Tips for Collecting, Reviewing, and

Analyzing Secondary Data

The specific types of information and/or data
needed to conduct a secondary analysis will
WHAT IS SECONDARY DATA REVIEW depend, obviously, on the focus of your study.
AND ANALYSIS? For CARE purposes, secondary data analysis is
Secondary data analysis can be literally defined usually conducted to gain a more in-depth
as “second-hand” analysis. It is the analysis of understanding of the causes of poverty in the
data or information that was either gathered by various countries and/or regions where CARE
someone else (e.g., researchers, institutions, works. Secondary data review and analysis
other NGOs, etc.) or for some other purpose involves collecting information, statistics, and
than the one currently being considered, or other relevant data at various levels of
often a combination of the two (Cnossen 1997). aggregation in order to conduct a situational
analysis of the area (see Data & Indicator List
If secondary research and data analysis is in Appendix 1; refer to the LRSP Guidelines,
undertaken with care and diligence, it can Annex 5, July 1997). The following is a
provide a cost-effective way of gaining a broad sampling of the types of secondary data and
understanding of research questions. information commonly associated with poverty
analysis:
Secondary data are also helpful in designing • Demographic (population, population
subsequent primary research and, as well, can growth rate, rural/urban, gender, ethnic
provide a baseline with which to compare your groups, migration trends, etc.),
primary data collection results. Therefore, it is • Discrimination (by gender, ethinicity,
always wise to begin any research activity with age, etc)
a review of the secondary data (Novak 1996). • Gender equality (by age, ethnicity, etc)
• Policy environment
• Economic environment (growth, debt
RESEARCH DESIGN AND PURPOSE ratio, terms of trade)
Secondary data analysis and review involves • Poverty levels (poverty and absolute),
collecting and analyzing a vast array of • Employment and wages (formal and
information. To help you stay focused, your informal; access variables),
first step should be to develop a statement of • Livelihood systems (rural, urban, on-
purpose – a detailed definition of the purpose of farm, off-farm, informal, etc),
your research – and a research design. • Agricultural variables and practices
(rainfall, crops, soil types, and uses,
Statement of Purpose: Having a well-defined irrigation, etc.),
purpose – a clear understanding of why you are • Health (malnutrition, infant mortality,
collecting the data and of what kind of data you immunization rate, fertility rate,
want to collect, analyze, and better understand – contraceptive prevalence rate, etc.),
will help you remain focused and prevent you • Health services (#/level, services by
from becoming overwhelmed with the volume level, facility-to-population ratio;
of data. access by gender, ethnicity. etc.),
• Education (adult literacy rate, school
Research Design: A research design is a step- enrollment, drop-out rates, male-to-
by-step plan that guides data collection and female ratio, ethnic ratio, etc),
analysis. In the case of secondary data reviews • Schools (#/level, school-to-population
it might simply be an outline of what you want ratio, access by gender, ethnicity, etc.),
the final report to look like, a list of the types of • Infrastructure (roads, electricity,
data that you need to collect, and a preliminary communication, water, sanitation,
list of data sources. etc.),
• Environmental status and problems
• Harmful cultural practices

Special attention should be given to collecting

disaggregated data. That is, data that is broken
WHAT TYPES OF DATA AND/OR down in the following ways: gender, age,
INFORMATION ARE NEEDED? ethnicity, location, etc..
M. Katherine McCaston, HLS Advisor June 2005
Updated from M.Katherine McCaston (1998) -Partnership & Household Livelihood Security Unit 1
 Scholarly Journals: Scholarly journals
Even when highly disaggregated; however, generally contain reports of original research or
these “raw” data points alone are often only experimentation written by experts in specific
static or indirect measures of the situation or fields. Articles in scholarly journals usually
problems that exist in countries and regions – undergo a peer review where other experts in
partial or imperfect reflections of reality (UNDP the
1997). It is through reviewing, interpreting, and same field review the content of the article for
cross-analyzing the secondary that these pieces accuracy, originality, and relevance.
of information allow us to gain a better
understanding of a specific situation, Literature Review Articles: Literature review
population, sector, etc. Analysis of data gives articles assemble and review original research
you the information that you need to make dealing with a specific topic. Reviews are
judgements, recommend areas of intervention, usually written by experts in the field and may
and/or design follow-up studies. Cross- be the first written overview of a topic area.
analyzing data will also help you understand not Review articles discuss and list all the relevant
only what is happing in a particular area but publications from which the information is
also WHY it is happening. derived.

Trade Journals: Trade journals contain articles

SOURCES OF SECONDARY DATA that discuss practical information concerning
Official Statistics: Official statistics are various fields. These journals provide people in
statistics collected by governments and their these fields with information pertaining to that
various agencies, bureaus, and departments. field or trade.
These statistics can be useful to researchers
because they are an easily obtainable and Reference Books: Reference books provide
comprehensive source of information that secondary source material. In many cases,
usually covers long periods of time. specific facts or a summary of a topic is all that
is included. Handbooks, manuals,
However, because official statistics are often encyclopedias, and dictionaries are considered
“characterized by unreliability, data gaps, over- reference books (University of Cincinnati
aggregation, inaccuracies, mutual Library 1996; Pritchard and Scott 1996).
inconsistencies, and lack of timely reporting”
(Gill 1993), it is important to critically analyze
official statistics for accuracy and validity. WHERE TO FIND SECONDARY DATA
There are several reasons why these problems There are numerous sources of secondary data
exist: and information. The first step in collecting
secondary data is to determine which
1. The scale of official surveys generally institutions conduct research on the topic area
requires large numbers of enumerators or country in question.
(interviewers) and, in order to reach those
numbers enumerators contracted are often Large surveys and country-wide studies are
under-skilled; expensive and time-consuming to conduct;
2. The size of the survey area and research therefore, they are usually done by governments
team usually prohibits adequate supervision or large institutions with a research orientation.
of enumerators and the research process; Thus, government documents and official
and statistics are a good starting place for gathering
3. Resource limitations (human and technical) secondary data; however, as previously stated,
often prevent timely and accurate reporting the quality of the documents will vary
of results. depending on the country of study and the
amount of resources dedicated to data
Technical Reports: Technical reports are collection.
accounts of work done on research projects.
They are written to provide research results to
colleagues, research institutions, governments,
and other interested researchers. A report may
emanate from completed research or on-going Make Use of Local Experts
research projects.
When searching for secondary data
or questioning the quality of a
M. Katherine McCaston, HLS Advisor June 2005 source that you have already
collected,
Updated from M.Katherine McCaston (1998) -Partnership & Household seek advice
Livelihood Security Unit from sector 2
specialists and other experts in your
country office. Your colleagues are
valuable sources of information and
expertise.
Other major sources of international
development data are the World Bank, the
United States Agency for International
Development (USAID), the United Nations
Development Programme (UNDP), the Food
and Agriculture Organization of the United
Nations (FAO), the International Fund for
Agricultural Development (IFAD), the World
Health Organization (WHO), International EVALUATING THE QUALITY OF YOUR
Center for Research on Women (ICRW), the INFORMATION SOURCES
Chronic Poverty Research Center (CPRC), the One of the advantages of secondary data review
Center for Research on Poverty (CROP), and analysis is that individuals with limited
Overseas Development Institute (ODI), and research training or technical expertise can be
Institute of Development Studies (IDS) to name trained to conduct this type of analysis. Key to
a few. the process, however, is the ability to judge the
quality of the data or information that has been
International development institutes commonly gathered. The following tips will help you
share information sources and have libraries for assess the quality of the data.
archiving these materials. Thus, a data-
gathering visit to one office might yield Determine the Original Purpose of the Data
numerous sources of information on the topic Collection: Consider the purpose of the data or
area of interest. publication. Is it a government document or
statistic, data collected for corporate and/or
University libraries are good sources of marketing purposes, or the output of a source
information and should be consulted. Also, it whose business is to publish secondary data
would be beneficial to establish contact with (e.g., research institutions). Knowing the
experts at local university departments that are purpose of data collection will help to evaluate
dedicated to research on the topic areas that you the quality of the data and discern the potential
are interested in (e.g., Departments of level of bias (Novak 1996).
Agricultural Sciences, Public Health,
Economics, Anthropology, Sociology). These Attempt to Ascertain the Credentials of the
experts can be important sources of information Source(s) or Author(s) of the Information
on on-going research projects as well as for What are the author’s or source’s credentials --
guiding you toward other sources of topic area educational background, past works/writings, or
information or individuals that can be experience -- in this area? For example, the
contacted. following sources are generally considered
reliable sources of data and information:
Local NGOs also often conduct empirical research reports documenting findings from
research and can be valuable sources of agricultural research published by the FAO or
information. This in particularly true when you IFAD; socioeconomic data reported by the
Secondary Data Sources
are searching for local-level information and World Bank; and survey health data reported in
data. In some cases, NGOs might also have USAID’s Demographic Health Surveys.
Government
small libraries that provideDocuments
additional
information. Official Statistics Does it include a methods section and are the
Technical Reports methods sound? Does the article have a section
Scholarly Journals that discusses the methods used to conduct the
Trade Journals study? If it does not, you can assume that it is a
Review Articles popular audience publication and should look
Reference Books for additional supporting information or data. If
Research Institutions the research methods are discussed, review
Universities
M. Katherine
Libraries,McCaston, HLS Advisor
Library Search Engines June 2005
Updated from M.Katherine McCaston (1998) -Partnership & Household Livelihood Security Unit
Computerized Databases 3
The World Wide Web
(Shell 1997)
them to ascertain the quality of the study. If you secondary data analysis, the aggregated
are not a research methods expert, have percentage figure, rather than the number of
someone else in your County Office review the “cases” reported, should be used.
methods section with you.
Another area of data analysis that requires a
What’s the Date of Publication? When was the skeptical eye is employment-related data. It is
source published? Is the source current or out- difficult to count the employed accurately,
of-date? Topic areas of continuing or rapid especially in developing countries. Employment
development, such as the sciences, demand data often do not take into account the number
more current information. of people involved in informal or unrecorded
activities, seasonal agricultural laborers,
Who is the Intended Audience? Is the women’s agricultural labor, or child labor.
publication aimed at a specialized or a general
audience? Is the source too elementary -- aimed Thus, official employment statistics should be
at the general public? viewed in light of these inadequacies. Labor
force data that provides a list of the categories
What is the Coverage of the Report or used (e.g., employed, unemployed,
Document? Does the work update other underemployed, own-account workers, unpaid
sources, substantiate other materials/reports that family workers) will help you determine the
you have read, or add new information to the quality of the measure (Worldbank 1997).
topic area?
When you feel that the employment data is
Is it a Primary or Secondary Source? Primary unreliable, looking at other economic indicators
sources are the raw material of the research will help you develop a clearer understanding of
process, they represent the records of research the situation. For example, if your employment
or events as first described. Secondary sources data state that only 25 percent of the population
are based on primary sources. These sources is economically active. However, data from a
analyze, describe, and synthesize the primary or recent poverty survey state that only 5 percent
original source. If the source is secondary, does of the population live below the absolute
it accurately relate information from primary poverty line, you can conclude that the
sources? employment data is not a good measure to use.

Importantly, Is the Document or Report Well-

Referenced? When data and/or figures are
given, are they followed by a footnote, endnote Questions for Evaluating
-- which provides a full reference for the
information at the end of the page or document
Data Quality
-- or the name and date of the source (e.g.,
Burke 1997)? Without proper reference to the • What are your source’s
source of the information, it is impossible to credentials?
judge the quality and validity of the information • What methods were used?
reported. • Is the information current or
DO THE NUMBERS DO NOT MAKE out-of-date?
SENSE? • Is the intended audience
Data reporting characteristics vary according to other researchers or the
what the data is being collected for and the general public?
stage of reporting. For example, health clinics • Is the document’s coverage
might report quarterly the number of cases of of the topic area broad or too
diarrhea, upper respiratory infection, or narrow?
malnutrition that they have been treated at a • Is it a primary or secondary
clinic. source? If it is a secondary
source, does it accurately cover
This information is useful for healthcare
professionals who will later analyze the and report on the primary
information to ascertain the percentage of the sources?
population in a municipality or province that • Does the author provide
were diagnosed with these problems over a references for the data and
given period of time. For the purpose of information reported?
• Do the numbers make
M. Katherine McCaston, HLS Advisor June 2005 sense? Are they the numbers
you want
Updated from M.Katherine McCaston (1998) -Partnership & Household Livelihood – cases
Security Unit versus 4
percentages? When compared
to related data are the
measures somewhat
consistent?
 interval, for example: the number of persons
that migrated to urban areas in the last five
years.

Disaggregated Data: These are data on

individuals or single entities, for example: age,
sex, level of education, income, occupation, etc.
These data are generally more informative and
useful than aggregate data.

There have been increasing efforts over the last

couple of decades to encourage more data
disaggregation, particularly in the international
development arena. Strong emphasis has been
placed, for example, on data disaggregation by
gender, ethnicity, age, and location. This type
WHAT DO YOU DO WHEN DATA of data enables researchers and development
SOURCES DISAGREE? practitioners to obtain a more comprehensive
When conducting secondary data analysis, it is understanding of how groups within a society
not uncommon to come across data sources that react differently to and/or are effected
disagree or conflict with each other. To help differently by various conditions or events (e.g.,
overcome this problem you should: structural adjustment, other socioeconomic
conditions, environmental degradation,
1. Decide if the source of the data is a primary government policies, development projects,
or a secondary source. In other words, look etc.).
for a citation. If the source is simply
quoting a number or statistic, it may not be For example, although women have been found
accurate, and should be taken cautiously. to be the primary agriculturists in many
societies, until recently the majority of data
2. If you cannot find the original source of the related to farming was gathered from male
data in question, look for more data sources farmers and often did not represent the same
covering the topic and determine the most reality experienced by women farmers. Thus
widely held conclusion. If two independent the lack of gender-disaggregated agricultural
secondary data sources agree, the data has been a major constraint for effective
information is probably more believable. integration of women in the planning and
implementation of agriculture and rural
development programs (FAO 1997).
3. Consult a local expert in the topic area.
Make use of the valuable resources around When conducting analysis of poverty
you. More than likely, there are colleagues determinants it is always advisable to gather
at your country office, in local government your data at the lowest possible level of
offices, or other institutions that can easily geographic/political unit aggregation (e.g.,
help resolve an issue, answer your municipality, county). This allows you to not
questions, or direct you to the answers. only ascertain what is happening at the
department or state level but also within these
areas. In some situations, you might find that
THE IMPORTANCE OF DATA the level of data disaggregation varies across or
DISAGGREGATION between political/geographic unit.
The level of data aggregation or disaggregation
simply refers the extent to which the
information or data is broken down.
Aggregate versus
Aggregate Data: Aggregate data are data that Disaggregate
describe a group of observations, with the
grouping made on a defined criterion. For In a nutshell, you can say that
example, geographic data are often grouped by
the more aggregated the data,
spatial units such as region, state, census tract,
etc. Aggregate data can also be defined by time the more invisible the people.

M. Katherine McCaston, HLS Advisor June 2005

Updated from M.Katherine McCaston (1998) -Partnership & Household Livelihood Security Unit 5
You might have nutritional data that are
disaggregated only at the department level for Therefore, cross-analyzing key indicators and
10 out of 14 departments in a country, with only using additional information sources help us
four departments having both department and understand or make reasonably sound inferences
municipal level data. In this case you would about unmeasured conditions or situations; thus
present data at the department level because it allowing us to better understand not only what
represents the broadest level of consistent data is happening and where it is happening but also
collection (i.e., you have it for all departments why it is happening.
in the country).

If time permits, it is also advisable to briefly USING SECONDARY INFORMATION TO

discuss the municipal level data for the four STRENGHTEN PRIMARY RESEARCH
departments as an example of what is happening Secondary information is also valuable for
at the municipal level. While this information generating hypotheses and identifying critical
is not generalizable at the country level, it will areas of interest that can be investigated during
enhance your knowledge of the nutritional primary data gathering activities. For example,
situation in the specific geographic secondary data analysis conducted prior to the
areas/municipalities. When complemented with Tanzania Urban Food and Household
other data (livelihood systems, levels of Livelihood Security Assessment, identified key
poverty, agroecological zone, etc.), these data research areas that should be closely studied
can help us further characterize and understand during the assessment (e.g., the influence of
the local situation and make inferences to seasonality on urban livelihoods; the impact of
higher levels of data aggregation. increased privatization; gender differentiation in
urban land tenure policies; and social cohesion
and locality). Questions were then developed
GETTING FROM WHAT AND WHERE TO and included in the household, community
WHY group, and key informant interview
Secondary data is generally referred to as questionnaires to allow analysis of these key
outcome data. This is because secondary data areas of interest.
generally describe the condition or status of
phenomena or a group; however, these data Without thorough analysis of secondary data,
alone do not tell us why the condition or status these key constraints to urban food and
exists. This limitation can be overcome in two livelihood security possibly would not have
ways. been identified, thus the problem analysis – the
why of the primary research exercise – was
First, it can be overcome by using information strengthened through secondary data and
from case studies and other research to fill in information analysis.
the gaps. For example, data on child
malnutrition rates and women’s level of
education provide information relevant for THE ADVANTAGES AND
understanding why some children are more DISADVANTAGES OF SECONDARY
likely to be malnourished than others. The DATA REVIEW AND ANALYSIS
child health research literature tells us that Advantages:
children whose mothers have low levels of • Secondary data analysis can be carried
education will likely exhibit higher malnutrition out rather quickly when compared to
rates than children of mothers with higher levels formal primary data gathering and analysis
of education. Thus, consulting relevant exercises.
literature can help illuminate causal
relationships. • Where good secondary data is
available, researchers save time and money
Second, analysis of additional key data and by making good use of available data rather
indicators can help us acquire more explanation than collecting primary data, thus avoiding
as to why a problem exists. For example, if low duplication of effort.
farm income has been identified as a problem,
data on land tenure, land size, types of crops, • Using secondary data provides a
production value, cost of inputs, and so on, can relatively low-cost means of comparing the
be compared to help identify who has this level of well-being of different political
problem and possible causes and solutions. units (e.g., states, departments, provinces,

M. Katherine McCaston, HLS Advisor June 2005

Updated from M.Katherine McCaston (1998) -Partnership & Household Livelihood Security Unit 6
 counties). However, keep in mind that data • Because the data were collected by
collection methods vary (between other researchers, and they decide what to
researchers, countries, departments, etc.), collect and what to omit, all of the
which may impair the comparability of the information desired may not be available
data. (Israel 1993).
• Much of the data available are only
• Depending on the level of data indirect measures of problems that exist in
disaggregation, secondary data analysis countries and regions (University of
lends itself to trend analysis as it offers a Cincinnati 1996).
relatively easy way to monitor change over
time. • Secondary data can not reveal
individual or group values, beliefs, or
• It informs and complements primary reasons that may be underlying current
data collection, saving time and resources trends (Beaulieu 1992).
often associated with over-collecting
primary data.
THE IMPORTANCE OF PROPERLY
• Persons with limited research training REFERENCING YOUR SECONDARY
or technical expertise can be trained to DATA REVIEW
conduct a secondary data review (Beaulieu Secondary data review and analysis is a form of
1992; University of Cincinnati 1996). research and data compilation that is demanding
and time-consuming; however, without proper
Disadvantages: citation (i.e., author, date, title) of materials that
• Secondary data helps us understand the you used, your work will often be disregarded
condition or status of a group, but as it will only have limited use by those who
compared to primary data they are wish to follow in your footsteps.
imperfect reflections of reality. Without
proper interpretation and analysis they do • A well-documented secondary data
not help us understand why something is review and analysis allows for easier use of
happening. the material by other interested parties.

• The person reviewing the secondary data can • Properly citing the publication date of
easily become overwhelmed by the volume the sources you used will allow subsequent
of secondary data available, if selectivity is researchers to use your work to make
not exercised. comparisons over time and between
countries, communities, towns, regions, etc.
• It is often difficult to determine the quality of
some of the data in question. • Proper citation allow subsequent
researchers to use your work, thus
• Sources may conflict with each other. preventing unnecessary duplication of
research efforts (Qualidata 1997)
• Because secondary data is usually not
collected for the same purpose as the
original researcher had, the goals and
purposes of the original researcher can SUMMARY
potentially bias the study.
Secondary data can be a valuable source of
information for gaining knowledge and insight
Secondary versus into a broad range of issues and phenomena.
Primary Data Review and analysis of secondary data can
provide a cost-effective way of addressing
issues, conducting cross-national comparisons,
Secondary data complements,
understanding country-specific and local
but does not replace, primary conditions, determining the direction and
data collection and should be magnitude of change -- trends, and describing
the starting place for any the current situation. It complements, but does
research activity. not replace, primary data collection and should
be the starting place for any research

M. Katherine McCaston, HLS Advisor June 2005

Updated from M.Katherine McCaston (1998) -Partnership & Household Livelihood Security Unit 7
 Pritchard, Eileen and Paula R. Scott
1996 Literature Searching in Science,
Technology, and Agriculture.
Westport, CT: Greenwood Press.

Originally Prepared by: M. Katherine Qualidata

McCaston, Food Security Advisor & Deputy 1997 Guidelines for Depositing Qualitative
Household Livelihood Security Coordinator, Data, ERSC Qualitative Data Archival
PHLS Unit, CARE, February 1998. Updated by Resource Centre. Available
same June 2005 online (telnet):
www.essex.ac.uk/qualidata

References Cited Shell, L.W.

1997 Secondary Data Sources: Library
Beaulieu, Lionel J. Search Engines, Nicholls State
1992 Identifying Needs Using Secondary University.
Data Sources, Institute of Food and
Agricultural Services, University of Trochim, William
Florida. 1997 The Knowledge Base Homepage:
Research Methods, Cornell University.
Cnossen, Christine Available online (telnet):
1997 Secondary Reserach: Learning Paper 7, trochim.human.cornell.edu/kb.
School of Public Administration and
Law, the Robert Gordon University, University of Cincinnati
January 1997. Available online 1996 Critically Analyzing Information, the
(telnet): Reference Library, University of
jura2.eee.rgu.ac.uk/dsk5/research/mate Cincinnati. Available online (telnet):
rial/resmeth www.libraries.uc.edu/libinfo
FAO
1997 User/Producer Workshop on Gender- UNDP
Disaggregated Agricultural Statistical 1997 Sustainable Livelihoods: Concepts,
Data, FAO Women in Development Principles and Approaches to Indicator
Service and FAO Regional Office for Development (Draft Discussion Paper).
Africa, Harare, Zimbabwe, September Available online (telnet):
1997. www.undp.org/seped/sl/ind2.htm.

Gill, Gerard J. Worldbank

1993 O.K., The Data’s Lousy, But It’s All 1997 World Development Indicators.
We Got (Being a Critique of Washington, D.C.: Worldbank.
Conventional Methods). London:
International Institute for Environment
and Development.

Israel, Glenn D.
1993 Using Secondary Data for Needs
Assessment, Fact Sheet PEOD-10,
Program Evaluation and
Organizational Development Series,
Institute of Food and Agricultural
Services, University of Florida.

Novak, Thomas P.
1996 Secondary Data Analysis Lecture
Notes. Marketing Research,
Vanderbilt University. Available online
(telnet):www2000.ogsm.vanderbilt.edu
/marketing.research.spring.1996.

M. Katherine McCaston, HLS Advisor June 2005

Updated from M.Katherine McCaston (1998) -Partnership & Household Livelihood Security Unit 8