M&E Training Manual
M&E Training Manual
M&E Training Manual
on
Monitoring and Evaluation
Concepts, Tools and Strategies
for
Social Sector Programmes
Tools Series II
Ministry of Social Development
Monitoring & Evaluation Division
September 2003
TABLE OF CONTENTS
PREFACE ...................................................................................................................................... 1
INTRODUCTION ........................................................................................................................... 2
Needs Assessment.................................................................................................................. 10
Baseline Assessment.............................................................................................................. 11
Conceptual Model as an Integral Part of Programme Design............................................. 11
Establishing a Programme Theory ....................................................................................... 12
Developing a Logical Framework ........................................................................................ 12
Defining Goals and Objectives ............................................................................................. 13
Impact Objectives.................................................................................................................. 14
Choosing Activities ............................................................................................................... 15
Identifying Indicators ............................................................................................................ 15
Process Indicators................................................................................................................. 15
Results Indicators.................................................................................................................. 16
Criteria for Selecting Indicators ........................................................................................... 16
MODULE 3: PLANNING A USEFUL EVALUATION ................................................................ 19
Experimental Design............................................................................................................. 31
Non Experimental Design .................................................................................................. 31
Pre Test and Post Test Design ........................................................................................ 32
Static Group Comparison ..................................................................................................... 32
Quasi Experimental Design ............................................................................................... 32
Ministry of Social Development/Monitoring and Evaluation Division September 2003
Quantitative Data.................................................................................................................. 38
Qualitative Data.................................................................................................................... 38
Tips for Quantitative and Qualitative Data.......................................................................... 39
Coding................................................................................................................................... 40
Data Analysis ........................................................................................................................ 40
Analysing Qualitative Data................................................................................................... 40
Analysing Quantitative Data................................................................................................. 40
Preparing an Evaluation Report ........................................................................................... 42
Developing a Solutions Framework...................................................................................... 42
ii
Preface
This training manual was developed as a reference document for an introductory training course
in Monitoring and Evaluation Concepts, Tools and Strategies for Social Sector Programmes. It
is intended that the course material covered will assist in building monitoring and evaluation
capacity in the social sector, through the exposure of participants to five modules of training and
the transfer of knowledge by these persons to others in their respective Ministries.
This publication is the second in a series of Tools being developed by the Monitoring and
Evaluation Division of the Ministry of Social Development, with the support of the Evaluation
Advisor seconded to the Ministry through The Commonwealth Fund for Technical Cooperation
(C.F.T.C.).
Permanent Secretary
Ministry of Social Development
INTRODUCTION
Most of you must have heard of, or participated in, an evaluation exercise. You may even have
initiated an evaluation by carrying it out using internal resources or by contracting a consultant to
do it. As managers of social sector programmes, you will, in the course of your career, be called
upon to make decisions about the viability of a particular programme, or to compare the relative
benefits of various programmes. When this happens, you would be required to support whatever
decision you make with concrete facts and you will realise that evaluation can be a useful tool to
achieve your objective. If you are not too familiar with evaluation, you may begin to wonder
why there is a need for it.
In their seminal book on evaluation, Peter Rossi et al, argued limited resources for social
programmes in every country, including the United States, make it critical that such investments
yield demonstrable and proportionate social benefits1
Evaluation is also undertaken to strengthen governance and institutional development. Findings
from evaluations help to ascertain the extent to which ongoing programmes are (i) achieving the
objectives for which they were designed; (ii) responding to the actual needs of the beneficiaries;
(iii) conforming to existing policies and procedures; and (iv) contributing to institutional learning
and capacity development.
Evaluation has become a key component of results-based and performance-based management.
Findings from evaluations are used to initiate policy and programme changes. Also, they can
contribute to organisational learning through constant exchange of information among key
stakeholders.
Other reasons for undertaking an evaluation are:
1
2
Peter Rossi et al, Evaluation. A Systematic Approach, 6th Edition, 1999, pp6-7
Ibid
3
4
Ibid
Ibid
With increasing emphasis on results-based management, evaluation findings are being used as a
management tool to determine if a programme should be continued, discontinued, improved,
expanded or curtailed.
It should be noted that in addition to using evaluation findings to improve programmes,
evaluations could be conducted to contribute to social science knowledge. This may necessitate
the use of rigorous social science research methods such as experimental and quasi-experimental
design.
Furthermore, evaluation findings can be used for public relations purposes. In this context,
evaluation can be launched to justify a political decision or simply to promote the image of the
organisation that initiates it.
Need for Reliable Data
Whatever the purpose, an evaluation should produce reliable data that can stand both internal and
external scrutiny. Evaluation findings should guide programme managers and decision- makers
to make decisions to improve ongoing/future interventions. A fundamental principle, when
considering undertaking an evaluation, is to aim for quality in the design of the study so that the
findings can stand the test of time. Also, evaluation must be credible, impartial and costeffective.
Credibility
It is advised that the process for data collection be transparent. As indicated before, key
stakeholders must be consulted and their views must be taken into account when determining the
questions to focus on. Most importantly, the core competencies or the skills set of the evalua tors
must be sound.
Impartiality and Independence
Evaluators should remain impartial throughout the evaluation process. Above all, findings and
conclusions should be free of personal biases. Judgements must be based on proven and reliable
data that can withstand scrutiny.
Cost-Effectiveness
The benefits of evaluation should outweigh the costs. The evaluation should be well tailored and
focused to minimise costs. Under most circumstances, the total cost of an evaluation should not
exceed 10% of the annual expenditure on the programme being assessed.
Question (s)
Programme relevance
Programme efficiency
Programme accountability
Programme impact
Catalytic effect
Programme sustainability
It should be noted that evaluation could cover any or all of the issues raised. When all of the
issues are covered it is known as a comprehensive evaluation or an in-depth evaluation. The
type of evaluation to be undertaken is invariably determined by the uses of the evaluation
findings. Above all, whoever is requesting the evaluation could also influence the types of
questions that could be asked. For instance, if the request comes from sponsors of the
intervention such as the government, they may want to know whether the established policies
and procedures are being followed as well as the ultimate effects of the intervention on the
beneficiary group. If it comes from those who are managing the intervention, they might be
interested in finding out about the internal workings of the intervention in terms of the timeliness
and cost effectiveness of the activities being implemented. These issues are discussed further in
Module 3.
Remember this caveat: Keep the evaluation as focused and simple as possible. Too many
questions could result in a superficial assessment and/or inefficient use of valuable time as well
as human and financial resources.
Evaluation versus Social Science Research
Evaluation makes use of social science data collection methods and as mentioned earlier, every
effort must be made to ensure that the data, which is collected during evaluation, is reliable and
can stand any scrutiny. However, evaluation is more art than science because it is often tailored
to meet the needs of decision- makers and programme implementers who may be impatient for
the results. The nature of the request may not give the evaluator sufficient time to apply the most
rigorous of methods.
Social science research is undertaken to produce knowledge about a given problem. The
methods used tend to be more rigorous (for example: experimental design) than the tools used to
design social interventions. The reason being that it may be unethical to withhold services from
a segment of a population just for experimental purposes! It may even be politically
unacceptable to do so.
One cannot possibly undertake a thorough evaluation if appropriate mechanisms have not been
put in place to facilitate the monitoring of the intervention on a regular and systematic basis.
Without a good monitoring system it would be difficult to compare pre- intervention and postintervention conditions to ascertain if an intervention has indeed made a difference in the lives of
the beneficiaries.
What is Monitoring?
Monitoring is used to determine how well a programme is carried out at different le vels and at
what cost. It tracks changes that occur over time in terms of resource inputs, production, and
use of services. Above all, a monitoring system provides information on progress towards
the achievement of stated objectives.
Steps in a Monitoring Process
For a monitoring system to work effectively, it requires the development of a management
information system (MIS) for data capture, storage, retrieval and analysis. This could be based
on manual and/or electronic templates. It may be advisable to develop electronic templates for
more complex monitoring requirements. For instance, to date, there are over 150 social
interventions across Trinidad and Tobago aimed at improving the lives of the most marginalized
segments of society. These interventions are carried out through several Government Agencies
and reach thousands of beneficiaries. The data for such interventions should be available in
electronic formats and stakeholders should be able to access the data at any point in time.
Role of Performance Indicators in Monitoring
Monitoring is relatively straight forward if right from the on-set of an intervention, thought is
given to developing indicators for the defined objectives. The data collected should be based on
the agreed indicators. The information derived could then be used to improve the activities (see
Module 2 for additional discussion on objectives and indicators). In conclusion, it should be
underlined that routine collection of intervention data is necessary. It helps to improve on
programme management and performance and it enables us to know how well a programme is
doing.
If lapses are detected midstream, measures could be taken to streamline them.
Moreover, it facilitates accountability in terms of determining if established policies and
procedures are being adhered to.
References:
1.
2.
3.
4.
Peter H. Rossi et al, Evaluation: A Systematic Approach, 6th Edition, Sage Publication,
1999.
International Programme for Development Evaluation Training (IPDET) July 2003,
Module 1 Introduction to Development Evaluation and Module 2 Evaluation
Models.
World Bank, Operations and Evaluation Department, Monitoring and Evaluation: Some
Tools, Methods and Approaches, 2002.
Carol H. Weiss, Evaluation Research: Methods of Assessing Programme Effectiveness,
1972, Princeton Hall.
10
Baseline data
Intervention
Sensitization
& Training
Programme
Average rating of
the same Homes
June 2003 = 6.4
Identification
of a problem
Design of an
intervention
Results
There are some key questions worth asking when designing a conceptual model. These are:
What are the specific problems affecting the focus population?
Which of these problems can the intervention realistically deal with?
11
What changes must occur in the beneficiary population to reduce the problem?
What activities should be carried out to bring about the change? etc.
Conceptual Model for the Civilian Conservation Corps
Exposure to Induction
training: eg. Physical
Training, Lectures in
life skills development
Participation
of Youth at
Risk in
6 month
CCC Program
Increased
Employability
Increased confidence,
development of
positive work ethic,
development of
relevant skills
Exposure to On
the Job
Training in one
of 10 skill areas
Transformation of
Youth at Risk into
productive, well
balanced and
responsible citizens
Reforestation and
conservation of the
environment
12
Project Goal:
Objective 1:
Key Activities
Indicators
Process:
No of mentor
applications approved.
No of groups
providing beneficiary
listings.
No of persons
participating in mentor
training.
Results:
Percentage of mentees
satisfied with mentor
support system.
Means of
Verification
Frequency Person
Responsible
Admin. Records
Annual
J.Smith/M.Doe
Admin. Records
Annual
J.Smith
Admin. Records
Annual
J.Smith /
Consultant
Post-intervention Annual
survey of mentees
J.Smith
13
14
Ultimately impact objectives are key since in practice most interventions aim to effect change in
the circumstances of the beneficiary population. We should endeavour at all times to write
smart objectives. SMART stands for:
Specific avoid differing interpretation
Measurable monitor and evaluate progress preferably in quantitative terms
Appropriate to the problem being addressed
Realistic- achievable yet challenging
Time-bound within a specific timeframe
Choosing Activities
After identifying the objectives, the next step is to choose activities. Activities refer to specific
tasks. They are what the organisation proposes to do to achieve the defined objectives. It is
advisable to review available skills and expertise before identifying the activities. Examples of
activities:
Organise counselling sessions for teenage girls who could be at risk of getting pregnant.
Conduct 10 hours of training for recipients of hampers under the SHARE programme.
Identifying Indicators
An indicator is a measure of a concept or behaviour. An indicator is used as a road map to assess
how far and the extent to which specific project objectives have or have not been attained.
There are two types of indicators, namely process and results indicators.
Process Indicators
Process indicators provide information on the activities that are being implemented in terms of
types of activities, the number, who the activities are directed at, etc. These indicators provide
information that would enable us to determine if an intervention is moving in the right direction
in order to achieve the stated objectives. This type of information is collected throughout the life
of the intervention. Examples of process indicators are: (i) Number of people who have visited
the San Juan Community Mediation Centre in the last quarter of 2002; (ii) Number of cases that
have been successfully resolved at the San Juan Community Mediation Centre in the first quarter
of 2003.
Process indicators are useful for monitoring. Data collected using process indicators help in
determining the reasons for the success or failure of an intervention. For example: Number of
participants at the M&E training programme organised by the Ministry of Social Development.
15
Results Indicators
These types of indicators are closely linked to the stated objectives of an intervention. They are
meant to provide a framework for assessing whether or not as a result of the intervention there
has been a visible change in the circumstances of the beneficiary population. The extent of
change can be measured at the programme level or the population level. Results indicators are
expressed as a percentage, ratio or proportion. It provides a basis for assessing the degree of
change in relation to the beneficiaries and/or their environment. Although results indicators are
meant to measure change, they should not be anticipative. For example: instead of reduction in
the number of teenage pregnancies it is more appropriate to write percentage of adolescent
girls of 10-19 years old who have had babies in the last year5
Results Indicators and Objectives
Since results indicators provide an indication of whether or not an objective has been achieved, it
is advisable to include at least one result indicator when designing the intervention.
Principles for selecting Results Indicators
Indicators should be precise and clear. If indicators are written as percentages both the
numerator and the denominator should be specified. For example: Percentage of beneficiaries
of the SHARE programme who have successfully set up a micro enterprise.
Number of beneficia ries who have successfully set up micro enterprises Numerator
The total number of beneficiaries participating in the SHARE programme Denominator
Criteria for Selecting Indicators
There should be emphasis on the selection of indicators that are clear and concise. The
following criteria must be considered when selecting indicators 6 :
Relevance
Reliability
IPPF/WHR, Guided for Designing Results-Oriented Projects and Writing Successful Proposals,
December 2000, p18
Ibid p18
16
Validity
Availability of
information
Ease in
measuring
Easy to
understand
Cost effectiveness
Robustness
Timeliness in
data collection
Data collection and analysis should take place within a welldefined timeframe in terms of the frequency of data collection and
the currency of the data.
It may be prudent to select more than one indicator for an objective. This is because an objective
may have different dimensions and complexity. In view of time constraints and in order to be
cost effective, it is advisable to identify only a few indicators that are manageable.
An example of a poor indicator:
Increase in the percentage of youth using condoms during first sexual experience.
An appropriate indicator:
Percentage of sexually active youth (ages 15 to 24) participating in a sensitisation project that
report- using condoms at last intercourse by age and sex.
17
References:
1.
2.
3.
Peter H. Rossi et al, Evaluation: A Systematic Approach, 6th Edition, Sage Publication,
1999.
International Planned Parenthood Federation, Guide for Designing Results-Oriented
Projects and Writing Successful Proposals, Western Hemisphere Region, Inc, December
2002.
Susan Adamchack et al, A Guide to Monitoring and Evaluating Adolescent Reproductive
Health Programmes, Focus on Young Adults, Tools Series 5, June 2000.
18
19
2.
Programme
A programme consists of several activities or projects with defined goals and objectives that aim
at improving the social and economic circumstance of the beneficiaries. An example is the
SHARE programme.
Note that both projects and programmes can be subjected to mid-term and final evaluations. A
mid-term evaluation is undertaken in order to determine the overall effectiveness, efficiency, and
impact of a project or programme. Findings from a mid-term evaluation could result in making
changes to the project/programme. A final evaluation or ex-post evaluation is an assessment of a
project/programme after it has been completed.
3.
Thematic Evaluation
This type of evaluation focuses on selected interventions within a sector that addresses specific
priorities. For example: teenage mothers, drug addicts, and alcoholism.
4.
Sector Evaluation
Sector evaluation focuses on a cluster of development interventions.
education, training, agriculture, micro-enterprise, etc.
5.
Country Evaluation
This type of evaluation is common with donor-funded programmes/projects.
organisation can decide to evaluate its activities in a given country.
A donor
6.
Programme Effectiveness
It is often assumed that once a policy/programme is initiated one can expect successful
implementation. This is always not the case since the intervention may not have been effectively
implemented. This could be due to poor design, inadequate inputs and a host of many reasons.
This explains why a key component of evaluation is to focus on how well a programme has been
implemented by looking at the inputs, processes, outputs and outcomes.
Inputs
This refers to resources (i.e. personnel, facilities, space, equipment and supplies) that make it
possible for successful implementation of a project/programme.
Programme Processes
This refers to the use of inputs to undertake activities, which will lead to the realisation of the
objectives that have been designed for the intervention. Programme processes range from
management procedures to research and evaluation systems.
20
Programme Outputs
These are the results obtained at the programme level through activities and use of inputs.
Programme outputs can be classified into functional outputs, service outputs and service
utilisation. Some examples of functional outputs are: the number of beneficiaries trained to set
up their own business; the number of training sessions conducted; etc. Service outputs focus on
access to services and quality of care, such as number of drug addicts who are currently being
treated for addiction and number of those treated who are satisfied with the type of services
being offered at the treatment centres. Service utilisation denotes the ability of a programme to
retain repeat clients and more significantly to attract new clients, such as the number of new
clients who have access to the SHARE programme.
Programme Outcomes
These are results that occur after implementation of activities. Outcomes can be at the
programme level or the population level and are medium term or long term in nature.
The relationship among the programme components can be illustrated as follows:
Input
Process
Output
Outcome
In sum, evaluation that focuses on programme effectiveness/processes attempt to find out what
works and what does not work when carrying out the implementation of activities. The
following issues are often considered: How well is the programme functioning? To what extent
are the services delivered as intended? Are the services being provided congr uent with the needs
of the target beneficiaries? What are the management capabilities of the implementers? 7
Bertrand, Magnani and Tutenberg, Evaluating Family Planning Programes. The Evaluation Project, September
1996
21
7.
Programme Efficiency
It is often said that knowledge of programme results is not sufficient enough to declare success in
producing outputs and outcomes. Results must be measured against their costs. Due to
competing demands on the resources of the government, it behoves programme managers to
demonstrate that their programmes are not excessively expensive and that all things considered,
their programmes are providing value for money. Programmes could be terminated or retained
on the basis of their comparative costs. Of course in the realm of politics, it is not always
feasible to kill a programme as a result of inefficiencies or cost over runs. Inefficient
programmes may be kept purely for political expediency.
The following questions should be asked when conducting efficiency assessment:
Is a programme producing sufficient results in relation to the overall costs/resources
deployed?
Which approach to providing the service results in the lower cost per unit in relation to
the outcome?
22
There are two main techniques for assessing programme efficiency, namely cost-benefit
analysis and cost-effectiveness analysis.
Cost-Benefit Analysis
Cost-benefit analysis focuses on the relationship between programme costs and outcomes with
both cost and outcomes expressed in monetary terms:
Hospital Care
Salaries
Overheads
Food
Utilities
Hospital bed, etc.
Community Care
Food
Rent
Training
Service delivery, etc.
To over simplify, the benefits derived from a programme should outweigh the cost of providing
it. This method could be applied to a single programme, two programmes with the same goals
competing for the same funds, or two programmes with different goals competing for the same
funds.
Generally speaking, cost-benefit analysis is difficult to do because it requires making
assumptions about the monetary va lue of programme benefits.
Cost-Effectiveness
Cost-effectiveness analysis examines the relationship between programme costs and outcomes,
in terms of the cost per unit of outcomes achieved. This method compares two programmes
which both aim to achieve a comparable outcome. For example, the unit cost of rehabilitating
one drug addict could be $100,000 per year in Programme X but $80,000 per year in Programme
Y, making Programme Y more cost effective. Cost effectiveness could also be considered from
more than one point of view, for instance, the cost-effectiveness of two dispute resolution
options (i) to the client or (ii) to the state.
23
24
Impact assessment determines the extent to which a programme delivers on the intended
objectives in such a way that it results in improvements in the social conditions of the target
beneficiaries. Some questions to consider when doing an impact assessment:
To what extent could programme outcomes be attributed to the intervention?
To what extent did the programme succeed in producing change in the social conditions
of the beneficiaries?
To what extent can one attribute changes that have occurred to the specific interventions?
These questions seek to establish a cause and effect relationship. The bottom line is to
establish the net effect of an intervention. In order to do so, it is useful to define outcome
variables. It may be possible to use classic experimental design (treatment and control
group-subjects with the same characteristics randomly selected.) This type of design is
useful for a pilot project-See Module 4.
Programme objectives designed for impact assessment:
Changing behaviour e.g. sexual behaviour to avoid contracting sexually transmitted
diseases
Lowering functional illiteracy
Reducing homelessness
Minimising teenage pregnancies
Reducing incest
Reducing child abuse
Reducing unemployment among the youth
A Diagram Depicting Impact Theory/Assumptions:
Skills
Job
training
Job
search/Int
.
Better job
Improved
Economic
Status
Motivation
to work
25
9.
Programme Sustainability
It denotes the extent to which an intervention can continue to be viable and produce benefits after
the completion or closure of the interventio n.
10.
Overall Programme Evaluation and Management Audit (OPE/MA)
An OPE/MA is an evaluation that covers all facets of programme evaluation components such as
those described above.
Evaluation Assessment
Evaluation assessment studies are undertaken to ensure that resources designated for evaluation
are judiciously used to answer the right questions in an appropriate and credible manner. It may
be noted that evaluation assessment is not a substitute for the actual evaluation. It is meant to
underscore the importance of creating demand and a buy- in of the findings. The ultimate
objective is that the findings will be accepted and used.
What is the rationale for undertaking evaluation assessment?
The following are the underlying reasons for initiating an evaluation assessment:
To understand the intervention and the environment in which it operates
To define the purpose of the evaluation
To identify the users of the evaluation findings (stakeholders)
To define key evaluation issues and related questions
To identify key evaluation methods and strategies which would be utilized to determine
the availability of data and quality of data which is available (if any)
To estimate the cost and resources required for data collection, data analysis and
producing the report
To define the timeframe
To formulate options in terms of the questions and methods to be used to collect data
To formulate recommendations.
The Intervention and its Context
It is important to have a good understanding of the intervention in terms of its goals, objectives,
targets, the activities, the reach and the expected outcomes. It may be useful to establish
plausible relationships among the activities. Discuss the origins of the intervention by reviewing
legislative instruments, Acts of Parliament, etc. and assess whether or not the intervention is
26
doing what is it was set up to do. Review the appropriateness of the underlying assumptions, the
programme reach and determine if the expected outcomes are still realistic. An example of a
programme profile is presented in the box below:
Profile of an Intervention
Typically an intervention should have the following core profile:
Goal:
Description:
Objectives:
Target Group: Identify who are the intended beneficiaries of the intervention.
Activities:
Expected
Outcomes:
Budget:
Stakeholder Analysis
Assess the main actors of the intervention such as programme managers, programme staff, policy
makers and beneficiaries. Indicate how the evaluation will be of immediate benefit to them.
Define Key Issues/Questions
There is a need to take into consideration the information requirements or needs of the
stakeholders, namely decision- makers, programme managers and beneficiaries. Try to tailor the
evaluation questions to answer the issues that are of importance to them. In other words, the
concerns of the stakeholders must be translated into specific questions. The questions will form
the basis of data collection. It is advised that the list of questions should be shared with
programme managers and policy makers for their input. Below are samples of questions:
27
Programme Rationale:
Is the intervention still relevant?
Is it meeting the purpose for which it was designed?
To what extent are the activities and outputs consistent with the stated goals?
Objectives Achievement:
Did the intervention achieve the stated objectives?
How realistic were the objectives?
To what extent were the targets achieved?
Were the objectives in line with the needs of the beneficiary population?
Outcomes and Effects:
Did the intervention lead to a change in the circumstances of the beneficiary population?
Were there any unintended effects?
Were there any catalytic effects?
Programme Efficiency:
To what extent were the resources (i.e. financial, human, equipment, etc) appropriately
utilized?
Were any savings made?
Is there an alternative way of achieving the stated objectives in a cost-effective manner?
Define Evaluation Methods/Strategies
The evaluation must describe the evaluation approaches that will be used to answer the defined
questions. The methods selected should be time-tested/rigorous to facilitate the collection of
appropriate information to meet the needs of the client(s). The methods should be consistent
enough to allow for both internal and external validity.
Identify Tools for Data Collection
Data collection tools must be rigorous. However it is to be noted that evaluation is not a
longitudinal research to produce knowledge or definitive statements. It is a tool for providing
timely information that is appropriate for policy and decision-making. Here are a few examples
of data collection tools: rapid assessment, focus group discussions, case studies, time series
analysis to detect changes over time, pre- and post-intervention surveys, etc. If the size of the
beneficiary population is too large, it may be useful to use some sampling techniques. There are
two types of sampling techniques: non-probability sampling (i.e. purposive samples and quota
samples) and probability sampling (i.e. random samples, stratified samples and cluster samples).
See Module 4 for further discussion on sampling.
Ministry of Social Development/Monitoring and Evaluation Division September 2003
28
Data can be obtained through interviews, survey questionnaire, direct observation, review of
documents and records, etc. It is advisable to use several sources of information to minimize
personal biases (i.e. data triangulation).
Estimation of Cost and other Resources
Estimate the total cost of the proposed evaluation, including other resources. Describe the team
composition and the competencies/skills required to undertake the evaluation. Do an
implementation plan with specific time lines/benchmarks.
How long does it take to do an Evaluation Assessment?
Depending on the scope and complexity of the intervention, an evaluation assessment may take
up to a month to complete.
Conclusions and Recommendations
The conclusion should entail a recommendation regarding the specific evaluation questions that
could be covered during the actual evaluation. Options should be given with regard to possible
evaluation approaches and the related costs of each approach.
Scope of the Evaluation
Once a decision is reached to undertake an evaluation, the next step is to prepare the scope of the
evaluation to clarify the programme context, what issues to focus on, who will do what, when it
will be done and the resources required for undertaking the evaluation.
Step 1:
Define the programme context. Define the goals and objectives of the programme
including the conceptual framework that maps out the linkages between inputs,
processes, outputs, and outcomes.
Step 2:
Step 3:
Define the methodological approach in terms of how you intend to carry out the
evaluation (i.e. study design, indicators and data sources).
Step 4:
Describe the implementation plan in terms of the main deliverables, who (individuals
and institutions that will be responsible for aspects of the evaluation), when (time
table for carrying out specific activities) and with what funds (budget).
Step 5:
29
References:
1.
2.
3.
4.
5.
6.
30
Time
O1
31
Intervention (X) is introduced and sometime after the introduction, an attempt is made to
undertake some measurement (O 1 ). Considering that there is no control group, it is not possible
to make a comparison. O1 provides descriptive information and other data that could be useful
for planning an intervention. It is used for conducting diagnostic studies to determine why a
problem exists.
Pre Test and Post Test Design
(Experimental Group)
Time
O1 X
O2
There is no control group for this type of design. However, there is an earlier measurement
observation (O 1 ) that makes it possible for the evaluator to assess changes overtime after an
intervention (O 2 ). If there is a significant change in the circumstances of the beneficiaries then
one could attribute those to the intervention. The findings could be influenced by threats to
validity, history, maturation, etc.
Static Group Comparison
(Experimental Group) Time (Control Group)
X O1
O2
In this design, the experimental group receives an interve ntion (X) followed by observation (O 1 ).
This measurement observation is compared against a second observation (O 2 ) from a control
group that did not receive the intervention. The design is not based on random sampling, which
means that there are differences between the two groups (not equal) and hence strong threats to
validity.
Quasi Experimental Design
We had indicated earlier on that due to ethical, programmatic and political considerations, it
might not always be possible to implement a true experimental design. Nonetheless, a similar
design could be applied to obtain almost identical results by using quasi experimental design
such as time series design.
Time Series Design
Time series design is similar to the nonexperimental, pretest | post-test design, except that it
has the advantage of repeated measurement observations before and after programme
intervention (X).
32
Experimental Group
O1
O2 X
O3
Time
O4
O5
O6
We may initially find that there is no difference between O1 and O2. Then it is observed that
there has been a sudden improvement in the circumstances of the beneficiary group, which is
observed by O3 and O4 . The increase is subsequently maintained in O5 and O6 . Therefore, it can
be concluded with some degree of confidence that the increases were probably due to the effects
of the intervention (X). This design is often used to perform trend analysis in terms of what
pertained before and after the intervention.
It would be interesting to apply this approach to assessing changes brought about by any of the
existing social sector programmes. For example to determine if, as a result of the distribution of
hampers under the SHARE Programme, the nutritional in take of the beneficiaries has indeed
improved.
Issues to Consider when Selecting an Evaluation Design
The following should be taken into consideration when selecting a design.
Ethical issues Try to avoid a design that could lead to the application of unethical
procedures, a violation of peoples rights and dignity, or a denial of services that would
otherwise be available.
The ideal situation is the use of experimental design with randomly selected subjects
based on a single population. If this is not possible, find a comparison group that is
nearly equivalent to the experimental group.
If a randomly assigned control group is not available or there is no similar comparison
group, you may consider using a time series design that can provide information on
trends before and after the programme intervention.
If it is not possible to use a times series design, try to obtain baseline (pre-test)
information that can be compared against post intervention information (pre-test post-test
design = O1 X O2 ).
The issue of validity should be considered at all times by asking such questions as: How
true are the measurements? Are there possible threats to validity? etc.
Another principle to bear in mind is that a good design does not stop the provision of
services.
Always use multiple data sources to obtain information, this is also known as data
triangulation.
33
34
Sampling
A distinction must first be made between a total population and a sample of that population. A
population contains elements, each of which is a potential case. If an evaluation involves a small
population, all the elements can be studied. For example evaluating the health status of 50 people
living with HIV in a hospice in Arima. Since 50 people is a relatively small population, there is
no need to draw a sample, the entire population can be evaluated. On the other hand, it would be
costly and time consuming to evaluate the health status of the entire population of 50,000 people
living with HIV in Trinidad and Tobago. One may need to draw up a sample of elements/cases
within the 50,000 population to study.
Usually, the practice is to concentrate on a select few or a representative sub-set to interview. If
the sample were truly representative, the information obtained would be similar to the
information that would be obtained from the entire population.
There are two basic types of samples.
samples.
35
Systematic Sampling
This is a modification of simple random sampling which is less tedious and less time consuming.
The estimated number of elements in the larger population is divided by the desired sample size,
yielding a sampling interval. Let us call it n. Using a sample frame with the population
elements listed in an arbitrary order and selecting every nth case, starting with a randomly
selected number between one and n, the sample size is drawn.
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20
Systematic drawn cases are 1 5 9 13 and 17
Stratified Sampling
Stratification can be used for either simple random sampling or systematic sampling to ensure
the desired representation of specific sub groups. For example, elements in the larger population
can be arranged by age, education, income, location, profession, political affiliation, etc.
Cluster Sampling
This method is used to simplify sampling by selecting clusters of elements, using simple random,
systematic, or stratified sampling techniques and then proceeding to study all the elements in
each of the sampled clusters. Usually the clusters are geographic units such as provinces,
districts, towns, villages, units or organisational units such as centres, clinics, or training groups.
NonProbability Sample
This is also known as convenience sampling. It refers to the selection of cases that are not based
on known probabilities. The selection could be accidental (choosing from whatever case that is
available) or purposive (selection from specific type cases). This type of sampling is not
representative of the larger population since there can be over selection or under selection of
cases. If it is too expensive to use probability-sampling technique, then non-probability
sampling may be the most appropriate method to use.
Sample Size
The size of the sample is determined by two main things, namely the availability of resources
and the proposed plan of analysis. For example, if you intend to analyse cross-tabulations of
variables, take into consideration that each category of an independent variable included in crosstabulation must contain at least 50 cases.
36
Confidence Level
In practice it is recommended that a study should aim at obtaining a 95% confidence level that
has a + or - 5% interval/error.
Example of sample sizes
Population = 500
Population = 8,000
Population=500,000
95%|+/-5%
99%|+/-5%
95%|+/-5%
99%|+/-5%
95%|+/-5%
99%|+/-5%
217
286
367
615
384
665
References:
1. Susan Adamchak, A Guide to Monitoring and Evaluating Adolescent Reproductive
Health Programmes, Tools Series, 5 June 2000
2. Michael Q. Patton, Utilization Focused Evaluation, Sage Publications, 1978
3. Andrew A. Fisher et al, Handbook for Family Planning Operations Research Design,
Population Council, New York, 1998
4. Bertrand et al, Evaluating Family Planning Programmes, The Population Project,
September 1996
37
38
Qualitative
It makes it possible to collect information
from respondents whose views are based
on gut feelings
Helps to probe social and cultural
attitudes
Allows for probing for unintended results
Allows assessment of goals that are not
amenable to quantitative analysis. For
example: empowerment, self-esteem,
negotiation skills, etc.
39
Coding
It may be useful to develop a codebook as part of designing a questionnaire. A numerical or
symbolic code may be assigned. For example, to find out the level of education of respondents,
the responses may be coded as follows:
(1) None
(2) Primary
(3) Secondary
(4) Tertiary
Data Analysis
Once data is collected, either qualitative or quantitative, the next step is analysing it. There are
various techniques for analysing both qualitative and quantitative data.
Analysing Qualitative Data
Qualitative data is often presented in a narrative form. It is not always feasible to assign a code
or even a numerical character to qualitative data. Instead qualitative data can be coded as
categories (thematic coding) and presented as a narrative.
The following shows how qualitative data can be categorised and presented:
1. Case Studies Based on narratives or interpretations of respondents understanding of
the workings and benefits of an intervention.
2. Process Analysis - Visual depiction of a programmes processes and outcomes.
3. Causal Flow Charts Shows how things work.
4. A decision Tree Model This graphically outlines the realm of choices and priorities
that go into decision- making.
5. Taxonomy A visual representation/diagram showing respondents relate categories of
language and meaning.
Analysing Quantitative Data
This usually involves mathematical calculations through the application of statistics. The most
commonly used statistics are descriptive and inferential statistics.
Descriptive Statistics
Descriptive statistics is the first step in quantitative analysis. Descriptive statistics are used to
describe the general characteristics of a set of data. Descriptive statistics include frequencies,
counts, averages and percentages.
This method is used to analyse data from monitoring, process evaluation and outcome /impact
evaluation.
Frequencies
A frequency denotes a univariate (single variable) number of observations or occurrences. For
example, when you say that 100 out of 200 homeless people have been given shelter you are
40
stating a frequency. When the frequencies related to a single variable (homeless people) are
listed together, it is known as a frequency distribution. For example, of the 300 people listed as
homeless, 100 were given shelter, 105 were sent for a medical check up, 95 received training in
basic skills, etc. Data for a single variable (homeless people) can be further disaggregated suc h
as of the 100 homeless people who were given shelter, 50 were females and 50 males. Of the
100 homeless, 10 were young people of between 10-18 years of age. This is known as bivariate
and multivariate.
Bivariate and multivariate frequencies can be classified and presented in a table format. This
display of labelled rows and columns is known as cross tabulation (See example below).
Number of Children by Disability Type
District
Hearing Learning
Caroni Educational District
Other
Physical
Visual
All
225
1,210
29
82
418
1,964
47
16
32
104
80
2,054
65
52
172
2,423
86
620
84
97
318
1205
165
1,853
174
68
257
2,517
123
665
94
90
332
1,304
10
400
15
435
54
275
63
400
750
7,124
471
400
1607
10,352
All Districts
Percentages
Percentages are calculated by dividing the frequency in one category by the total number of
observations then multiplying by 100:
Variable
Shelter
Medical Check Up
Males
Females
Total
Frequency
100
100
50
50
300
Percent (%)
33.3
33.3
16.7
16.7
100
41
Inferential Statistics
Inferential statistics allow the evaluator to make inferences about the population from which the
sample was drawn, based on probabilities or stratified sampling. Testing for statistical
significance helps to ensure the differences observed in data, however small or large, were not
due to chance.
Preparing an Evaluation Report
See Step Nine of Appendix I.
Developing a Solutions Framework
Sample of a Solutions Framework (Rapid Assessment of the Mediation Centres):
Time Frame
Person/Organization
Responsible
Aug-Oct 2003
Sep-Oct 2003
Directors
Mediation Centres
Sep-Oct 2003
Recommendations
Priority
Level
References:
1. Peter H. Rossi et al, Evaluation: A Systematic Approach, Sage Publications, 1999
2. Susan Adamchak, A Guide to Monitoring and Evaluating Adolescent Reproductive
Health Programmes, Tools Series, 5 June 2000
3. Michael Q. Patton, Utilization Focused Evaluation, Sage Publications, 1978
4. Andrew A. Fisher et al, Handbook for Family Planning Operations Research Design,
Population Council, New York, 1998
5. Bertrand et al, Evaluating Family Planning Programmes, The Population Project,
September 1996
42