1 Monitoring and Evaluation Use of statistics Module 5 Session 9
Mar 28, 2015
1
Monitoring and EvaluationUse of statistics
Module 5
Session 9
2
Use of Statistics in Monitoring and Evaluation
1. Why consider statistics in M&E?
2. Where are the entry points for using stats?
3. Improving the quality of indicators
4. Enhancing monitoring of indicators
5. Informing evaluations
3
1. Why use statistics in M&E?
“When you can measure what you are speaking about and express it in numbers, you know something about it; but when you cannot measure it, then you cannot express it in numbers, your knowledge is of the meager and unsatisfactory kind.” – Lord Kelvin (British Physicist)
4
2. Entry points
Planning – informing the selection of indicators
Monitoring – measuring progress against quantitative indicator targets
Evaluation – using statistics from different sources to inform on achievement in context
5
3. Planning – enhancing the quality of indicators
Definition:Indicators are signposts of change along the path to development.
Indicators are what we observe in order to verify whether – or to what extent – it is true that progress is being made towards our goals, which define what we want to achieve. Indicators make it possible to demonstrate results. Indicators can also help in producing results by providing a reference point for monitoring, decision-making, stakeholder consultations and evaluation. In particular, indicators can help to:
Measure progress and achievements; Clarify consistency between activities, outputs, outcomes and
goals; Ensure legitimacy and accountability to all stakeholders by
demonstrating progress; Assess project and staff performance.
6
Indicator qualities
Good indicators have the following five characteristics: Numeric. While not always more objective, numerical precision
lends itself better to an agreement over the future interpretation of data. On the other hand, factual indicators provide only a very crude “measurement” due to their limited scale (mostly yes/no). For a set of factual indicators, no monitoring system is needed, since the status is mostly known by stakeholders (e.g.: law passed by parliament: yes or no). If it not possible to avoid factual indicators due to the nature of the project, factual indicators should be at least be supplemented by numeric indicators in a comprehensive set.
Objective. An indicator which involves a subjective judgement by somebody is not objective. For a good indicator, there has to be a general agreement over interpretation of data.
7
Indicator qualities (2)
Specific. The indicator needs to be as specific as possible in terms of quantity, quality, time, location, target groups, baseline, targets etc.
Relevant. The indicator needs to relate directly to the respective output, outcome or impact. In other words, a good indicator is a relevant “measure” for the objective.
Feasible. Even if an indicator fulfils all other criteria, it is not useful if the data collection for the indicator is not feasible. First, data for the indicator needs to be easily available.
8
Examples
Numeric indicators
� Number � Nr. of entrepreneurs trained
� Nr. of new jobs created by small enterprise sector
� Percentage � % of government budget devoted to social sectors
� % of rural population with access to basic health
care
� Ratio � Ratio of doctors per 1.000 people
� Ratio of female to male school enrolment
9
Description of indicatorPrecise definition: The definition must be detailed enough to ensure that different people at
different times, given the task of collecting data For a given indicator, would collect identical types of data. Potentially
ambiguous terms (for example: small farmers, poor households, disadvantaged groups) need to be clearly defined (for example: farmers with < 1 hectare of land, households below national poverty line).
Unit of measure: Define the precise parameter used to describe the magnitude or size of
the indicator (for example: number of individuals, percentage, shillings, hectares, cumulative, average, etc.).
Disaggregated by: Identify how data will be separated to improve the breadth of
understanding of results reported (for example: gender, district, urban/rural, etc.).
10
Baseline and Targets
Baseline: The baseline is the value of the indicator prior to an action. The
baseline value establishes the starting point from which change can be measured.
Benchmarks: Benchmarks are values of the indicator while an action is still
ongoing. Benchmarks are therefore intermediate targets.
Target: The targets is the expected value of the indicator after an action.
11
4. Thinking about monitoring
Planning for data acquisition Attributes of good data Timing of relevant data collection exercises
12
Planning for data acquisitionData collection method and timing: Describe exactly and in detail how and when you will collect the data.
Identify what methods and instruments you will use. Note any tool or survey required to collect the data. Attach data forms
when necessary. Examples of data collection methods are secondary data, surveys, expert judgments, etc.).
Data source: The data source is the entity from which the data are obtained (e.g. a
government department, an NGO, other donors, etc.).Estimated cost of data acquisition: Provide a rough estimate of what it will cost to collect and analyze the
data.Individual responsible and location of data storage: Describe who will take the lead for collecting this indicator. Describe
how data will be stored over time and in what format.
13
Data quality issues
Known data limitations and significance: Identify where data may be weak or limited.
Identify actions taken or planned to address data limitations:
14
Attributes of ‘good’ data The source of data should be known. Typically
someone at the agency that collected the data should be available to clarify and explain details;
The reason for which quantitative data was collected should be known and documented. This is important because it helps to understand if for example there were any biases in response or if you need to account for interviewer/response biases/errors;
Codified responses should be carefully documented at some place. Frequently responses are ranks or ordinal responses;
15
Attributes of ‘good’ data (2) The date of collection, scale, frequency, sampling unit, enumeration
units, selection process and coverage of data should be known. Thus the number (scale) of households (sampling unit) covered on a certain date, should be known. The number of times it is collected (frequency) and what region of the country it covers (coverage) should be known. If data is collected on livestock then the units that this is recorded in (thousands or millions) should be known. If the data does not include all the observations as initially planned, it is good to know the reason for this;
The method used to identify respondents should be known. There are many methods to collect data and identify respondents. These can be random sample over a country, or a village, or variations thereof. Thus for example if data was collected randomly or to reflect proportionate representation of ethnic groups or in an interview format – then all of this should be known. Data can via a rapid rural assessment. The method for identifying respondents helps to know the extent to which data can be scaled-down or disaggregated.
16
Attributes of ‘good’ data (3) Spatially explicit data that is either derived from
satellite images or ground collected are also another type of data. Documentation for these data is also extremely important. Additional knowledge about scale and format of data is critical.
Errors or faults in data collection (there are always some) should be known well.
People working with the data should be very familiar with the software they use to extract data, transform data and construct data. Frequently a lot of errors in analyses occur because researchers are not familiar with the way their software handles data.
17
Timing of data collection exercises
Monitoring indicators require updating quarterly or biannually.
Identify which reliable instruments may provide some useful data for district level monitoring (e.g. consumer price index in urban centres; biomass data; migration statistics etc)
18
5. Informing Evaluations
What type of data do evaluations require? What sources of data exist? How to access and manage that data?
19
Evaluation data requirements Evaluations ask the question of not only ‘what’ has
and has not been achieved, but also ‘why’ it has or has not been achieved.
The ‘what’ question requires answers to often quantitative indicators. For example, the PEAP evaluation needs to address the question of whether poverty headcount has gone down as per the target.
Equally, an agricultural project evaluation might ask ‘what’ has changed against indicators such as:
Increased yield/acre Expansion of acreage Share of high value crops in total production
20
Evaluation data requirements (2)
The ‘why’ question of evaluation requires a broader investigation into data:
Explanatory data: What else has been occurring that may have affected the intervention (e.g. reduction in international coffee prices; heavily floods in area etc)
Trend data: what has occurred over time in key indicators
Mixing quantitative and qualitative data: using qualitative approaches (e.g. focus groups) to better understand possible cause and effect relationships
21
What sources of data exist?
Best sources of quantitative data are:Surveys (representative at regional level)
Demographic health survey Household budget survey National household survey Sero behavioural survey National service delivery survey Informal cross-border survey
Other Evaluations and Reviews
Censuses (population, agriculture, etc)
22
“Without data, all you are is just another person with an opinion” - Unknown
Case Study Purpose Selection
QuotaSampling
SmallSample
LargeSample
Census
Direct Measurement
Questionnaire(quantitative)
Questionnaire(qualitative)
Structured meeting
Open Meetings
conversations
analytical games
participatoryobservation
Follow up sentinels sites
beneficiary evaluation
LSMS/multi-purpose
Income-ExpenditureSurvey
Census
Employment Survey(households)
Community Surveys
Focal Groups
Source: Statistical Literacy Project, Bureau of Development Policy, UNDP
23
How to access that information
Survey data sets are available in reports, and raw at UBOS
Partner evaluation reports are available online, e.g. World Bank
N.B. Often necessary to visit key offices to identify and access key data sets