knows the political environment – and the associated risks
of operating in such an environment – well enough to be able to
communicate in this fashion. Needless to say, mastering this skill is
yet another issue where the help of a committed collaborator is invalu-
able.
Not treating interviewees like data machines, but respecting the
risk they are taking, taking the time to answer questions about the
purpose of one’s research project, and sharing the insights gained
through literary reviews, interviews and data analysis not only allevi-
ates the worries outlined above but also satisfies the curiosity of the
interviewees. Officials, in particular, are curious about how things are
Coping in Collaborative Field Research 103
done in other locations, how policies could be improved in their lo-
cality and how certain issues are discussed in non-Chinese scholarly
publications. In my case, basing interactions on respect and taking the
time to build trust has resulted in re-invitations to certain research
locations, and those repeated visits have enabled me to deepen my
understanding about these places. This has benefitted not only cur-
rent research projects but also my general understanding of adminis-
trative processes in China. In addition, those locations are good can-
didates for field sites in future projects.
It needs to be pointed out that such an approach does not entail
being uncritical or overly accommodating or mean that sensitive is-
sues are being avoided. On the contrary, my experience has shown
that this kind of sensitivity is a necessary precondition for such topics
to be discussed at all.
Conclusion
Field research in China, as everywhere else, is challenging. Several
factors affect the quality and quantity of data that is collected in the
field, and it is the responsibility of the researcher and his or her peers
in the scientific community to judge the validity of the inferences that
are made from this data. A lot can go wrong, but I believe that the
choices field researchers face are far more complex than whether to
conduct field research at the cost of losing one’s integrity or whether
to ensure one’s integrity at the expense of depending on potentially
unreliable statistics, documents and secondary sources. In my opin-
ion, there are three variables that impact the outcome of field re-
search activities: first, the experience, preparedness and personality of
the researcher; second, the relationship between the researcher and
the interviewee; and, third, the nature of the topic that is being inves-
tigated.
As for the research topic, it goes without saying that topics con-
sidered sensitive by the Chinese central government are both more
difficult to research and riskier to publish. Collaborators are more
likely to refuse to participate in such a project, interview partners will
be harder to find, and places more difficult to visit (Thunø 2006).
The publication of the results is riskier, as well – a collaborator might
urge caution, and a researcher might refrain from publishing the re-
sults for fear of being denied a visa in the future. Although these
104Christian Göbel
circumstances also apply to topics that are extremely relevant for
understanding contemporary China, such as the student demonstra-
tions of the late 1980s, the crackdown at Tiananmen Square or the
relations between minorities and Han Chinese, such topics constitute
a minority. As numerous examples show, research on topics that are
locally considered sensitive or which might reflect negatively on the
central government – such as social unrest, land grabs, corruption or
the soundness of China’s financial sector – can be both researched
and published without fear of having to sell oneself out. In fact, the
number of non-Chinese scholars who have been denied visas to Chi-
na because of what they have published is small (Redden 2008).
The second factor, discussed in great detail above, pertains to the
researcher him- or herself. The more knowledgeable and prepared a
researcher is, the better he or she will be able to contextualize new
information and to reduce the blank spots on the imaginary research
map. Success in probing for “hidden” information depends on an-
other actor-specific factor: social and interview skills. This relates to
the third factor: If researcher and interviewee can create an atmos-
phere of mutual respect and trust, and if the interviewee is comfort-
able with or even enjoys the situation, then it is likely that he or she
will make an effort to help the researcher. That said, I have experi-
enced situations where it became quickly obvious that such an at-
mosphere could not be created, simply because, for whatever reason,
the interviewee and the interviewer did not strike a chord, or even
disliked each other. In such situations, I find it fairest to terminate the
interview as quickly as possible, but without being rude or impolite. A
similar issue pertains to the collaborator. Mutual respect and trust are
necessary conditions for collaborative research to succeed, as is shar-
ing a common objective: If both parties can reap the spoils of suc-
cessful collaborative research, then the likelihood that biases resulting
from case pre-selection can be addressed increases significantly. To
avoid disappointment, the terms of engagement must be spelled out
clearly and agreed upon before collaborative research commences.
Coping in Collaborative Field Research 105
References
Alpermann, Björn (2009), Political Science Research on China:
Making the Most of Diversity, in: Journal of Chinese Political Science,
4, 343–356.
George, Alexander L., and Andrew Bennett (2005), Case Studies and
Theory Development in the Social Sciences, Cambridge: MIT Press.
Gerring, John (2007), Case Study Research: Principles and Practices, New
York: Cambridge University Press.
Göbel, Christian (2010), The Politics of Rural Reform in China: State Policy
and Village Predicament in the Early 2000s, Milton Park, Abingdon,
Oxon, New York, NY: Routledge.
Gold, Thomas (1989), Guerilla Interviewing among the Getihu, in: E.
Perry Link, Richard Madsen and Paul Pickowicz (eds), Unofficial
China: Popular Culture and Thought in the People’s Republic, Boulder:
Westview Press, 175–192.
Hansen, Mette Halskov (2006), In the Footsteps of the Communist
Party: Dilemmas and Strategies, in: Maria Heimer and Stig
Thøgersen (eds), Doing Fieldwork in China, Honolulu: University
of Hawai’i Press, 81–85.
Heberer, Thomas, and Gunter Schubert (2012), County and Town-
ship Cadres as a Strategic Group. A New Approach to Political
Agency in China’s Local State, in: Journal of Chinese Political Science,
17, 221–249.
Heimer, Maria, and Stig Thøgersen (eds) (2006a), Doing Fieldwork in
China, Honolulu: University of Hawai’i Press.
Heimer, Maria, and Stig Thøgersen (2006b), Introduction, in: Maria
Heimer and Stig Thøgersen (eds), Doing Fieldwork in China,
Honolulu: University of Hawai’i Press, 1–26.
Holz, Carsten A. (2007), Have China Scholars All Been Bought?, in:
Far Eastern Economic Review, 3, 36–40.
Hurst, William (2010), Cases, Questions and Comparison in Research
on Contemporary Chinese Politics, in: Allen Carlson et al. (eds),
Contemporary Chinese Politics: New Sources, Methods, and Field Strate-
gies, 162–177.
Lieberthal, Kenneth, and Michel Oksenberg (1988), Policy Making in
China: Leaders, Structures, and Processes, Princeton, NJ: Princeton
University Press.
McKeown, Timothy (20102), Case Studies and the Limits of the
Quantitative Worldview, in: Henry E. Brady and David Collier
106Christian Göbel
(eds), Rethinking Social Inquiry: Diverse Tools, Shared Standards,
Lanham, MD: Rowman & Littlefield Publishers, 140–168.
O’Brien, Kevin (2006), Discovery, Research (Re)Design, and Theory
Building, in: Maria Heimer and Stig Thøgersen (eds), Doing
Fieldwork in China, Honolulu: University of Hawai’i Press, 27–41.
Redden, Elisabeth (2008), The Blacklist Academic Leaders Ignore, Inside
Higher Education, online: <www.insidehighered.com/news/2008/
07/14/china> (20 January 2014).
Sæther, Elin (2006), Fieldwork as Coping and Learning, in: Maria
Heimer and Stig Thøgersen (eds), Doing Fieldwork in China,
Honolulu: University of Hawai’i Press, 42–57.
Silver, Nate (2012), The Signal and the Noise: Why So Many Predictions
Fail-but Some Don’t, New York: Penguin Press.
Smith, Graeme (2013), Measurement, Promotions and Patterns of
Behavior in Chinese Local Government, in: Journal of Peasant
Studies, 6, 1027–1050.
Smith, Graeme (2009), Political Machinations in a Rural County, in:
The China Journal, July, 29–59.
Solinger, Dorothy J. (2006), Interviewing Chinese People: From
High-Level Officials to the Unemployed, in: Maria Heimer and
Stig Thøgersen (eds), Doing Fieldwork in China, Honolulu: Uni-
versity of Hawai’i Press, 153–167.
Thøgersen, Stig (2006), Beyond Official Chinese: Language Codes
and Strategies, in: Maria Heimer and Stig Thøgersen (eds), Doing
Fieldwork in China, Honolulu: University of Hawai’i Press, 110–
126.
Thunø, Mette (2006), In the ‘Field’ Together: Potentials and Pitfalls
in Collaborative Research, in: Maria Heimer and Stig Thøgersen
(eds), Doing Fieldwork in China, Honolulu: University of Hawai’i
Press, 245–261.
J
ournal of Current Chinese Affairs 2/2014: 1–2
Contents
The Entanglement between Science and Politics
Editorial
Karsten GIESE
Editorial
3
Introduction
Sascha KLOTZBÜCHE
R
W
estern-Chinese Academic Collaboration in the Social
Sciences 7
Analyses
H
eike HOLBIG
S
hifting Ideologics of Research Funding: The CPC’s
N
ational Planning Office for Philosophy and Social
S
ciences 13
D
oris FISCHE
R
T
he Impact of Changing Incentives in China on
I
nternational Cooperation in Social Science Research on
China 33
J
osef Gregory MAHONEY
Changes in International Research Cooperation in China:
P
ositive Perspectives 47
Sascha KLOTZBÜCHE
R
“Embedded Research” in Collaborative Fieldwor
k
65
Christian GÖBEL
L
et’s Not Go There: Coping with (Pre-) Selection Bias
i
n Collaborative Field Research 87
2 Contents
Research Articles
KAO Ya-ning
R
eligious Revival among the Zhuang People in China:
P
ractising “Superstition” and Standardizing a Zhuang
R
eligion 107
Olivia KRAEF
Of Canons and Commodities: The Cultural Predicaments of
N
uosu-Yi “Bimo Culture” 145
M
artin SAXE
R
R
e-Fusing Ethnicity and Religion: An Experiment on
T
ibetan Grounds 181
Contributors 205
