You are on page 1of 10

Social Science Research 36 (2007) 149158

www.elsevier.com/locate/ssresearch
0049-089X/$ - see front matter 2005 Elsevier Inc. All rights reserved.
doi:10.1016/j.ssresearch.2005.10.003
Convergence and conWdentiality? Limits
to the implementation of mixed methodology

Erin Leahey

Department of Sociology, University of Arizona, P.O. Box 210027, Tucson, AZ 85721-0027, USA
Available onilne 19 December 2005
Abstract
The purpose of this paper is to identify and explain a rather large impediment to the use of mixed
method research designs that begin with survey data. Although supplementing quantitative survey
analyses with insights derived from qualitative work is increasingly popular, there is also heightened
concern about ethical issues in research, particularly the conWdentiality of human research subjects.
Surprisingly, there has been little discussion about how these two trends intersect, and the extent to
which they are compatible. In this paper I argue that because of the important obligation to protect
human research subjects identities (and, to a lesser extent, the proliferation of publicly available
datasets), few researchers are actually in a position to implement this type of mixed methodology.
Despite their noted advantages over mono-method designs, mixed method designs that start with
quantitative data and subsequently supplement it with qualitative insights may not achieve the dom-
inance that many have supposed and desired.
2005 Elsevier Inc. All rights reserved.
Keywords: Mixed methodology; Research ethics; Quantitative and qualitative research; Primary and secondary data

I would like to pay tribute to the exemplary work of Barbara Entwisle, Jill Bouma, Gail Henderson, Susan
Short, Zhai Fengying, and Lisa Pearce, which prompted the argument outlined in this paper. I thank Kieran He-
aly for convincing me that my argument was important and publishable, and Grant Blank for helpful comments.
*
Fax: +1 520 621 9351.
E-mail address: leahey@arizona.edu
150 E. Leahey / Social Science Research 36 (2007) 149158
1. Introduction
Just as concerns about ethical issues in survey research seem to be cyclical (Presser, 1994),
so too are what we might call styles of research practice; and these days, mixed methods are
making headlines. Various researchers have lamented the traditional bifurcation of quantita-
tive and qualitative approaches (Bourdieu and Wacquant, 1992; Geertz, 1979; Hughes, 1984;
King et al., 1994), and others have promoted the use of mixed methods designs (Alford, 1998;
Brewer and Hunter, 1989; Bryman, 1989; Cook and Reichardt, 1979; Giddens, 1976; King
et al., 1994; Maynard and SchaeVer, 2000; Tashakkori and Teddlie, 1998; Tashakkori and
Teddlie, 2002). Although there are multiple ways to mix methods, and various names ascribed
to this process, mixed methods can be deWned generally as the integration of more than one
method or data source to investigate a phenomenon.
1
The posited beneWts of mixed method designs are multiple. Denzin (1988) contends that
mixing methods permits more exact understandingor, in methodological terms, enhanced
validity (internal and external) as well as reliability. Other advantages include greater conW-
dence in results; assistance in uncovering deviant or surprising dimensions of a phenomenon;
enriched explanations, and theory integration or synthesis (Newman and Benz, 1998, p. 84).
Because each method has its strengths and weaknesses, combining methods is purported to
give the best of both worlds (Chen, 1997). In fact, integrating qualitative and quantitative
data may be the best way to gain a complete understanding of social phenomena if, as Cole-
man (1986) attests, larger-scale quantitative work can tap the more structural eVects of inter-
est, and qualitative work can better address interpretive reasons and mechanisms.
These advantages contribute to the recent popularity of mixed method designs (Creswell,
2003). Datta (1994) argues that mixed methods can and should constitute the dominant
methodology. Some contend that it already holds this dominant position. Brewer and Hunter
(1989, p. 28) state that multi-method research is now discussed, planned, and conducted as a
routine matter, part and parcel of normal social science. Whether mixed-methodology has
already achieved a dominant status or not, it is unquestionably a rapidly growing area of
interest in research methodology (Tashakkori and Teddlie, 2002, p. ix). Within sociology,
new interest in mixed methodology is apparent from a review of recent American Sociologi-
cal Association meeting schedules: no session on mixed methods was scheduled in 2000, but
in the following 5 years, there were regular sessions and didactic seminars on the topic.
Despite the advantages and popularity of mixed methods, practical guidance is very lim-
ited. Most discussions of mixed methodology focus on the conceptual frameworkthe
ideaof integrating methods. For example, King, Keohane and Verba (1994) present what
they see as a common, underlying logic of inference in both quantitative and qualitative
research. After describing published research in the multivariate and interpretive tradi-
tions, Alford (1998) concludes with a chapter on the theoretical power of multiple para-
digms. The aim of Newman and Benzs (1998, p. 111) bookto place in perspective the
relationship between qualitative and quantitative researchis also conceptual. Even
Tashakkori and Teddlies (2002) reference book on Mixed Methodology, while providing
1
ClassiWcations of ideal-types of mixed method research (Caracelli and Greene, 1993) are relevant to my argu-
ment only to the extent that the limitation I identify applies to quantitative studies that integrate a qualitative
component, whether sequentially or concurrently. Of the various terms used to describe mixed method approach-
es [including triangulation, integration, synthesis, quantitative, and qualitative methods, and multi-method (Cre-
swell (2003)], I use the most recent terminology, mixed method.
E. Leahey / Social Science Research 36 (2007) 149158 151
various examples for illustrative purposes, primarily serves to classify ideal types of mixed
method approaches.
While articles that actually employ mixed methods provide a greater understanding of
the processes involved, many of the practical limitations of mixing methods are over-
looked. A common mixed method designone that is the focus of this papersupple-
ments a largely survey-based project with qualitative insights obtained from interviews,
focus groups, and/or observation. Examples from Entwisle and colleagues Designing
Points of Convergence (forthcoming) and Pearces (2002) article in Sociological Method-
ology (Integrating Survey and Ethnographic Methods for Systematic Anomalous Case
Analysis) are illustrative. Although Entwisle et al. illustrate multiple ways of using
qualitative data to supplement wide-scale survey data and even detail the various chal-
lenges they encountered, an air of optimism surrounds their mixed method solutions,
and the accessibility of their procedures to other researchers is discussed only brieXy. In
her presentation of an extension of deviant case analysis, Pearce (2002) argues that com-
bining survey analysis, sampling techniques, and ethnographic methods constitutes an
eVective way to choose cases for a qualitative study and to improve quantitative mea-
sures. She failed to realize that the beneWts she describes are not beneWts of the method
per se, but beneWts linked to her position on a research team involved in the primary sur-
vey data collection.
Perhaps it is understandable that as a reaction to longstanding debate between quanti-
tative and qualitative factions, advocates of mixed method designs would focus on its
advantages over its limitations, and on the idea of mixing methods over practical guidance.
In many ways, researchers implementing and promoting mixed methods are acknowledg-
ing the watershed that Coleman (1986) lamented: with statistical advances, increasing
national funding, the growth of data archives, and survey research dominance in the mid-
20th century, two elements essential to understanding the social world were neglected: con-
textual eVects and actors personal volition (i.e., individual reasons for engaging in action).
Mixed method researchers were enthusiastic about resurrecting these two elements and
joining them with more common statistical and survey methods. But their enthusiasm for
this new mode of research often took precedence over more practical matters.
An important limitation of this kind of mixed method designwhich begins with survey
data and adds a qualitative componenthas been overlooked by most researchers who write
about and use mixed methods. Paradoxically, the neglect of practical limitations is most
apparent in works, like those mentioned above, which exhibit what Maynard and SchaeVer
(2000) call conditional complementarity. The aim of conditional complementarity, which is
essentially a mixed method approach, is to integrate qualitative data, such as in-depth inter-
views, observation, archival material, and focus groups into a primarily quantitative, often
survey-based, study. The purpose of this paper is draw attention to a single limitation of such
mixed method designs: in short, integrating a qualitative component into a survey research pro-
ject is limited to a rather privileged subset of researchers. Either the researchers must be
involved in the original collection of survey data, or they must have access to geographical,
personal, or other unique identiWers to collect additional qualitative data from the research
site or original research participants. Two important and sweeping changes in the research
environment have reduced the size of the subset of researchers eligible to use such mixed
method designs. These include (1) an increase in the availability of secondary data and a cor-
responding decline in primary data collection eVorts, and (2) increased awareness about data
conWdentiality and the protection of human research subjects.
152 E. Leahey / Social Science Research 36 (2007) 149158
1.1. Greater availability of secondary data
EVorts to collect survey data intended for wide-scale use and to archive smaller-scale
surveys have proliferated in the last 50 years. In the Wrst half of the 20th century, often the
only option available to researchers interested in empirical investigation was to collect
their own data. Indeed, personal and local community studies were the norm. However,
starting in the 1940s there was a watershed in empirical research in which dense commu-
nity or organizational samples were replaced by national samplesand the struggling
eVort to use survey research to make statements about communities, organizations, or
social subsystems was overwhelmed by the greater statistical rigor of characterizing popu-
lations and analyzing behavior of individuals as independently drawn from the popula-
tion (Coleman, 1986, p. 1315). Changing funding priorities and increasing availability of
individualistic statistical techniques lent much legitimacy to large-scale, typically national,
survey research (Platt, 1996; Pontille, 2003). Since the instigation of various large, nation-
ally representative studies (such as NLS in 1966 and PSID in 1986) and the founding of
ICPSR in 1962, researchers have had the option of foregoing primary data collection in
favor of using data collected and archived by other researchers or research institutes and
funded by government agencies and foundations.
While primary data collection remains essential to researchers who study unique topics
for which no secondary data exist, its use has declined over the last 70 years. During in-
depth interviews that I conducted with social science journal editors, chairs of grant-giving
agencies, and directors of data archives (Leahey, 2005), a few lamented what they perceived
as a decline in primary data collection. This perception resonates with more systematic
archival evidence concerning the use of primary data in articles published in leading sociol-
ogy journals. Based on a 20% stratiWed random sample of all journal articles published in
the American Sociological Review and the American Journal of Sociology between 1935 and
2000, I found that the use of primary data declined over time (see Fig. 1).
2
In a logistic
regression of primary data collection on time, the coeYcient for the variable representing
time (with values ranging from 35 for 1935 to 100 for the year 2000) was statistically
signiWcant (p 6.01) and negative. At least among the more quantitative research published
in journals (Clemens et al., 1995), the use of primary data has declined between 1935 and
2000, even though the data also reveal that the ratio of empirical to theoretical studies has
increased over this same period. Therefore, researchers must be relying on secondary data
sources to a greater extent than they did in the past.
1.2. Increasing concerns about conWdentiality
However, the information required to supplement survey data with qualitative data,
such as respondents metropolitan areas, the names of schools or organizations sampled
for the study, or even the names of respondents themselves, typically cannot be made avail-
able to researchers accessing secondary data. Institutional review boards and institutions
2
Perhaps the slight increase in primary data collection during the 1980s and 1990s can be partly attributed to
Colemans (1986) call, and partly attributable to trends toward mixing methods (which often require primary
data collection). A reviewer also suggested that it might also be attributable to the dominance of comparative-his-
torical papers in ASR and AJS during these decades.
E. Leahey / Social Science Research 36 (2007) 149158 153
that collect and archive survey data have become rightfully more concerned about the pos-
sibility of deductive disclosure: the identiWcation of a particular research subject made pos-
sible when detailed demographic information is combined with geographic data. When
large- and small-scale survey datasets are made publicly available, direct identiWers like
names and addresses must be stripped, and indirect identiWers (typically geographic infor-
mation) are given arbitrary codes and/or collapsed to reduce the risk of participant identi-
Wcation. This is customary for the National Longitudinal Study (NLS), General Social
Survey (GSS), Panel Study of Income Dynamics (PSID), and National Longitudinal Study
of Adolescent Health (AddHealth), as well as smaller datasets archived at ICPSR.
3
For
example, a dataset may indicate which individuals were students at the same school, but
the school itself is not identiWed. This is adequate for quantitative assessments of contex-
tual school eVects on student achievement, but insuYcient for researchers wanting to
supplement their analyses with qualitative data. Identifying information, such as the name
of the residential area, school, organization, or even the respondent is required if supple-
mental qualitative investigations (e.g., observation, archival research, focus groups, or fol-
low-up interviews) are to be conducted.
It is true that many publicly available datasets have restricted-use versions that contain
potentially identifying informationmostly geographic in nature. To use such geo-coded
data, researchers must either apply for a license to obtain the data for a speciWed duration
or sign a contract of agreement to access the data at the site where it is collected and man-
aged. The NLS, PSID, and AddHealth datasets all have conWdential geo-coded data (often
including state, county, metropolitan area of residence, or zip code) available to
3
According to the ICPSR website, Each data collection in the ICPSR archive has been examined to ensure
that the contents of the collection do not violate explicit or implicit pledges of conWdentiality given to respondents
or research subjects. Data items that could be used to identify individual respondents are typically removed,
masked, or collapsed (for more information see http://www.icpsr.umich.edu/ACCESS/restgeneral.html). Because
of the longitudinal nature and type of data (often videotapes, audiotapes, and pictures) held at the Henry Murray
Research Center at RadcliVe, this data archive center has a less stringent policy about the removal of indirect
identiWers. At the discretion of the principal investigator, they allow other researchers to re-contact research par-
ticipants for follow-up study.
Fig. 1. ASR and AJS Articles using Primary Data: 19352000.
p
e
r
c
e
n
t
a
g
e
year
1940 1950 1960 1970 1980 1990 2000
40
50
60
70
154 E. Leahey / Social Science Research 36 (2007) 149158
researchers who apply in writing for a license to access, properly use, and protect such data.
Some of the larger datasets housed at ICPSR also have restricted-use versions available.
The restricted geo-codes can be used to match respondent survey data with, for example,
more aggregate level census data about the surrounding area.
However, even restricted-use data is insuYcient for a researcher planning a mixed
method investigation. Even if researchers gain access to restricted-use geographical infor-
mation, they are not permitted to use that information for non-statistical purposes. Non-
statistical purposes include many of the steps required for mixed method investigations:
deducing the identity of a few participants who seemed to provide contradictory responses
to several survey questions in order to re-interview them; determining the identity of a par-
ticular school or organization included in the study so that additional information about
the institution can be obtained through archival work, key-informant interviews, or focus
groups; and visiting a particular geographic area to conduct supplementary Weld work.
Using identifying information for non-statistical purposes is also forbidden with any data-
set housed at ICPSR, whose bylaws state that data are to be used solely for statistical
analysis and reporting of aggregated information, and not for investigation of speciWc indi-
viduals or organizations (ICPSR Bylaws, Article I, Section 2.C). These restrictions serve
as a critical safeguard against breaches of participants conWdentiality.
2. Unrecognized limitations
Given the declining use of primary data collection and increasing concerns for human
research subjects conWdentiality (Singer and Levine, 2003), it is surprising that the limits to
implementing a mixed method design that begins survey data have not been recognized.
Even researchers who detail the procedures of this kind of mixed method design overlook
its prerequisites, which are not available to all. Pearces (2002) proposal requires both a
quantitative identiWcation of anomalous cases and follow-up interviews with original
respondents, but she does not address the extent to which these requirements can be met by
other researchers, particularly those not involved with the original data collection.
4
Pearce
did not recognize that the beneWts of using methodological convergence to improve her sci-
entiWc understanding of the phenomenon were not transferable to many researchers situa-
tions, as two of the three beneWts of method integration described in her conclusion (p.
129) apply only to researchers who are involved in the initial data collection eVorts. First,
she states that a large-scale survey can provide a useful sampling frame from which to sys-
tematically select deviant cases as ethnographic informants, but this is only true if
researchers have access to the surveys sampling frame. Second, she states that access to
informant characteristics measured in the survey data enables researchers to learn valuable
information prior to observations or interviews; this too is only beneWcial if researchers
have access to information that identiWes geographic areas and/or respondents. Although
Pearces results support the call for continued creativity in the design of multi-method
projects in the advancement of a holistic social science (p. 106), there is a grave limitation
to the extent to which researchers can contribute to such creativity.
To further support this claim, I reviewed all of the works listed in Tashakkori and Tedd-
lies (2003, p. 623) chronology of works on mixed methods research: eight books, Wve book
4
Even for those involved with primary data collection, the original informed consent form given the research
participants must state that contact for possible follow-up research is permitted.
E. Leahey / Social Science Research 36 (2007) 149158 155
chapters, and nine journal articles. Like the texts cited earlier in this paper, most of these
works focus on the idea of a mixed method approach to social research and/or the advan-
tages inherent in mixed method designs. Some address disadvantages, such as the diYculty
of replicating such studies, and costs in terms of time and money that mixed method
approaches incur (Datta, 1994; Jick, 1979; Newman and Benz, 1998, p. 84) . A few others
mention ethical issues in research, but not ethical issues that might arise speciWcally with
the implementation of mixed method approaches (e.g., see Creswell, 2003, pp. 6466). Chen
(1997) questions the usefulness and potential dominance of mixed methods primarily for
theoretical, but not practical, reasons: his brief discussion of practical limitations does not
mention the availability of identifying data or concerns about conWdentiality.
Only one of these 22 works, Brewer and Hunters (1989, p. 194) Multimethod Research:
A Synthesis of Styles, brieXy alludes to the limitation I have proposed:
Finally, we must address an ethical problem posed by multimethod research. This
problem stems from the ability of multimethod research to identify and combine a
variety of discrete data points from diVerent methods, thereby linking information
about individuals and groups that could not be linked if the methods were used
separately. This is [a] . kind of threat to the invasion of privacy.
Brewer and Hunter cite Humphreys (1970) research on public sexual encounters as an
example of the heightened ethical concerns that researchers conducting mixed method
research face (recall that Humphreys engaged in observation in tearooms as well as
subsequent household survey eVorts). They do not mention, however, the additional obsta-
cles confronted by users of secondary survey data, from whom participants personal iden-
tiWers must be withheld to ensure conWdentiality.
Even the most cautious and critical discussions of mixed method research neglect to
mention the issue of accessibility and availability that I raise here. Bryman (1992, p. 62)
presents his concerns regarding methodological integration, with the hopes of injecting a
sense of caution into the somewhat over-enthusiastic discussions, but the diYculties of
pursuing a mixed method design encountered by a researcher who starts with secondary
survey data are not among them. Jick (1979, p. 610) notes that triangulation is a strategy
that may not be suitable for all research purposesvarious constraints may prevent its
eVective use; but does not elaborate on constraints besides noting the additional time
costs. Datta (1982, p. 141) comes close to recognizing this limitation when she wonders why
so few of the multi-site and multi-method studies that dominate evaluations of educational
programs have involved national surveys. Her answer focuses on the situational charac-
ter of educational programs and successes, rather than on the feasibility of beginning with
a large-scale national survey and supplementing it with qualitative data collection and
analysis, which, as I have argued, is limited by essential concerns for the conWdentiality of
research participants.
While on the surface this obstacle seems to apply to any study that involves repeated
contact with research participants, it is rarely encountered in mono-method studies.
5
If a
researcher begins her quantitative analysis using a cross-sectional secondary dataset and
pines for another wave of data collectioneven one using the same methodshe will
5
I limit this discussion to quantitative mono-method studies. Data from qualitative mono-method studies that
involve repeated contact with individuals are rarely archived or shared with others outside the research team,
making the problem of accessing identiWers irrelevant.
156 E. Leahey / Social Science Research 36 (2007) 149158
certainly face the same obstacle as a researcher who hopes to supplement the analysis with
a qualitative component. However, most researchers know at the outset whether their
research question requires cross-sectional or longitudinal data and will choose a secondary
dataset appropriately. If it happens to require longitudinal data, identiWcation numbers
permit researchers to link research participants over time without knowing their identity.
Because many of the touted functions and advantages of mixed methods cannot always
be anticipated at the outset of a study, the limitation I have identiWed is particularly rele-
vant to mixed method designs that supplement survey data and analysis with qualitative
research. In such projects, qualitative data are typically used to adjudicate between com-
peting interpretations of quantitative results, Xesh out surprising Wndings, or explore devi-
ant cases; they are not typically part of the initial research design (Bryman, 1992, p. 67) .
Rather, supplemental qualitative data and analysis are integrated at a later date, on an as-
needed basis. For example, Pearce (2002) turned to qualitative investigation after Wnding
that her statistical model could not adequately capture the variation she found across her
cases. Entwisle (forthcoming) turned to qualitative techniques when extant coding categories
were too detailed to digest and when statistical models appeared, based on goodness-of-Wt
tests, to be incorrectly speciWed. Because the advantages of supplementing survey-based
analyses with a qualitative component are most apparent after the quantitative work has
begun, the limitation I have identiWed is more pronounced in mixed method studies than in
other studies involving follow-up. The contributions of qualitative researchers who design
research from the outset to elaborate and improve quantitative research (Lavin and May-
nard, 2001; Maynard and SchaeVer, 2000) are not subject to this limitation.
6
Few researchers have the foresight to know that their quantitative analyses might pro-
duce surprising Wndings, suggest deviant cases, or allow for conXicting interpretations. If
they did, perhaps they could design a multi method research project from the outset, col-
lecting both quantitative survey data and more qualitative information, as William Julius
Wilson did for his project culminating in When Work Disappears (1996). But unlike Pearce
and Entwisle et al., many researchers begin their analyses with secondary survey data, and
therefore do not have the option to add a qualitative component (at least not in a way that
is linked to the survey data)
7
even when it could be useful and informative. Thus, the appli-
cation of a popular and potentially dominant methodology is actually quite limited.
Although moving from a lone-scholar model to a more team-oriented approach to
research would open the possibility of mixed methods to more researchers, the necessity of
limiting research team sizes and of protecting participants conWdentiality suggests that
limited access to this kind of mixed method design will remain an issue for some time. And
in any case, selective access to mixed method collaborative teams is likely to perpetuate the
structure of privilege I have identiWed.
Given that research methods are techniques for investigation, their limitationsboth
scientiWc and practicalshould be recognized and explained, especially by researchers
expounding their beneWts, and especially in texts devoted to mixed method approaches.
Though drawing attention to this practical limitation is an important Wrst step, research is
clearly needed to help resolve the tension between giving researchers greater access to data
6
I thank an anonymous reviewer for making this point.
7
The addition of qualitative data that is not linked to the original quantitative data may also be able to aid the
interpretation of surprising quantitative Wndings (see, for example, Sutton and Rafaeli (1988)), but cannot help re-
searchers explore deviant cases (Pearce, 2002) or devise better measures (Entwisle, forthcoming). The limitations
highlighted in this research apply only to studies that require linked qualitative and quantitative data.
E. Leahey / Social Science Research 36 (2007) 149158 157
collected by others and protecting the conWdentiality of those providing the data (Singer
and Levine, 2003, p. 163).
References
Alford, R.R., 1998. The Craft of Inquiry: Theories, Methods, Evidence. Oxford University Press, New York.
Bourdieu, P., Wacquant, L.J.D., 1992. An Invitation to ReXexive Sociology. University of Chicago Press, Chicago.
Brewer, J., Hunter, A., 1989. Mulitmedia Research: A Synthesis of Style. Sage, Newbury Park, CA.
Bryman, A., 1989. Quantity and Quality in Social Science Research. Routledge, London.
Bryman, A., 1992. Quantitative and qualitative research: further reXections on their integration. In: Julia, B. (Ed.),
Mixing Methods: Qualitative and Quantitative Research. Avebury, BrookWeld, VT.
Caracelli, V.J., Greene, J.C., 1993. Data analysis strategies for mixed-method evaluation designs. Education Eval-
uation and Policy Analysis 15, 195207.
Chen, H., 1997. Applying mixed methods unders the framework of theory-driven evaluations. In: Gary, T.H., Jen-
nifer, C.G. (Eds.), Advances in Mixed-Method Evaluation: The Challenges and BeneWts of Integrating Diverse
Paradigms. Jossey-Bass, San Francisco.
Clemens, E.S., Powell, W.W., McIlwaine, K., Okamoto, D., 1995. Careers in print: books, journals, and scholarly
reputations. American Journal of Sociology, 101.
Coleman, J.S., 1986. Social theory, social research, and a theory of action. American Journal of Sociology 91,
13091335.
Cook, T.D., Reichardt, C.S. (Eds.), 1979. Qualitative and Quantitative Methods in Evaluation Research. Sage,
Beverly Hills, CA.
Creswell, J.W., 2003. Research Design: Qualitative, Quantitative, and Mixed Method Approaches. Sage, Thou-
sand Oaks, CA.
Datta, L., 1982. Strange bedfellows: the politics of qualitative methods. American Behavioral Scientist 26, 133144.
Datta, L., 1994. Paradigm wars: a basis for peaceful coexistence and beyond. In: Reichardt, C.S., Rallis, S.F.
(Eds.), The Qualitative-Quantitative Debate: New Perspectives. Jossey-Bass, San Francisco.
Denzin, N.K., 1988. The Research Act. McGraw-Hill, New York.
Entwisle, B., Bouma, J.E., Short, S.E. Henderson, G.E., Fengying. Z., forthcoming. Designing Points of Conver-
gence: Methods for Integrating Qualitative Approaches into Quantitative Analysis. University of California
Press, Berkeley, CA.
Geertz, C., 1979. From the natives point of view: on the nature of anthropological understanding. In: Rabinow,
P., Sullivan, W. (Eds.), Interpretive Social Science. University of California Press, Berkeley.
Giddens, A., 1976. New Rules of Sociological Method. Basic Books, New York.
Hughes, E.C., 1984. The Sociological Eye: Selected Papers. Transaction Books, New Brunswick.
Humphreys, L., 1970. Tea Room Trade: Impersonal Sex in Public Places. Aldine Publishing Company, Chicago.
Jick, T.D., 1979. Mixing qualitative and quantitative methods: triangulation in action. Administrative Science
Quarterly 24, 602611.
King, G., Keohane, R.O., Verba, S., 1994. Designing Social Inquiry: ScientiWc Inference in Qualitative Research.
Princeton University Press, Princeton, NJ.
Lavin, D., Maynard, D.W., 2001. Standardization vs. rapport: respondent laughter and interviewer reaction dur-
ing telephone surveys. American Sociological Review 66, 453479.
Leahey, E., 2005. Overseeing Research Practice: The Case of Data Editing. Unpublished manuscript.
Maynard, D.W., SchaeVer, N.C., 2000. Toward a sociology of social scientiWc knowledge: survey research and eth-
nomethodologys asymmetric alternates. Social Studies of Science 30, 323370.
Newman, I., Benz, C.R., 1998. Qualitative-Quantitative Research Methodology: Exploring the Interactive Contin-
uum. University of Illinois Press, Carbondale.
Pearce, L.D., 2002. Integrating survey and ethnographic methods for systematic anomalous case analysis. Socio-
logical Methodology 32, 103132.
Platt, J., 1996. A History of Sociological Research Methods in America. Cambridge University Press, Cambridge,
England.
Pontille, D., 2003. Authorship practices and institutional contexts in sociology. Science, Technology, and Human
Values 28, 217243.
Presser, S., 1994. Presidential address: informed consent and conWdentiality in survey research. Public Opinion
Quarterly 58, 446459.
158 E. Leahey / Social Science Research 36 (2007) 149158
Singer, E., Levine, F.J., 2003. Protection of human subjects of research: recent developments and future prospects
for the social sciences. Public Opinion Quarterly 67, 148164.
Sutton, R.I., Rafaeli, A., 1988. Untangling the relationship between displayed emotions and organizational sales:
the case of convenience stores. Academy of Management Journal 31, 461487.
Tashakkori, A., Teddlie, C., 1998. Mixed Methodology: Combining Qualitative and Quantitative Approaches.
Sage, Thousand Oaks, CA.
Tashakkori, A., Teddlie, C., 2002. Handbook of Mixed Methods in the Social and Behavioral Research. Sage,
Thousand Oaks, CA.
Tashakkori, A., Teddlie, C., 2003. Handbook of Mixed Methods in the Social and Behavioral Research. Sage,
Thousand Oaks, CA.
Wilson, W.J., 1996. When Work Disappears: The World of the New Urban Poor. Knopf, New York.

You might also like