Coping Strategies Index Manual Second Edition FINAL
Coping Strategies Index Manual Second Edition FINAL
Coping Strategies Index Manual Second Edition FINAL
net/publication/259999318
CITATIONS READS
95 11,021
2 authors:
Some of the authors of this publication are also working on these related projects:
All content following this page was uploaded by Daniel G. Maxwell on 11 September 2017.
Richard Caldwell
Executive Director, TANGO International
Bob Bell, Director, Food Resources Coordination Team (FRCT), CARE USA: bellr@care.org
CARE grants permission to Feinstein International Center, Tufts University, TANGO International, WFP and all
not-for-profit organizations engaged in humanitarian activities to reproduce this work, in whole or in part. The
following notice shall appear conspicuously with any reproduction: CSI Field Methods Manual, Copyright © 2008
Cooperative for Assistance and Relief Everywhere, Inc. (CARE). Used by Permission.
This publication was made possible through support provided by the Office of Food for Peace, Bureau of
Democracy, Conflict and Humanitarian Assistance, U.S. Agency for International Development and CARE, under
the terms of the Institutional Capacity Building Grant (Cooperative Agreement No. AFP-A-00-03-00017-00).
The opinions expressed herein are those of the authors and do not necessarily reflect the views of the U.S. Agency
for International Development
The Coping
Strategies
Index:
Field Methods Manual
Second Edition
Coping Strategies Index: Field Methods Manual Second Edition
TABLE OF CONTENTS
Acknowledgements
Executive Summary
1. Overview of the Coping Strategies Index
Introduction
The Basic Logic of the CSI
How Does the CSI Work?
Uses of the CSI
2. Constructing the Coping Strategies Index
a. Step 1. Coping Behaviors: Getting the Right List for the Location
b. Step 2. Frequency: Counting the Frequency of Strategies
c. Step 3. Severity: Categorizing and Weighting the Strategies
d. Step 4. Scoring: Combining Frequency and Severity for Analysis
3. Using the Coping Strategies Index
a. Interpreting the CSI Score
b. Analysis: Correlating CSI with Other Information
4. Two Different Kinds of CSI for Different Applications
a. The Original (Context-Specific) CSI
b. The Comparative (Reduced) CSI
5. Applications of CSI: Informing Decision Making
a. Quantitative and Qualitative Applications
b. Early Warning and Food Security Monitoring
c. Food Security Assessment and Targeting of Interventions
d. Monitoring and Evaluation of Interventions
e. Use in Conjunction with Other Tools
6. References
Appendix 1. Notes on Collecting and Analyzing Information
a. Sampling
b. Respondents
c. Some Miscellaneous Methodological Concerns
d. Examples of Questionnaires and CSI Module Formats
e. List of (Appropriate) Individual Coping Behaviors
f. Examples of Complimentary Tools
Appendix 2. Qualitative (PRA) Example of the CSI
Appendix 3. Examples of Studies
a. The Kenya Pilot Study
b. The Accra Urban Food and Nutrition Study
c. The CHS Surveys
d. Post-Tsunami Recovery—Sri Lanka
Appendix 4. Review of Literature: Other Food Security Indicators
iv
Coping Strategies Index: Field Methods Manual Second Edition
ACKNOWLEDGEMENTS
This manual describes the CSI tool and how to develop and use it. The original manual, written
in 2003, was based on a collaborative research project, implemented by WFP and CARE in
Kenya, with financial support of the UK Department for International Development via WFP,
The Bill and Melinda Gates Foundation, and CARE-USA.
Since 2003, the CSI has been widely used by the World Food Programme, CARE International
and other non-governmental organizations and humanitarian agencies, governments, and
researchers as an indicator of food insecurity in a variety of applications. During that time,
several flaws were noted in the original methods manual, which this edition addresses. It also
introduces new innovations since 2003.
This second edition was funded through support provided by USAID and CARE, and we
gratefully acknowledge their financial support.
We would also like to thank the contributing authors of the first edition who gave valuable
feedback to this edition, including Ben Watkins and Greg Collins, as well as Patrick Webb,
Jennifer Coates, and Chris Barrett for various insights incorporated here. We are grateful to Tom
Spangler for his help in preparing the literature review, Mark Langworthy for helping with the
analysis and validation of some of the new innovations in this edition, and Joyce Maxwell for
editing and desk-top publishing. We would also like to thank Joyce Luma and her colleagues in
the Vulnerability Assessment and Mapping (VAM) Unit of the World Food Programme for their
support in the application of the CSI—much of the innovation here results from work conducted
with VAM in the field. Lastly, we would like to thank Bob Bell and his colleagues at CARE-
USA for their support of field testing and this effort to publish a second edition of the Field
Methods Manual.
Any part of this manual may be reproduced for training or explanatory purposes, provided the
source is cited. The correct citation is:
Coping Strategies Index: Field Methods Manual. Copyright © 2008 Cooperative for Assistance
and Relief Everywhere, Inc. (CARE). Used by permission.
The Authors
January 2008
v
Coping Strategies Index: Field Methods Manual Second Edition
EXECUTIVE SUMMARY
Overview
Measuring food insecurity is a costly and complicated exercise. In highly food insecure countries
operational agencies need regular measurements for monitoring changes and for assessing the
impact of food aid interventions. Often these interventions take place in emergency conditions.
Time is limited, and field conditions do not permit lengthy and intensive data collection or
analysis processes. Tools are needed that are quick and easy to administer, straight-forward to
analyze, and rapid enough to provide real-time information to program managers.
The Coping Strategies Index (CSI) is one such tool. It was developed in Uganda, Ghana, and
Kenya but has now been used for early warning and food security monitoring and assessment in
at least nine other African countries and several in the Middle East and Asia.
The CSI measures behavior: the things that people do when they cannot access enough food.
There are a number of fairly regular behavioral responses to food insecurity—or coping
strategies—that people use to manage household food shortage. These coping strategies are easy
to observe. It is quicker, simpler, and cheaper to collect information on coping strategies than on
actual household food consumption levels. Hence, the CSI is an appropriate tool for emergency
situations when other methods are not practical or timely.
The CSI can be used as a measure of the impact of food aid programs, as an early warning
indicator of impending food crisis, and as a tool for assessing both food aid needs and whether
food aid has been targeted to the most food insecure households. During food aid needs
assessments the tool serves to identify areas and population groups where the needs are greatest.
It can also shed light on the causes of high malnutrition rates, which are often very difficult to
identify. Finally, if coping strategies are tracked over a long period, CSI is useful for monitoring
long-term trends in food insecurity.
1
Coping Strategies Index: Field Methods Manual Second Edition
2
Coping Strategies Index: Field Methods Manual Second Edition
(Corbett 1988, Devereux 1993). The CSI captures and systematizes this logic to reflect food
insecurity at the household level
As noted above, there are two basic types of coping strategies. One includes the immediate and
short-term alteration of consumption patterns. The other includes the longer-term alteration of
income earning or food production patterns, and one-off responses such as asset sales. While it is
important to understand longer-term livelihood strategies in an emergency, research has shown
that the management of short-term consumption strategies is an accurate indicator of acute food
security (Coates et al. 2006, Bickel et al. 2000, Maxwell et al. 1999). But people do not wait until
an outright shortfall of food is upon them before they begin to change their behaviors. Thus
changes in coping behavior not only reflect current status, they also reflect the best judgment of
household decision makers about the foreseeable future—giving the measurement of coping
behaviors some predictive ability too (Christaensen and Boisvert 2000).
Experience with the CSI has shown that, typically, food insecure households employ four types
of consumption coping strategies.
First, households may change their diet. For instance, households might switch food
consumption from preferred foods to cheaper, less preferred substitutes.
Second, the household can attempt to increase their food supplies using short-term strategies
that are not sustainable over a long period. Typical examples include borrowing or
purchasing on credit. More extreme examples are begging or consuming wild foods,
immature crops, or even seed stocks.
Third, if the available food is still inadequate to meet needs, households can try to reduce the
number of people that they have to feed by sending some of them elsewhere (for example,
sending the kids to the neighbors house when those
neighbors are eating). Box 1. What Is “Household Food
Fourth, and most common, households can attempt Security”?
Food security was defined years ago
to manage the shortfall by rationing the food by the World Bank as “access by all
available to the household (cutting portion size or people at all times to sufficient food for
the number of meals, favoring certain household an active, healthy life.” In practical
members over others, or skipping whole days terms, this encompasses the
without eating). physiological needs of individuals, the
complementarities and trade-offs
It will be clear that all these types of behavior indicate among food and other basic
a problem of household food insecurity, but not necessities (especially health care and
education, but others as well), changes
necessarily problems of the same severity. A household
over time in terms of people’s livelihood
where no one eats for an entire day is clearly more strategies, the assets to which they
food insecure than one where people have simply have access, and uncertainty and risk
switched from consuming rice to cassava. The basic (that is, vulnerability). Clearly, food
idea is to measure the frequency of these coping security is about more than just how
much people have to eat. Yet, having
behaviors (how often is the coping strategy used?) and
“enough” food to eat is the most
the severity of the strategies (what degree of food important outcome of being food
insecurity do they suggest?). Information on the secure; while physiological
frequency and severity is then combined in a single requirements differ, people largely
score, the Coping Strategies Index, which is an know whether they have “enough” or
not.
indicator of the household’s food security status. It
3
Coping Strategies Index: Field Methods Manual Second Edition
considers only the coping strategies that are important in a particular local context.
From the discussion of household food security in Box 1, it is clear that there are other factors
besides just short-term food consumption that must be considered in assessing food security.
These include longer-term livelihood strategies, labor opportunities, alternative income-
generating strategies, levels of physical and financial (and other) assets, and one-off asset sales
or bartering. Unfortunately, it is rarely possible to collect all this data in an emergency.
The Kenya Pilot Study (See Appendix 3) collected data to control for all these factors, and still
found that the CSI itself was both an accurate reflection of current food security status at the
household level, and a good predictor of future vulnerability. The CSI works because households
tend to use both consumption coping strategies and longer-term strategies to ensure that they
have enough to eat. Although a complete analysis of household food security would require a
detailed understanding of livelihoods, assets and consumption behaviors, the CSI is perfectly
adequate as a rapid indicator of household food security (Christaensen and Boisvert 2000,
Maxwell et al. 1999).
4
Coping Strategies Index: Field Methods Manual Second Edition
monitoring, the CSI has been used as an assessment tool, and to help guide household targeting
of food assistance. Lastly, the CSI has been used as an indicator of longer-term changes in food
security status. The main focus of this manual is on monitoring food insecurity, but the basic
design of the tool is the same for the various other applications mentioned above. It should be
underlined, however, that the tool does need some “up-front” work to ensure that it is accurately
adapted to the local operating environment. With a little care to these details, the CSI will rapidly
generate the information needed to improve the effectiveness and efficiency of food operations
in an emergency.
5
Coping Strategies Index: Field Methods Manual Second Edition
1
This particular example came from the Kenya pilot study. Please note, however, that it is only an example. It is
used to demonstrate the construction of the CSI throughout this section, for the sake of using a consistent example.
It does not imply that this is the “right” set of individual behaviors—those must always be generated from the
context in which the CSI is being used.
6
Coping Strategies Index: Field Methods Manual Second Edition
7
Coping Strategies Index: Field Methods Manual Second Edition
2
The first edition of the CSI Manual used a 30-day recall period and relied on relative frequencies of the answers.
For reasons of recall accuracy, these guidelines have shortened the recall period and ask for specific numbers of
days that individual behaviors were practiced.
8
Coping Strategies Index: Field Methods Manual Second Edition
3. If a respondent answers “all the time,” be sure to clarify if that means each day for the past
seven days, and if so, record the answer as seven. If s/he says “never,” be sure to clarify if
this means zero days, and if so record the answer as zero. If it is something that the
household never does, a “not applicable” answer should be given.
4. Scoring the results is explained below.
You now have the basic tool you need to conduct a household survey using the CSI. It should
contain the following elements:
A set of coping strategies or individual behaviors that represents the consensus of diverse
groups in the community, location, or culture, which people rely on when they don’t have
enough food and don’t have enough money to buy food.
A column to record the frequency with which people are forced to rely on these strategies or
behaviors.
A code for a “not applicable” answer. Note that a “not applicable” answer doesn’t change the
score for the CSI, but it does enable other kinds of analysis.
In order to conduct the analysis of the CSI, however, you need a few more pieces of information.
The first is a way to “weight” the severity of the individual behaviors.
3
Again, this represents a change from the first edition of this manual. In the first edition, several different methods
of weighting were suggested, and more complex instructions were offered about which method to select. In fact,
analysis has shown that results vary little with different methods of weighting, so this edition opts for the simplest
method—a simple ordinal ranking of one to four, weighted in the same manner.
9
Coping Strategies Index: Field Methods Manual Second Edition
In fact, it is possible to ask the same focus groups to first help brainstorm the list, and then to
discuss severity (though it is sometimes useful to have separate discussions so that the list of
coping strategies or behaviors is established and agreed first—a process that requires several
focus groups). Then the exercise below is carried out to establish the severity of each strategy or
behavior.
10
Coping Strategies Index: Field Methods Manual Second Edition
b. Borrow 2 2 2 2 2 3 2 2 2 2 2 2 2.1 2
f. HH members - - 2 2 2 1 3 2 2 3 2 3 2.2 2
eat elsewhere
g. Beg 5 5 4 4 3 3 2 4 4 5 5 5 4.1 4
11
Coping Strategies Index: Field Methods Manual Second Edition
12
Coping Strategies Index: Field Methods Manual Second Edition
Figure 4 is an actual example of a household from the Kenya Pilot Study. Below are details for
how it was scored.
Question (a) “In the past 7 days, if there have been times when you did not have enough food
or money to buy food, how often has your household had to rely on less preferred and less
expensive foods?”
The answer was that this household had done that five out of the previous seven days.
The severity weighting for this particular behavior is 1.
So the weighted total recorded for the answer to Question (a) is 5 (5 x 1).
For Question (b) the frequency was two days out of the previous seven and the severity
weighting was 2, so the weighted total is 4.
This procedure is repeated for each question: multiply the frequency score by the severity
weighting and record the number in the final box of the row. Then the individual scores in
the boxes are summed to the bottom of the form. Needless to say, for large surveys, it is
better to do the calculations with a computer.
13
Coping Strategies Index: Field Methods Manual Second Edition
4
At this point, there is no firm evidence about whether raw CSI scores can be directly compared across different
locations. For the safest results, caution should be exercised in making comparisons of raw scores that are not from
the same location or culture for which a specific application is prepared.
14
Coping Strategies Index: Field Methods Manual Second Edition
such as income, food budgets, and asset data. The strength of the relationship between the CSI
and such indicators has been tested in over fourteen studies. With measures of food frequency or
dietary diversity, meal frequency, and caloric intake the correlations are largely negative (as
would be expected, meaning that as the CSI increases the related variable decreases) and
statistically significant. Similarly, the correlation is negative and significant with measures of
assets, and correlates positively (as expected) with percent of expenditure devoted to food. This
confirms earlier studies indicating that the CSI is significantly correlated with other measures of
consumption adequacy, and equally well correlated with measures of assets and expenditure
(Maxwell et al. 1999).
Convergence (Confirming)
However, the correlation with many of these indicators, while statistically significant, is not
always robust. So where possible, the use of more than one indicator of food insecurity is
recommended (i.e., the CSI with dietary diversity or WFP’s Food Consumption Score indicator).
This not only permits “triangulation” of findings, it deliberately uses indicators that capture
different elements of the complex notion of food insecurity. This enables convergence of
findings, providing greater confirmation of food security status.
15
Coping Strategies Index: Field Methods Manual Second Edition
5
It should be noted that the only rigorous test of this observation has been conducted with data from Sub-Saharan
Africa only, although the instrument has been used elsewhere, and the same behaviors show up in those surveys as
well.
16
Coping Strategies Index: Field Methods Manual Second Edition
17
Coping Strategies Index: Field Methods Manual Second Edition
18
Coping Strategies Index: Field Methods Manual Second Edition
If the CSI is incorporated into monitoring even before an intervention is begun, as part of an
early warning or food security information system, CSI can also be compared with other (non-
household) indicators of the general food security situation. It should be stressed, however, that
while most of the indicators tracked in an Early Warning System are “leading indicators”
(indicators that note a potential problem before it arises) the CSI is both a “current” and a
“leading” indicator (that is, it gives information about both current and future status).
Christaensen and Boisvert (2000) reported that the CSI was a better indicator of future food
consumption that either current income or current caloric intake. Having CSI information already
on hand greatly enhances both analytical capability of early warning, and the timeliness of
response.
Household Targeting
If households are targeted by an administrative targeting mechanism, the CSI can be used to
assess the validity of the targeting criteria. Those households fulfilling the targeting criteria
should have a higher CSI score than those that do not (note that this must be pre-intervention CSI
information). If a quick cross check shows that this is not the case, it is strong evidence that the
targeting criteria are wrong, and that the intervention is not really being targeted to the most
food-insecure households. Likewise, in a qualitative application to be used in conjunction with a
community-based targeting system, the CSI can be utilized to cross check against wealth ranking
or other mechanism used by the community or the relief committee to determine the eligibility of
individual households. For this kind of application—whether qualitative (in the case of
community-based targeting) or quantitative (in the case of administrative targeting)—the
context-specific CSI should be used, since the comparison among households will be within the
location in which the CSI was developed.
Geographic Targeting
The comparative (reduced) CSI was developed from the original (context-specific) CSI precisely
to be able to compare across different contexts or even different emergencies, to enable policy
makers to have some idea of the relative severity of different crises or compare the severity of
19
Coping Strategies Index: Field Methods Manual Second Edition
the same crisis in different locations. Among other things, such a comparison is useful for the
geographic targeting of emergency assistance.
Timing
Clearly, an early warning/monitoring system that uses the CSI can give a clear signal that
household food insecurity is increasing, and provide this information in adequate time to enable
a rapid response. Evidence shows that people begin to ramp up their coping behaviors well in
advance of actually facing an outright shortfall in food for consumption. CSI information can
also be used to inform program managers when it is time to phase out the emergency
intervention, or transition to a different kind of intervention. If used in a regular monitoring
system, the CSI can track household food security status, to help guide the timing of
programmatic transitions. Since information would be known about household food security
levels, it would help managers decide whether a food-for-work intervention or other recovery
intervention would be more appropriate. At the moment, little of this information is routinely
available to program managers.
Long-Term Interventions
Though intended here as an indicator of relatively short-term food security status, the CSI tool
could be used to track the impact on household food security of longer-term interventions (i.e.,
development projects and programs) in addition to short-term (emergency) interventions. Be
aware that the CSI is sensitive to short-term changes such as seasonality, or the effects of shocks,
however major or minor. If being used to track long-term interventions, ensure that short-term
influences such as seasonality are factored out of the analysis (for example, by conducting a
baseline survey and an impact evaluation survey at the same time of the year/harvest cycle, etc.).
20
Coping Strategies Index: Field Methods Manual Second Edition
However, nutrition surveys themselves provide little information about causal factors unless
complemented with other information—and the information required deals mainly with food
security and health status. Using the CSI in conjunction with a nutrition survey provides
information about food security status—and can be complemented with questions about health
information. This gives the analyst some sense of the main causes of a nutritional problem,
which a nutrition assessment alone usually does not do.
As noted, the CSI can be used as a part of a Post Distribution Monitoring system—specifically
with an end-use form for food assistance—to provide information about the impact of food aid.
21
Coping Strategies Index: Field Methods Manual Second Edition
6. REFERENCES
Bickel, Gary, Mark Nord, Cristofer Price, William Hamilton and John Cook (2000). Guide to
measuring Household Food Security in the USA. Washington: USDA.
Christiaensen, Luc, and Richard Boisvert (2000). “On Measuring Household Food Vulnerability:
Case Evidence from Northern Mali.” Working Paper. Department of Agricultural, Resource,
and Managerial Economics, Ithaca, New York: Cornell University.
Coates, Jennifer, Edward Frongillo, Beatrice Rogers, Patrick Webb, Parke Wilde, and Robert
Houser (2006). Commonalities in the Experience of Household Food Insecurity across
Cultures: What Are Measures Missing? Supplement to the Journal of Nutrition (136)5:
1438–1448.
Corbett, Jane (1988). “Famine and Household Coping Strategies.” World Development 16(9):
1099–1112.
Darcy, James, and Charles-Antoine Hoffman (2003). According to Need? Needs Assessment and
Decision-Making in the Humanitarian Sector. Humanitarian Policy Group Report 15.
London: Overseas Development Institute.
Davies, Susannah (1996). Adaptable Livelihoods. Coping with Food Insecurity in the Malian
Sahel. London: MacMillan Press.
Devereux, Stephen (1993). “Goats Before Ploughs: Dilemmas of Household Response
Sequencing During Food Shortages.” IDS Bulletin 24(2): 52–59.
Food and Agriculture Organization (2006). The Integrated Phase and Humanitarian Phase
Classification Technical Manual. Nairobi: Food Security Analysis Unit for Somalia.
Kennedy, Eileen (2002). “Qualitative Measures of Food Insecurity and Hunger.” Keynote paper
on “Methods for the Measurement of Food Deprivation and Undernutrition.” Proceedings of
the International Scientific Symposium on Measurement and Assessment of Food
Deprivation and Undernutrition. FAO, Rome.
Maxwell, Daniel, Clement Ahiadeke, Carol Levin, Margaret Armar-Klemesu, Sawudatu
Zakariah, and Grace Mary Lamptey (1999). “Alternative Food Security Indicators:
Revisiting the Frequency and Severity of ‘Coping Strategies.’” Food Policy 24(4): 411–429.
Maxwell, Daniel (1996). “Measuring Food Insecurity: The Frequency and Severity of ‘Coping
Strategies.’” Food Policy 21(3): 291–303.
Maxwell, Simon, and Timothy Frankenberger (1992). Household Food Security: Concepts,
Indicators, Measurements. A Technical Review. New York and Rome: UNICEF and IFAD.
Watts, Michael (1983). Silent Violence: Food, Famine and Peasantry in Northern Nigeria.
Berkeley: University of California Press.
Young, Helen, and Susanne Jaspars (2006). The Meaning and Measurement of Acute
Malnutrition: A Primer for Decision-Makers. Humanitarian Practice Network Paper No. 56.
London: Humanitarian Practice Network. London: Overseas Development Institute.
22
Coping Strategies Index: Field Methods Manual Second Edition
23
Coping Strategies Index: Field Methods Manual Second Edition
will be too imprecise to be useful and are unlikely to be adequate to support statistical inference.6
More details, including various sample sizes and strategies for baseline, mid-term and final
surveys, as well as information about how to select households in the absence of a proper
sampling frame, are presented in Appendix 4.
Cluster Sampling
Where an exhaustive sampling frame does not exist for households, the next lowest aggregation
of these units for which an exhaustive sampling frame exists must be used to select the sample.
6
Statistical inference is the ability to infer something about the larger population (N) from which the sample (n) was
taken on the basis of probability theory.
7
The remainder of this section on sampling methods comes from the first edition, and was written by Greg Collins.
24
Coping Strategies Index: Field Methods Manual Second Edition
These aggregated units are often villages, but other appropriate aggregations may exist,
especially for urban and/or nomadic populations.
The cluster sampling approach entails selecting clusters at the first stage of sampling and then
selecting households from within these clusters during the second stage of sampling. To maintain
the criteria that all households have an approximately equal probability of selection, clusters
must be weighted according to size (e.g., large clusters have a higher probability of selection
than small villages such that all households, regardless of village size have an approximately
equal probability of selection).
For this example use a random numbers table select numbers between 1 and 891 to choose
clusters. If we require 3 clusters, choose 3 numbers randomly between 1 and 891. Let us say we
have selected 439, 831, and 558. This would mean that we would take two clusters from village
2 (e.g., village/cluster 2 was selected twice) and one cluster from village 3.
25
Coping Strategies Index: Field Methods Manual Second Edition
Divide the total number of units in the sampling frame (column three above, or 6 in this
example) by the number of clusters needed (let us say we require two clusters). This yields a
sampling interval of 2. Select a random start in the sampling frame (for this example let us say
we selected the second unit of Village A). We then add the sampling interval 2 in order to select
the next cluster and we get the second unit of Village C. Therefore, the first cluster will be
Village A and the second cluster will be Village C. Note that once a cell in the sampling frame
has been selected, it cannot be selected again. This is sampling without replacement. Clusters
with multiple cells (e.g., Village A and Village C in the example) in the sampling frame can be
selected more than once (e.g., village A can be selected up to two times and village C can be
selected up to three times). In practice there will be many more clusters to choose from and more
clusters needed to make up the sample (see Sample Size and Clusters below), but the concept is
the same.
26
Coping Strategies Index: Field Methods Manual Second Edition
Enumerator 1
Enumerator 2
27
Coping Strategies Index: Field Methods Manual Second Edition
change between survey rounds is increased. In other words, the larger the sample size, the
smaller the minimum magnitude of detectable change. This is important given that the purpose
of impact monitoring is to gauge the change that occurs over time in response to food
distribution operations. However, the increased minimum magnitude of detectable change
associated with the compromised and minimalist approaches will still provide a useful means of
tracking meaningful change over time in the CSI score.
Figure 4.Three Options for Sample Size and Clusters for Use in Cluster
Sampling*
Number of Minimum magnitude
Number of HH within Total sample of detectable
Approach clusters each cluster size (n) change in CSI
Ideal 30 20 600 6.5
Compromised 25 20 500 7.0
Minimalist 20 20 400 8.0
* These were calculated using data from the Ghana research as a basis for parameter estimates in the sample size calculation.
Note that these calculations apply to the full (context-specific) CSI, not the reduced CSI. The latter will not result in
comparable changes of magnitude in CSI scores.
Sample Size for Simple Random Sampling (SRS) where an Exhaustive Sampling
Frame of Households Exists
The sample size for an SRS is half that of the cluster sampling approach (e.g., the cluster
sampling entails a design effect of two, doubling the sample size). Therefore, a sample size of
300 households should be randomly selected from the household-level sampling frame.
28
Coping Strategies Index: Field Methods Manual Second Edition
intervals (the minimalist approach will have wider confidence intervals than the compromised
approach and the compromised approach will have wider confidence intervals than the ideal
approach given the same mean and variance within the sample).
Stratification
Stratification is used when separate CSI estimates are desired for sub-groups within the study
population. For example if your study population includes two districts, it may be desirable for
each district to have a separate CSI estimate. We would then consider each district a “strata” and
apply the required sample size and number of clusters to each stratum (in this case, districts).
Be certain that separate estimates are required before stratifying the sample. In the above
example of two districts treated as separate strata, the sample size, number of clusters, and much
of the cost of the survey are doubled. For three strata, the sample size, number of clusters, and
much of the cost of the survey are tripled and so on. Therefore, be very critical before stratifying
your sample, weighing the advantages and disadvantages, particularly cost and resources, against
one another.
b. Respondents
Once you have selected the appropriate households, you also need to find the right respondent
within the household. The best person to ask about coping is the person in the household that is
in charge of preparing food and seeing to it that members eat. Usually, but not necessarily
always, that person is the senior female member of the household—typically the wife, the
mother, or female head of household. However, there are cases of households that do not have
such a member, or there may be another person in the household who is responsible.
In extreme emergencies, households may be broken up, and the notion of a “household” may
require modification. In general, households are usually defined in terms of the group of people
who “eat together or eat from the same pot.”
29
Coping Strategies Index: Field Methods Manual Second Edition
To track the impact of food aid, obviously, you will need information about not only food
security status (from CSI) but also about the receipt and end-use of food aid. Additionally,
contextual information (early warning information) is also useful. See the section below on
examples of complimentary tools for details for using CSI in conjunction with food aid.
Other information needed will depend on the objectives of the use of the CSI tool. The primary
purpose being described in this manual is monitoring the impact of food aid in emergencies, but
as noted, CSI has many other applications.
30
Coping Strategies Index: Field Methods Manual Second Edition
changes in general food security status (a harvest, changes in prices, changes in labor
opportunities, etc.).
31
Coping Strategies Index: Field Methods Manual Second Edition
K14 Sold jewelry or household items to purchase food? Yes _______ No ________
Sold livestock or farm implements to purchase
K15
food? Yes _______ No ________
32
Coping Strategies Index: Field Methods Manual Second Edition
33
Coping Strategies Index: Field Methods Manual Second Edition
In the past six months, how many times has your household received food aid?
Yes No
Did you have to pay anything to receive food aid?
Distribution Chief Friends Other: _________
Of total food aid consumed, how much of the food (kgs.)
aid did you get from:
Day Month Year
When was the last distribution you received?
34
Coping Strategies Index: Field Methods Manual Second Edition
i. Restrict consumption by adults in order for small children to eat? OOOO OOOOOO
This can be made a little more sophisticated by asking questions about relative frequency, rather
than just the “yes/no” response depicted above.
35
Coping Strategies Index: Field Methods Manual Second Edition
The important point is to allocate the same number of stones or beans for each question, so that
answers can be compared. This exercise can be a useful way of ensuring that the set of coping
strategies is complete and accurate for the given location. It can also be used in conjunction with
other rapid appraisal or PRA methods to give a quick overview of the situation at the community
level. It is less appropriate for tracking the impact of an intervention because it does not give any
disaggregated information about vulnerable households.
36
Coping Strategies Index: Field Methods Manual Second Edition
Results
The CSI correlated significantly with
food frequency, asset ownership,
income, and other measures of
livelihood security. The CSI correlates
significantly with changes in early
warning indicators. Bi-variate and
multi-variate analysis was carried out to
demonstrate this relationship. Figure 2
shows the results of the bi-variate
correlation between the CSI and a food
frequency measure, intended to be an
alternate measure of food security. The
table is broken down by survey round
and district.
37
Coping Strategies Index: Field Methods Manual Second Edition
These results strongly confirm that the CSI and food frequency indicators were picking up the
same trends in terms of household food security, which tends to confirm part of the first
hypothesis (that correlated with other food security indicators).
Multi–variate analysis was conducted using the CSI as the dependent variable. Independent
variables include location, asset ownership, income sources, non-consumption coping strategies
(asset sales, alternative income sources, expenditure reduction, migration, etc.), as well as the
receipt and timing of food aid.8
Asset ownership was negatively associated with CSI, which stands to reason—more assets
would imply both a higher level of wealth generally, as well as a greater capacity to cope with a
shock without it necessarily affecting food security. CSI was positively but weakly correlated
with different kinds of income, including agriculture, livestock, and labor, but negatively
correlated with the number of income sources. This indicates that all kinds of incomes were
affected by the drought (note that both an agricultural and pastoral area were included in the
sample—analyzing the results separately for each district would likely have sorted out the
differences between livelihoods systems). But it also means that the greater the level of
livelihood diversity, the greater the household’s capacity to withstand shocks.
Many of the other non-consumption strategies were correlated (at varying levels of significance)
with the CSI. This is an extremely important finding, because it implies that various kinds of
coping tend to co-vary. This suggests that the limited set of consumption coping strategies that
can be easily measured with the CSI are an accurate reflection of other kinds of coping going on
at the household level, and the substantial additional information collected in the pilot study on
non-consumption strategies need not be collected to have an accurate picture of the level of
coping at the household level. In other words, this finding implies that the CSI is an adequate
stand-alone indicator.
Receipt of food aid was positively associated with CSI in all the models in which the relationship
was tested—at first glance a counter-intuitive finding. However, given the negative relationship
between assets and the CSI, this would imply that the food aid that was received was accurately
targeted on households that needed it, but that the amounts received were not adequate for the
needs of those receiving it. The variable being analyzed was the binomial (receipt of food aid or
not)—it is not a quantitative measure of per capita food aid receipts. In Garissa District, 88
percent of all households had received food aid during Round 1, 68 percent of households in
8
This analysis was conducted by Greg Collins.
38
Coping Strategies Index: Field Methods Manual Second Edition
Round 2, and 66 percent of households in Round 3. Yet the mean amount of food aid per capita
received at the household level was between 5.1 kilograms per person and 7.5 kilograms per
person (a full food basket is considered 15 kilograms per person per month).
CSI (rationing
0.910** 1.000
strategies only)
Kcal per adult per
-0.082* -0.138** 1.000
day
Food share of
0.195** 0.144** 0.164** 1.000
household budget
Income (per capita
-0.220** -0.215** 0.374** -0.497** 1.000
expenditure)
Height for age z-
-0.108** -0.104** 0.033 -0.118** 0.146** 1.000
score of child
Data Source: Maxwell et al. 1999
*
Correlation Significant (p < 0.05)
**
Correlation Significant (p < 0.01)
39
Coping Strategies Index: Field Methods Manual Second Edition
CSI
45.00
strategies.9 Each question used five relative
frequency categories ranging between 40.00
“every day per week” to “never.” The CHS 35.00
collected information on the relevance of
30.00
coping strategies among sample Rd1 Rd2 Rd3
communities and determined the relative
severity of each coping strategy. It then
assigned a value of between one and four Figure 5. CSI Beneficiary
to each strategy—or a severity score—with CSI for Beneficiary and Non-
one being the least severe and four being beneficiaries
the most severe. To analyze the data, the
frequency score recorded during the 60.00
household surveys was multiplied by the 55.00
50.00
severity score,10 producing a single score
CSI
45.00
for each strategy. This same procedure was
40.00
repeated during each round of the CHS. 35.00
Severity scores were recalibrated using 30.00
results from eighteen focus groups. Rd1 Rd2 Rd3
9
The coping strategies were: Limit portion size at mealtimes; Reduce number of meals eaten per day; Skip entire
days without eating; Borrow food or rely on help from friends or relatives; Rely on less expensive or less
preferred foods; Purchase/borrow food on credit; Gather unusual types or amounts of wild food / hunt; Harvest
immature crops (e.g., green maize); Send household members to eat elsewhere; Send household members to beg;
Reduce adult consumption so children can eat; and Rely on casual labor for food.
10
The frequency value is on a scale of 1-5 and is a measure of the number of times per week that a household
employs a given strategy while the severity value, also on a scale of 1–5, is a perception of how serious
households view a given behavior (with 5 being the most severe).
40
Coping Strategies Index: Field Methods Manual Second Edition
CSI
the lean season (Round 2) freed households 40.00
from relying as much on coping behaviors. 30.00
20.00
e
i
a
nd
o
aw
bw
qu
bi
th
la
m
so
al
bi
ba
i
Za
az
M
am
Le
Sw
The average CSI varied significantly among
Zi
oz
M
the six countries surveyed (Figure 6). Lesotho
and Zambia had significantly higher CSI means than the other four countries—Mozambique,
Malawi and Zimbabwe all had the same mean CSI, while Swaziland had the lowest. However,
the CSI in all countries fell significantly over the three rounds of the survey, with the exception
of Swaziland, where the CSI actually increased from a mean of 20.3 in Round 1 to 36.7 in Round
3. Despite Swaziland’s relatively small sample size, this increase in the CSI was statistically
significant and triggered further investigation as to its causal factors.
41
Coping Strategies Index: Field Methods Manual Second Edition
Another way used to view the same data was to calculate the average score of the frequency. The
scale used was 1 to 5, with 1 being “never” and 5 being “always.” As Figure 8 shows, no
frequency averaged more than 3.0, and most were below 2.0. This means that on average,
households were using these strategies less than one day per week or, at most, 1–2 days per
week. The most common strategies in the baseline were to rely on casual labor, to purchase food
on credit, and to rely on less expensive or less-preferred foods. These three strategies remained
important in the EOP study, but as the data shows, the frequency of using any of the nine coping
strategies declined following the baseline study (with the exception of gathering wild foods,
which has remained about the same, and may be associated with seasonal factors).
42
Coping Strategies Index: Field Methods Manual Second Edition
Each household has a coping strategy index which represents the weighted value of all nine
behaviors. Figure 9 provides average CSI values for each study period by zone and district.11 The
data showed that food security remained an issue in the East, even as other indicators of recovery
showed progress. Note that CSI values decreased significantly in the South—likely because
income had recovered and households had the means to return to their normal diet behaviors.
The CSI values in the East suggested that households had not been able to decrease the number
or frequency of coping strategies associated with diet. This was consistent with economic
recovery data showing households in the East with declining income and increased debt
associated with non-productive expenditures.
30
25
20 Baseline
CSI
15 EOP
10
0
a
ee
le
a
h
ra
ra
l
a
st
Al
ar
ot
ut
lo
al
Ea
ta
pa
al
So
co
at
nt
G
lu
m
Am
ba
M
Ka
tti
co
Ba
m
in
Ha
Tr
11
The East was represented by Ampara, Batticoloa, and Trincomalee districts, while the South was represented by
Kalutara, Galle, Matara, and Hambantota.
43
Coping Strategies Index: Field Methods Manual Second Edition
44
Coping Strategies Index: Field Methods Manual Second Edition
45
Coping Strategies Index: Field Methods Manual Second Edition
References
Barrett, C. B. (2002). Food Security and Food Assistance Programs. Published in Handbook of
Agricultural Economics, B. L. Gardner and G. C. Rausser (eds.).
Coates, Jennifer, Edward A. Frongillo, Beatrice Lorge Rogers, Patrick Webb, Parke E. Wilde,
and Robert Houser. (2006). Commonalities in the Experience of Household Food Insecurity
across Cultures: What Are Measures Missing? American Society of Nutrition (originally
published in a supplement to The Journal of Nutrition).
Development Information Services International (DISI). (2006). World Food Programme
Comprehensive Food Security and Vulnerability Analysis (CFSVA): An External Review.
Dewey, K. G., R. J. Cohen, M. Arimond, and M. T. Ruel. (2005). Developing and Validating
Simple Indicators of Complementary Food Intake and Nutrient Density for Breastfed
Children in Developing Countries. Report. Washington, D.C.: International Food Policy
Research Institute (IFPRI); Davis, CA: University of California at Davis.
46
Coping Strategies Index: Field Methods Manual Second Edition
Frongillo, Edward A., and Siméon Nanama. (2006). Development and Validation of an
Experience-Based Measure of Household Food Insecurity within and across Seasons in
Northern Burkina Faso. American Society of Nutrition (originally published in a supplement
to The Journal of Nutrition).
Hoddinott, J. and Y. Yohannes. (2002). Dietary Diversity as a Food Security Indicator.
(Discussion paper 136) Washington, D.C.: International Food Policy Research Institute (IFPRI).
Maxwell, Daniel, Ben Watkins, Robin Wheeler, and Greg Collins. (2003). The Coping Strategies
Index: A Tool for Rapid Measurement of Household Food Security and the Impact of Food
Aid Programs in Humanitarian Emergencies. Field Methods Manual. Developed for CARE
Eastern and Central Africa Regional Management Unit (CARE-EARMU) and World Food
Programme Vulnerability Assessment and Mapping (VAM) Unit.
Maxwell, Daniel, Clement Ahiadeke, Carol Levin, Margaret Armar-Klemesu, Sawudatu
Zakariah, and Grace Mary Lamptey. (1999). Alternative Food-Security Indicators:
Revisiting the Frequency and Severity of “Coping Strategies.” Food Policy 24 (1999): 411–
429.
Swindale, Anne, and Paula Bilinsky. (2006). Development of a Universally Applicable
Household Food Insecurity Measurement Tool: Process, Current Status, and Outstanding
Issues. American Society of Nutrition (originally published in a supplement to The Journal
of Nutrition).
TANGO International. (2004). Guide for Measuring Food Access. Developed for Food and
Nutrition Technical Assistance (FANTA).
Webb, Patrick, Jennifer Coates, Edward A. Frongillo, Beatrice Lorge Rogers, Anne Swindale,
and Paula Bilinsky. (2006). Measuring Household Food Insecurity: Why It’s So Important
and Yet So Difficult to Do. American Society of Nutrition (originally published in a
supplement to The Journal of Nutrition).
Weismann, Doris, John Hoddinott, Noora-Lisa Aberman, Marie Ruel. (2006). Review and
Validation of Dietary Diversity, Food Frequency and Other Proxy Indicators of Household
Food Security. International Food Policy Research Institute (IFPRI). Draft. May 17, 2006.
47