View sample Secondary Analysis Research Paper. Browse other research paper examples and check the list of research paper topics for more inspiration. If you need a religion research paper written according to all the academic standards, you can always turn to our experienced writers for help. This is how your paper can get an A! Feel free to contact our research paper writing service for professional assistance. We offer high-quality assignments for reasonable rates.
Secondary analysis refers to a set of research practices that involve utilizing data collected by someone else or data that has been collected for another purpose (e.g., administrative records). It is used, to varying degrees, across a wide range of disciplines and throughout the world. Given this breadth, it is not surprising that it has taken on numerous forms. It is also conducted for several distinct reasons. Although not a research methodology, per se, several features distinguish it from other research activities. In turn, these features create opportunities and limitations for the secondary analyst. Central issues concern: (a) data availability, access, and documentation; (b) maintaining conﬁdentiality and privacy pledges made by primary researchers; and (c) proprietary rights and data ownership.
1. Secondary Analysis As A ‘Methodology’
The use of secondary data in behavioral and social sciences is ubiquitous, appearing in a number of traditional (i.e., quantitative) and untraditional (i.e., qualitative) forms. Despite its rather extended roots in the social and behavioral sciences, it is not widely celebrated as a method. As partial evidence of its relative obscurity, consider the fact that between 1978 and the present, a search of PsychInfo uncovered only 36 articles and books using the keyword ‘secondary analysis.’ Furthermore, the phrase ‘secondary analysis’ appeared in fewer than half of the article or book titles. Books on the topic are also relatively scarce; fewer than a dozen were uncovered through the same search (e.g., Boruch et al. 1981, Elder et al. 1993, Hyman 1972, Stewart 1984).
Ironically, the apparent obscurity of secondary analysis as a methodology is due to its pervasiveness. That is, the use of secondary data sources is so commonplace in many ﬁelds (e.g., economics, education, and sociology) that there is little need in calling attention to it as a separable methodology. This makes sense because unlike ethnography, survey research, or quasi-experimentation which each have distinctive methodological procedures and practices, secondary analysis does not involve a new or diﬀerent set of tactics. Even from a statistical point of view, there is little to distinguish it from primary analysis, and most of the measurement, design, and statistical issues facing the secondary analyst are largely the same as those faced by a primary analyst. The obvious exception is that the secondary analyst is constrained by scope and nature of the design (e.g., the sample, sample sizes, attrition, missing data, measures, and research design) inasmuch as these have been speciﬁed by someone else (McCall and Appelbaum 1991). As such, secondary analysis boils down to a data resource, not a methodology, per se. However, the use of secondary data—especially when micro data records are used—does involve a unique set of logistical, ethical and practical considerations.
2. Varieties Of Secondary Analysis
Because primary data can assume a number of forms (e.g., data based on cross-sectional surveys, administrative records, panel surveys, observations), secondary analysis has taken on a variety of forms. Two main categories can be identiﬁed: (a) some data are developed for the explicit purpose of serving as a national resource; and (b) other forms of data are byproducts of the actions of individuals and organizations. The latter are included in the deﬁnition of secondary analysis because they play a large role in contemporary research. Given the increased use of the Internet, more administrative records, facts, and statistics (public and private) also will be readily available for use in research.
2.1 Traditional Forms Of Secondary Analysis
The most common forms of secondary data include population censuses, continuous or regular surveys, national cohort studies, multisource data sets, administrative records, and one-time surveys or research studies. In the US, interest in secondary analysis was prompted by the appearance of public opinion polls and surveys in the 1940s and 1950s, and continued in the 1960s, when the federal government began, in earnest, gathering data through large-scale surveys based on representative sampling (Hyman 1972).
Of particular interest to social science researchers are the large scale, nationally representative, longitudinal (panel) surveys. In the USA, even cursory literature searches reveal hundreds of publications in economics and sociology that use data drawn from, for example, the Panel Study of Income Dynamics (PSID), the National Longitudinal Survey of Youth (NLSY), and the General Social Survey (GSS). In the United Kingdom, the General Household Survey (GHS) has been conducted annually since the beginning of the 1970s. The reuse of large-scale survey data is also evident across many nations simultaneously, as when Ravillion and Chen (1997) examined the correspondence between the rate of growth in gross domestic product and changes in poverty rates. The willingness of domestic and foreign governments to invest in data collection prompted Cherlin (1991) to speculate that the term secondary analysis will become obsolete. Many large-scale surveys are being sponsored by governments with no primary data analyses in mind.
Whereas large-scale surveys like the PSID are deliberately ﬁelded to answer a multitude of research questions about populations and subgroups, substantial data gathering also is undertaken by governments as byproducts of their operation, to monitor their own processes and outcomes, or to assess environmental conditions or events (e.g., weather, rainfall). These are generally referred to as administrative or governmental records. The use of governmental records in research has a long tradition; over 100 years ago, Emile Durkheim used government statistics to examine some of the causes of suicide (Cherlin 1991).
Individual research studies also are fodder for secondary analysis. These reanalyzes are undertaken to check the accuracy of statistical conclusions in primary studies; for testing new statistical procedures or substantive hypotheses; resolving conﬂicts among researchers; for testing supplemental hypotheses suggested in primary analyses. The latter are usually conducted by the primary analyst, and this category represents the prevalent use of the term ‘secondary analysis.’ In recent years, secondary analysts reanalyzed multiple studies that use the same instrumentation, a tactic more akin to the spirit of metaanalysis.
A particularly important role for secondary analysis of individual studies is testing or demonstrating the superiority of new statistical methods. A thoughtful example is provided by Muthen and Curran (1997). They showed that latent growth curve models produced larger intervention eﬀects than had been previously reported by exerting greater control over extraneous sources of error. More generally, in the past 25 years, eﬀorts to address causal questions with extant data have produced substantial advances in statistical modeling (Duncan 1991). And, reanalysis of data from program evaluation studies has been undertaken to assure policy-makers that the results of primary studies are technically sound (Boruch et al. 1981).
2.2 Less Traditional Forms Of Secondary Analysis
Although not a traditional form of secondary analysis, Webb et al. (1965) demonstrated that a substantial amount of ‘data’ are produced by individuals and organizations (public and private) as byproducts of their daily transactions. In their now classic text Unobtrusi e Measures: Nonreacti e Research in the Social Sciences, they identiﬁed a host of unconventional ways in which researchers can reuse existing data. In making their case, they showed how physical traces (erosion and accretion); the content of running records (e.g., mass media, judicial records), private records (sales records, credit card purchases), and contrived observations (e.g., hidden cameras) can be used as sources of data in research. The literature is ﬁlled with studies based on creative uses of these artifacts. Examples include assessing: the type, amount, and ‘quality’ of household garbage to assess the dietary and recycling habits of Arizona residents; the amount of trash left in the streets by revelers at Marti Gras to estimate the size of the daily crowd; and the diﬀerential wear and tear seen in separate sections to ascertain the popularity of sections of the International Encyclopedia of the Social Sciences. Finally, whereas secondary analysis has been historically viewed as a quantitative endeavor, reanalysis of qualitative data is now regarded as a legitimate part of the enterprise.
It would appear that Webb et al. (1965) had a substantial inﬂuence on subsequent generations of primary researchers. The most underacknowledged uses of secondary data are within primary evaluation studies, where a mix of new and existing data are increasingly being used to evaluate the eﬀectiveness of social interventions. The use of mixed method evaluation designs represents a core feature of contemporary evaluation theory and practice (Lipsey and Cordray 2000).
2.3 Advantages And Disadvantages Of Secondary Analysis
Secondary analysis oﬀers several advantages over primary data collection in behavioral and social sciences, but it also has its shortcomings. On the positive side of the ledger, re-use of data is eﬃcient. These eﬃciencies include: (a) replication (or not) of ﬁndings across investigators; and (b) the discovery of biases in conventional statistical methods. Both of these beneﬁt science by winnowing false hypotheses more quickly and oﬀering new evidence (estimates) in their place. Testing new hypotheses, beyond those envisioned by the data developers, add to the eﬃciency of knowledge acquisition. These beneﬁts have served as partial justiﬁcation for the cost of conducting large scale, nationally representative surveys (using cross-sectional and panel designs) that query individuals on a wide array of topics. Some topics involving special populations (e.g., twins) or long time frames cannot be investigated without reliance upon archives and data sharing among investigators.
Alternatively, secondary data impose limits on what can be studied, where, and over what period of time. McCall and Appelbaum (1991) suggest using a Feasibility Matrix of Sample × Measure × Assessment Age as a tool for prescreening the potential utility of secondary data sets. In addition, technical problems (e.g., selection biases, sample attrition, and missing data) can be so severe as to limit the value of the primary data set. As such, assessing data quality probably needs to be included in the McCall– Appelbaum matrix. The existence of data can shortcircuit one facet of the scientiﬁc process, tempting the analyst to ‘mine’ the data rather than initiate analyzes based on a theory or hypothesis. Although interesting results may emerge, such practices can lead to unreliable ﬁndings or ﬁndings limited to a single operationalization. The extent to which these problems in secondary analysis have inﬂuenced knowledge development is unknown. But, a mismatch between theory and data can be avoided with proper consideration of the relevance and quality of each data source.
3. What Is Unique About Secondary Analysis?
Using data generated by someone else does raise several issues that make secondary analysis somewhat unique. Obviously, secondary analysis is possible only when it is available, easy to access, and in a form that is usable. Availability and access can be facilitated or impeded by a number of factors. In particular, because data do not ‘speak for themselves,’ they must be well documented. Data from government-sponsored, large surveys, panels and so on are often routinely archived and well documented. This is not uniform across all types of secondary data. Data sharing among individual investigators can become contentious, especially in light of questions about proprietary rights and data ownership. Researchers are obliged to honor original pledges of conﬁdentiality and privacy. Balancing the desire to share data with these ethical requirements may require conﬁguring the data in a way that may limit how it is disclosed and the methodological options available to the secondary analyst.
3.1 Availability, Access And Documentation
Establishment of data archives, advances in computer technology, and the emergence of the World Wide Web (WWW) have greatly facilitated access to data and solved some of the early problems that plagued the ﬁrst generation of secondary analysts (e.g., poor documentation, noncommon data formats and language). Since the 1960s, the Interuniversity Consortium for Political and Social Research (ICPSR) has functioned as a major repository and dissemination service in the US. The National Archives have served a similar function for some governmental data. Increasingly, these roles have been devolved to individual governmental agencies that have developed skills in storing and disseminating their own data. In addition, the Internet has the potential for revolutionizing the use and transfer of data.
Secondary analysis also has been institutionalized within journal and professional codes of conduct. To facilitate access to data appearing in scientiﬁc journals, many journals have adopted policies whereby contributing authors are expected to make the data from their studies available for a designated period (usually three to ﬁve years) of time. Similarly, professional associations (e.g., the American Psychological Association) have incorporated data sharing into their codes of ethical behavior. Whereas archives and agencies require that data be properly documented, journals and professional associations are generally silent on this aspect of the data sharing process. Because data documentation is a process that records analytic decisions as they unfold over the course of the study, primary study authors need to be aware of these nontrivial editorial and ethical demands (see Boruch et al. 1981).
3.2 Ethical Considerations And Disclosure
The need to protect the conﬁdentiality and privacy of research participants is sometimes at odds with the desire to make data available to others. Resolving these competing values requires careful consideration at the time that primary research is conducted, documented, stored, and disseminated. Concealing the identity of participants can often be accomplished by removing personal identiﬁers from the data ﬁle. To the extent that identiﬁcation is still possible through deductive disclosure (combining information to produce a unique proﬁle of an individual), alternative procedures are needed. If identiﬁers are needed to link records (as in longitudinal research), additional layers of protection are needed. Fortunately, Boruch and Cecil (1979) provide a comprehensive treatment of the available statistical procedures (e.g., inoculating raw data with known amounts of error, randomized response techniques, collapsing categories) and institutional procedures (e.g., third parties who would serve as ‘data brokers’) that can be used to relax these problems. Data sharing among individual researchers requires explicit attention to ethical, statistical, organizational, and logistical (e.g., video editing and image modiﬁcation) issues throughout the research process.
3.3 Proprietary Rights And Data Ownership
Changing policies, practices, and technology will facilitate data sharing and, as a consequence, increase use of data collected by someone else. Alternatively, secondary analysis of data for the purposes of addressing disputes among analysts can be quite diﬃcult to negotiate. When conﬂict arises, it often revolves around who ‘owns’ the data and if, when, how and how much of it should be disclosed to others. Establishing proprietary rights to research and evaluation data is not a simple matter. Data ownership depends on how the research was ﬁnanced, policies of the sponsoring and host organizations, conditions speciﬁed in laws (e.g., Freedom of Information Act, Privacy Act), data sharing polices of the journal in which the research is published, and ethical guidelines of professional associations to which authors belong. As researchers embark on primary or secondary analyses, it is necessary to understand these avenues and constraints.
4. A Summary Note
Access to quantitative and qualitative data from governments, businesses, and individual researchers has greatly facilitated the practice of secondary analysis. Changes in information technology—notably greater use of the World Wide Web—will undoubtedly enhance these practices even further at primary and secondary levels of research and evaluation.
- Boruch R F, Cecil J S 1979 Assuring the Conﬁdentiality of Social Research Data. University of Pennsylvania Press, Philadelphia, PA
- Boruch R F, Wortman P M, Cordray D S and Associates (eds.) 1981 Reanalyzing Program Evaluations: Policies and Practices for Secondary Analysis of Social and Educational Programs, 1st edn. Jossey-Bass, San Francisco, CA
- Cherlin A 1991 On analyzing other people’s data. Developmental Psychology 27(6): 946–8
- Duncan G J 1991 Made in heaven: Secondary data analysis and interdisciplinary collaborators. Developmental Psychology 27(6): 949–51
- Elder Jr. G H, Pavalko E K, Clipp E C 1993 Working with Archival Data: Studying Lives. Sage Publications, Newbury Park, CA
- Hyman H H 1972 Secondary Analysis of Sample Surveys: Principles, Procedures and Potentialities. Wiley, London
- Lipsey M W, Cordray D S 2000 Evaluation methods for social intervention. Annual Review of Psychology 51: 345–75
- McCall R B, Appelbaum M I 1991 Some issues of conducting secondary analysis. Developmental Psychology 27(6): 911–17
- Muthen B O, Curran P J 1997 General longitudinal modeling of individual diﬀerences in experimental designs: A latent variable framework for analysis and power estimation. Psychological Method 2: 371–402
- Ravillion M, Chen S H 1997 What can new survey data tell us about recent changes in distribution and poverty? World Bank Economic Review 11: 357–82
- Stewart D W 1984 Secondary Research: Information Sources and Methods. Sage, Beverly Hills, CA
- Webb E J, Campbell D T, Schwartz R D, Sechrest L 1965 Unobtrusive Measures: Nonreactive Research in the Social Sciences. Rand McNally, Chicago, IL