Ethics are a set of values and beliefs that are based on a person’s view of what is right, wrong, good and bad and that influence the decisions people make. They can be dictated by the organization and also by laws in the country in which the M&E practitioners work and what people consider to be ethical in that context

The gathered data provide an important source of information to decision makers about the intervention being monitored and/or evaluated. While positive evaluations can help secure more funds, expand a pilot project or enhance reputations, the identification of serious problems can lead to difficult situations where the credibility of the work done is at stake. Understanding and managing political situations and influence is crucial for maintaining the integrity of the monitoring and evaluation work and well defined and robust methodologies for the data collection and analysis play a critical role.4

Ethical guidelines and principles

When planning, designing, implementing, managing and reporting on M&E activities, M&E practitioners should ensure that their actions are informed by ethical guidelines, particularly those outlined below:

• IOM Evaluation Policy and Monitoring Policy (September 2018);
• IOM Data Protection Principles (IN/00138) (May 2009) (Internal link only);
• IOM Standards of Conduct (IN/15 Rev.1) (Internal link only); 
• UNEG Ethical Guidelines for Evaluation (March 2008, revised 2020).

Some of the common ethical principles presented in the above documents “should be applied in full respect of human rights, data protection and confidentiality, gender considerations, ethnicity, age, sexual orientation, language, disability, and other considerations when designing and implementing the evaluation”.5 They can be summarized as follows:

Adhering to common ethical principles also contributes guaranteeing that the information gathered is accurate, relevant, timely and used in a responsible manner (see chapter 2, as well as Annex 2.1. Ethical monitoring and/or evaluation checklist).

• Independence also means avoiding conflicts of interest and being able to retain independence of judgement and not be influenced by pressure from any party to modify evaluation findings.

• 4 Morra Imas and Rist, 2009. Please also see chapter 2 on norms, standards and management for monitoring and evaluation
• 5IOM, 2017b, p. 438.

When planning for data collection, basic considerations ensure that the data to be collected and analysed is valid and reliable: purpose for data collection, methodology for data collection, resources for data collection and timing for data collection. Qualitative, quantitative or mixed methods approach to collect data can be considered in that respect.

Purpose for data collection	Some questions to ask: Is the data collected for the purpose of monitoring or evaluation?  What are the main information needs? What are the objectives, outcomes, outputs and activities being monitored or evaluated? What are the expected results (intermediate and long term)? Which stakeholder’s information needs will the data address?
Methodology and methods for data collection	Several aspects need to be considered, such as identifying the source of the data, the frequency of data collection, knowing how data will be measured, by whom and how many people will collect data and selecting the appropriate methodology in order to design the right data collection tool/s Some questions to ask: What are the criteria and questions to be addressed in the data collection tools?  What type of data is needed to answer the information needs?  Are multiple data sources required/used to answer the information needs? What types of data already exist? What data is missing? Are the measures used to collect data valid and reliable? Will a structured or semi-structured approach be used to collect the data? What sampling approach is needed to monitor progress or answer the evaluation questions?
Resources for data collection	Resources will be enabling the implementation of choices. Some questions to ask: Are there enough resources, such as staff and budget, to collect the data on a frequent basis? Who is responsible for collecting the data? Will external enumerators need to be hired? How will enumerators get to the data collection sites? What are the related costs? Are additional costs for data collection and analysis required?
Timing for data collection	Timing may influence the availability of resources, as well as the relevance of data (avoid outdating of data). Some questions to ask: At which stage of the implementation cycle will data be collected? How long is data collection expected to last? Will data be collected in a timely manner for it to reflect a current status quo?

Identifying the purpose of data collection

Identifying the purpose of data collection aims to address different information needs, and information needs of monitoring may also differ from those of evaluation.

Data collection for monitoring, which occurs during implementation, feeds implementation-related information needs, using data collection tools that are designed to collect data for measuring progress towards results against pre-set indicators. Data collected for evaluation serves the purpose of assessing the intervention’s results and the changes it may have brought about on a broader level, using data collection tools designed to answer evaluation questions included in the evaluation terms of reference (ToR), matrix or inception report (see also chapter 5, Planning for evaluation).

The process of planning and designing the respective tools for M&E data collection may be similar, as data collected for monitoring can also be used for evaluation, which will feed the diverse information needs of either. Identifying whether data collection is for either monitoring or evaluation purposes is a first step in planning, which will then influence the choice of an appropriate methodology and tools for data collection and analysis. The following tables show how questions can determine what type of data to collect respectively for monitoring and evaluation.

 

Source: Adapted from International Federation of Red Cross and Red Crescent Societies (IFRC), 2011.

Source: Adapted from IFRC, 2011.

The IOM Project Handbook defines data sources as identifying where and how the information will be gathered for the purpose of measuring the specified indicators.6

In general, there are two sources of data that can be drawn upon for monitoring and/or evaluation purposes:

• Primary data, which is the data that M&E practitioners collect themselves using various instruments, such as key informant interviews, surveys, focus group discussions and observations.
•  Secondary data, which is data obtained from other pre-existing sources, such as a country census or survey data from partners, donors or government.

• Note that in cases where IOM works with implementing partners that are directly managed by IOM, the data collected is still considered primary data collected by IOM.

• 6 Module 2 of IOM Project Handbook, p. 143

When choosing sources of data, it is helpful to start with a desk review to better assess what type of data to use. For monitoring, this corresponds to the information included under the column “Data source and collection method” of the IOM Results Matrix and Results Monitoring Framework (see chapter 3). For evaluation, the type of data will be clarified in the evaluation ToR, inception report and/or evaluation matrix and can also include data derived from monitoring.

A desk review usually focuses on analysing existing relevant primary and secondary data sources and can be either structured or unstructured. Structured desk reviews use a formal structure for document analysis, whereas unstructured reviews are background reading. For detailed guidance on conducting a desk review, see Annex 4.2. How to conduct a desk review.

When planning for data collection and analysis, knowing the type of measurement, that is how data will be measured, may influence the decision to choose the appropriate methodology. This is of particular importance to inform the design of data collection tools such as surveys.

Questions to consider
	What is it that you want to measure?  What is the purpose of measuring it? How will you go about measuring it?

Measures of indicators identified in a Results Matrix or Evaluation Matrix can include categorical (qualitative) and/or numerical (quantitative) variables. A variable is any characteristic or attribute that differs among and can be measured for each unit in a sample or population (see section on “Sampling”).

• Categorical variables represent types of qualitative data that can be divided into groups or categories.7 Such groups may consist of alphabetic (such as gender, hair colour or religion) or numeric labels (such as female = 1, male = 0), or binary labels (such as yes or no) that do not contain information beyond the frequency counts related to group membership.8
• Numerical variables (also known as quantitative variables) are used to measure objective things that can be expressed in numeric terms such as absolute figures, such as the number of persons trained, disaggregated by sex, a percentage, a rate or a ratio

When designing indicators, the most important tasks are to logically link these to the intervention results and determine how the indicators will measure these results.

• 7Categorical variables can further be categorized as either nominal, dichotomous (binary) and ordinal.
• 8For more information, see Laerd Statistics, n.d. on the types of variables.

Any measure that is intended to be used should be relevant, credible, valid, reliable and cost-effective. The quality of indicators is determined by four main factors:

(a) Quality of the logical link between the indicator and what is being measured (such as the objective, outcome, output and/or impact of an intervention)

• What is being measured and why? What are the potential indicators?
• Why does/do the indicator(s) measure the objective, outcome, output and/or impact?
• How does the indicator measure the objective, outcome, output and/or impact?

(b) Quality of the measurement

• Are the indicators measuring what they are designed to measure (validity)?
• Do the indicators provide the same results when the measurements are repeated (reliability)?

(c) Quality of implementation

• Are the financial costs of measuring the indicators worth the information to be collected (cost-effectiveness)? 
• Are the data collection instruments the most appropriate given the established indicators for measuring the intervention objectives, outcomes, outputs and/or impact (relevancy)? Limited resources (time, personnel and money) can often prevent the use of the most appropriate data collection instruments.

(d) Quality of recognizing the measurement results and their interpretation

• To what extent are the measurement results and their interpretation accepted as a basis for decision-making by those involved (credibility)?

Table 4.1 provides a checklist for ensuring good quality measures.

Source: Adapted from Morra Imas and Rist, 2009, p. 293.

The values that a variable takes form a measurement scale, which is used to categorize and/or quantify indicators. They can be nominal, ordinal, interval or ratio scales. The levels of measurement used will determine the kind of data analysis techniques that can or cannot be used.

Levels of measurement

Nominal scales

Nominal scales consist of assigning unranked categories that represent more of quality than quantity. Any values that may be assigned to categories only represent a descriptive category (they have no inherent numerical value in terms of magnitude). The measurement from a nominal scale can help determine whether the units under observation are different but cannot identify the direction or size of this difference. A nominal scale is used for classification/grouping purposes.

The most important task of any indicator is to ensure the best possible allocation of the characteristics being measured to the measurement scale. This segregation of the characteristics “and their measurable statistical dispersion (variance) on the scale are the main insights gained because of the indicator (the variables)”.10

• 10Stockmann, A Practitioner Handbook on Evaluation, p. 204

When planning for data collection and thinking of the type of data that will be collected, it is important to assess the target audience from which the data will be collected. A crucial consideration that may influence decision-making is to determine the sample size and sampling strategy to select a representative sample of respondents, as this has budgetary implications.

While at times it may be feasible to include the entire population in the data collection process, at other times, this may not be necessary nor feasible due to time, resource and context-specific constraints, so a sample is selected.

Sampling is the process of selecting units from a population (that is, a sample) to describe or make inferences about that population (that is, estimate what the population is like based on the sample results).

Questions to ask when sampling

Who will be the respondents for data collection? How many people will data be collected from? Why is it better to collect data from group A rather than group B? How many responses will need to be collected to make the findings reliable, valid and representative of a larger population or group? How much data will need to be collected to enable an in-depth analysis? Will it be enough to speak to just a few beneficiaries about the results of a particular activity, or will the entire target population be needed? What is the ideal balance between information participants can provide and the number of participants required until the information needed is acquired?

Sampling applies to both qualitative and quantitative monitoring/evaluation methods. Whereas random sampling (also referred to as probability sampling) is often applied when primarily quantitative data collection tools are used for monitoring/evaluation purposes, non-random sampling (also referred to as non-probability or purposeful sampling) tends to be applied to monitoring/evaluation work that relies largely upon qualitative data11

Properly selecting a sample, ideally at random, can reduce the chances of introducing bias in the data, thereby enhancing the extent to which the gathered data reflects the status quo of an intervention. Bias is any process at any stage in the design, planning, implementation, analysis and reporting of data that produces results or conclusions that differ systematically from the truth.12 For more information on the types of bias, see Annex 4.3. Types of bias.

• 11 Adapted from Trochim, 2020a and Lærd Dissertation, n.d.
• 12 Adapted from Sackett, 1979.

Random sampling

Random sampling is an approach to sampling used when a large number of respondents is required and where the sample results are used to generalize about an entire target population. In other words, to ensure that the sample really represents the larger target population and that not only reflecting the views of a very small group within the sample, representative individuals are randomly chosen. Random sampling is an effective method to avoid sampling bias.

True random sampling requires a sampling frame, which is a list of the whole target population from which the sample can be selected. This is often difficult to apply. As a result, other random sampling techniques exist that do not require a full sampling frame (systematic, stratified and clustered random sampling).

Non-random/Purposeful sampling

Figure 4.7. Non-random sample
	Non-random sampling is used where: Large number of respondents are not required; The research is exploratory; Qualitative methods are used; Access is difficult; The population is highly dispersed.

Non-random/purposeful sampling is appropriate when there is a small “n” study, the research is exploratory, qualitative methods are used, access is difficult or the population is highly dispersed. For further information as to when it is appropriate to use non-random sampling, see Patton (2015) and Daniel (2012). The chosen sampling technique will depend on the information needs, the methodology (quantitative or qualitative) and the data collection tools that will be required.

Note: While the table shows the most common types of non-random/purposeful sampling, further types of non-random/purposeful sampling can be found in Patton, 2015.

Limitations of non-random/purposeful sampling

There are several limitations when using non-random/purposeful samples, especially convenience and snowball samples. First, generalizations to the entire target population cannot be made. Second, statistical tests for making inferences cannot be applied to quantitative data. Finally, non-random samples can be subject to various biases that are reduced when the sample is selected at random. If using a non-random sample, M&E practitioners should ask the following: “Is there something about this particular sample that might be different from the population as a whole?” If the answer is affirmative, the sample may lack representation from some groups in the population. Presenting demographic characteristics of the sample can provide insight as to how representative it is of the target population from which the sample was drawn.

Source: Adapted from Sheppard, 2020.

What you are trying to measure? What is the purpose? How you will measure it?

The size of the sample will be determined by what will be measured, for what purpose and how it will be measured. The size of the sample will also need to ensure, with the maximum level of confidence possible, that an observed change or difference between groups is the result of the intervention, rather than a product of chance. However, this may not always be the case for non-random/purposeful sampling.

Determining sample size: Random sampling

When a large number of respondents is required, the appropriate sample size is decided by considering the confidence level and the sampling error.

• For a study that requires a small number of participants, selecting small random samples can give highly misleading estimates of the target population. Therefore, non-random sampling is more appropriate.

Determining sample size: Non-random/purposeful sampling

For non-random/purposeful sampling, an indication of whether an adequate sample has been reached or not is data saturation. Once this point is reached, no more data needs to be collected. However, due to little guidance on how many interviews are needed to reach saturation, this can be sometimes difficult to identify.

The following questions can help determine how many people to include in the sample achieving both data saturation and credibility:

• Should all population segments be included in the sample?
• Should people with diverse perspectives be included in the sample?
• Should the findings be triangulated (see section on “Triangulation”)?

Surveys are a common technique for collecting data. Surveys can collect focused, targeted information about a sample taken from the target population for a project, programme or policy, especially data about perceptions, opinions and ideas. While surveys can also be used to measure intended behaviour, there is always room for interpretation, and any data gathered may be less “factual” as what people say they (intend to) do may not reflect what they in fact do in reality.

Generally, a survey is conducted with a relatively large sample that is randomly selected so that the results reflect the larger target population (see section on Sampling). The format of the survey can be structured or semi-structured, depending on the purpose of the data collection (see Table 4.8) and be implemented on a one-time basis (cross-sectional) or over a period of time (longitudinal).

Cross-sectional surveys are used to gather information on the target population at a single point in time, such as at the end of a project. This survey format can be used to determine the relationship between two factors, for example, the impact of a livelihoods project on the respondent’s level of knowledge for establishing an income-generating activity.

Longitudinal surveys gather data over a period of time, allowing for an analysis of changes in the target population over time, as well as the relationship between factors over time. There are different types of longitudinal surveys, such as panel and cohort studies.13

• 13Both panel and cohort studies are approaches to the design of longitudinal studies. Cohort studies follow people identified by specific characteristics in a defined time period, whereas panel studies aim to cover the whole population (Lugtig and Smith, 2019).

Surveys can be administered in different ways, such as in-person interviews, phone interviews or as paper or online questionnaires that require participants to write their answers.

For more information on how to design and implement a survey, see Annex 4.5. Survey design and implementation and Annex 4.6. Survey example

Interviews are a qualitative research technique used to shed light on subjectively lived experiences of, and viewpoints from, the respondents’ perspective on a given issue, or sets of issues, that are being monitored or evaluated for a given intervention. Interviews provide opportunities for mutual discovery, understanding, reflection and explanation. Interviews are of three types: (a) structured; (b) semi-structured; and (c) unstructured. Table 4.9 provides an overview of each interview approach, when to use it and some examples.

To know more about interviews, examples of interview structure and probing, see Annex 4.7. Interview structure and questions (examples provided throughout the annex) and Annex 4.8. Interview example.

A focus group is another qualitative research technique in the form of a planned group discussion among a limited number of people, with a moderator and if possible, note takers, as well as observers if also using observations.15 The purpose of a focus group is to attain diverse ideas and perceptions on a topic of interest in a relaxed, permissive environment that allows the expression of different points of view, with no pressure for consensus. Focus groups are also used to acquire an in-depth understanding about a topic or issue, which is generally not possible using a survey. For instance, a survey can tell you that 63 per cent of the population prefers activity Y, but a focus group can reveal the reasons behind this preference. Focus groups can also help check for social desirability bias, which is the tendency among survey respondents to answer what they think the enumerator wants to hear, rather than their actual opinions. For example, during the focus group discussion, one may discover that the actual preference of the participants is activity Z, not activity Y, as per their responses to the survey. However, focus groups provide less of an opportunity to generate detailed individual accounts on the topic or issue being explored. If this type of data is required, one should use interviews instead. If someone is answering too often, it is important to identify if this behaviour intimidates other participants and moderate the discussions inviting others to contribute. It is also important to understand who that person is, for instance, a political leader trying to impose answers to the group.

To know more about focus group discussions, see Annex 4.9. Preparing, conducting and moderating a focus group and Annex 4.10. IOM example of a focus group discussion guide.

• 15Usually, focus group discussions should not exceed 15 participants. For more participants, community group interview techniques may be used.

A case study is a qualitative data collection method that is used to examine real-life situations and if the findings of the case can illustrate aspects of the intervention being monitored and/or evaluated. It is a comprehensive examination of cases to obtain in-depth information, with the goal of understanding the operational dynamics, activities, outputs, outcomes and interactions of an intervention.

Case studies involve a detailed contextual analysis of a limited number of events or conditions and their relationships. It provides the basis for the application of ideas and extension of methods. Data collected using a case study can help understand a complex issue or object and add strength to what is already known.

A case study is useful to explore the factors that contribute to outputs and outcomes. However, this method of data collection may require considerable time and resources, and information obtained from case studies can be complex to analyse and extrapolate.

Observation is a research technique that M&E practitioners can use to better understand participants’ behaviour and the physical setting in which a project, programme or policy is being implemented. To observe means to watch individuals and their environments and notice their behaviours and interactions by using all five senses: seeing, touching, tasting, hearing and smelling.

Observations should be used on the following:

• Gathering data on individual behaviours or interactions between people and their environment;
• When there is a need to know about a physical setting;
• When data collection from interviews/surveys with individuals is not feasible.16

Observations can be conducted in a structured, semi-structured or unstructured approach.

 For more information, tips on and examples of observations, as well as planning and conducting observations, see Annex 4.11. Examples of observations and planning and conducting observations.

• 16CDC, 2018.

• 17The following information is adapted from IFAD, 2002

Once the M&E design has been identified and the method(s) and tools have been developed, the data collection can start. It is also recommended to organize a training with the data collection team(s) on the methodology. The training should cover in detail each data collection tool that will be used and include practical exercises of how to implement them.

Developing a data collection guide with clear instructions for the enumerators is a useful reference tool, both during the training and after, for the actual data collection; see the example provided below for an excerpt from a survey included in a data collection guide. Taking these steps will ensure that the collected data will be accurate with a minimum amount of error. In certain cases, however, conducting a full training is not feasible due to time and resource constraints, and having a data collection guide can be an important reference.

The data collected needs then to be transferred onto a computer application, such as Microsoft Word or Excel. Having the data in an electronic format will facilitate the data clean-up and data analysis. For quantitative data, the first step in data entry is to create the data file(s) to achieve a smooth transfer between a spreadsheet and a statistical programme package, such as SPSS and Stata for conducting statistical analyses.

For qualitative data, the first step in the data entry process is transferring all the interview, focus group and observation notes to a Word document for conducting content analysis using qualitative data programme packages, such as NVivo or MAXQDA.

Another component of the data entry is assigning each subject (or unit of analysis) a unique identifier (ID) (for example: 01, 02, 03 and so on), unless this is done directly during the data collection process. To do this, a separate file should be created that matches the identifying information for each subject (unit of analysis) with their unique ID. Assigning a unique identifier to each respondent ensures that the data cannot be traced back to them if the data is disclosed to other parties.

Once the data has been transferred from the medium used to record the information to a computer application (Word or Excel), it needs to be screened for errors. Following this, any errors need to be diagnosed and treated.

Data errors can occur at different stages of the design, implementation and analysis of data (see Figure 4.8):

• When designing the data collection instruments (such as improper sampling strategies, invalid measures, bias and others);
• When collecting or entering data;
• When transforming/extracting/transferring data;
• When exploring or analysing data;
• When submitting the draft report for peer review.26

* United Nations High Commissioner for Refugees (UNHCR), 2015.

• 26UNHCR, 2015.
• 27Ibid.

Once the suspect data has been identified, the next step is to review all the respondent’s answers to determine if the data makes sense given the context in which it was collected. Following this review, there are several possible diagnoses for each suspect data point identified:

• The data point is missing. Missing data can be a result of omitted answers by the respondents (no response), questions that are skipped over by the enumerator (erroneous entry or skip pattern) or the data entry agents, or there are dropouts (for longitudinal research).
• The data point is a true extreme value. True extreme values are answers that seem high but can be justified by other answers.
• The data point is a true normal value. True normal values are valid answers.
• The data point is an error. Errors can be either typos or inappropriate answers (questions asked were misunderstood by the respondents). Sometimes, errors can be rapidly identified when there are pre-defined cut-offs because the values are logically or biologically impossible. For example, the sample comprises only of respondents between the ages of 18 and 35; however, on the survey, a respondent is listed as being 80 years old, which is not possible.28

• 28 Ibid.

Once the problematic observations have been identified, these need to be treated before the data can be analysed. The following are some of the key approaches to dealing with data errors:

• Leave the data unchanged. This approach is the most conservative as it entails accepting the erroneous data as valid response(s) and making no changes. For large-n studies, leaving one erroneous response may not affect the analysis. However, for small-n studies, the decision of leaving the data unchanged may be more problematic.
• Correct the data, however without modification of the intention of or meaning given by the respondent.
• Delete the data. It is important to remember that leaving out data can make it seem as if the data is being “cherry-picked” to obtain the desired results. Alternatively, a binary variable can be created (1 = suspicious record; 0 = not so) and use this new variable as a record filter in Pivot tables or in-table filtering to understand the impact of potentially erroneous data in the final results.
• Re-measure the suspect or erroneous values, if time and resources permit.29

• 29 a b Ibid.

Missing values require attention because they cannot be simply ignored. The first step is to decide which blank cells need to be filled with zeros (because they represent negative observation; for example “no”, “not present” and “option not taken”) and which to leave blank (if using blanks to indicate missing or not applicable). Blank cells can also be replaced with missing value codes; for example, 96 (I don’t know), 97 (refused to answer), 98 (skip question/not applicable) and 99 (blank/missing).  

If the proportion of missing or incomplete cases is substantial for a category of cases, this will be a major M&E concern. Once a set of data is known to be missing, it is important to determine whether the missing data are random or whether they vary in a systematic fashion, and also the extent to which the problem exists. Random missing values may occur because the subject inadvertently did not answer some questions. The assessment may be overly complex and/or long, or the enumerator may be tired and/or not paying attention, thereby missing the question. Random missing values may also occur through data entry mistakes. If there are only a small number of missing values in the data set (typically, less than 5%), then it is extremely likely to be random. Non-random missing values may occur because the key informant purposefully did not answer some questions (confusing or sensitive question, no appropriate choices such as “no opinion” or “not applicable”). 

The default option for handling missing data is filtering and excluding them from the analysis:

(a) Listwise/casewise deletion: Cases that have missing values on the variable(s) under analysis are excluded. If only analysing one variable, then listwise deletion is simply analysing the existing data. If analysing multiple variables, then listwise deletion removes cases if there is a missing value on any of the variables. The disadvantage is a loss of data, because all data from cases who may have answered some of the questions, but not others (such as the missing data), are removed. 

(b) Pairwise deletion: All available data is included. Unlike listwise deletion, which removes cases (subjects) that have missing values on any of the variables under analysis, pairwise deletion only removes the specific missing values from the analysis (not the entire case). In other words, all available data is included. If conducting a correlation on multiple variables, this technique allows to conduct the bivariate correlation between all available data points and ignore only those missing values if they exist on some variables. In this case, pairwise deletion will result in different sample sizes for each correlation. Pairwise deletion is useful when the sample size is small or missing values are large, because there are not many values to begin with, so why omit even more with listwise deletion.

Note: Deletion means exclusion within a statistical procedure, not deletion (of variables or cases) from the data set. 

(c) Deletion of all cases with missing values: Only those cases with complete data are retained. This approach reduces the sample size of the data, resulting in a loss of power and increased error in estimation (wider confidence intervals). While this may not be a problem for large data sets, it is a big disadvantage for small ones. Results may also be biased if subjects with missing values are different from the subjects without missing values (that is, non-random) resulting in a non-representative sample. 

(d) Imputation (replace the missing values): All cases are preserved by replacing the missing data with a probable value based on other available information (such as the mean or median of the observations for the variable for which the value is missing). Once all missing values have been imputed, the data set can then be analysed using standard techniques for complete data. More sophisticated imputation methods, involving equations that attempt to predict the values of the missing data based on a number of variables for which data are available, exist. Each imputation method can result in biased estimates. Detailing the technicalities, appropriateness and validity of each technique goes beyond the scope of this document. Ultimately, choosing the right technique depends on the following: (i) how much data are missing (and why); (ii) patterns, randomness and distribution of missing values; and (iii) effects of the missing data and how the data will be used in the analysis. It is strongly recommended to refer to a statistician if M&E practitioners are faced with a small data set with large quantities of missing values.

In practice, for M&E purposes with few statistical resources, creating a copy of the variable and replacing missing values with the mean or median may often be enough and preferable to losing cases through deletion methods.

During the data clean-up process, certain variables may need to be recoded and new variables created to meet the analytic needs for the M&E exercise. Variables may be recoded in various scenarios, including the following: 

• Formatting: Date (day, month and year), pre-fixes to create better sorting in tables and rounding (in continuous variables).
• Syntax: Translation, language style and simplification.
• Recoding a categorical variable (such as ethnicity, occupation, an “other” category and spelling corrections).
• Recoding a continuous variable (such as age) into a categorical variable (such as age group).
• Combining the values of a variable into fewer categories (such as grouping all problems caused by access issues).
• Combining several variables to create a new variable (such as building an index based on a set of variables).
• Defining a condition based on certain cut-off values (such as “at risk” versus “at acute risk” population). 
• Changing a level of measurement (such as from interval to ordinal scale).
• A distinction is needed between values (conceptually). 

Categorical variables can be recoded in three ways:

(a) Collapse a categorical variable into fewer categories by combining categories that logically go together or eliminate categories that have small numbers of observations;

(b) Break a categorical variable up into several variables with fewer categories;

(c) Combine several categorical variables into fewer variables with more categories. 

Two good data management practices are transparency and the proper documentation of all the procedures followed, including the data cleaning process.

Documenting errors, changes and additions is essential to the following:

• Determining and maintaining data quality;
• Avoiding duplication of error checking by different data entry staff;
• Knowing what and by whom data quality checks have been carried out;
• Recovering data cleaning errors;
• Informing data users of the changes made to the last version of the data accessed.

To keep track of all the changes made to the data, a change log can be created. By keeping track of all the modifications made, it will be possible to roll back to the original values, when necessary. The following are some of the fields that are included in a change log:

• Table (if using multiple tables)
• Column, row
• Date changed
• Changed by
• Old value
• New value
• Comments

Qualitative data analysis is a process aimed at reducing and making sense of vast amounts of qualitative information – very often from multiple sources, such as focus group discussion notes, individual interview notes and observations – in order to deduce relevant themes and patterns that address the M&E questions posed. When analysing qualitative data, the focus is on the words spoken by the respondents, the context in which the data was collected, the consistency and contradictions of respondents’ views, the frequency and intensity of participants’ comments, their specificity and emerging themes and patterns. For example, as part of monitoring an ongoing project, it is decided to conduct 10 focus groups with a select number of beneficiaries. What should be done once all the discussion notes are collected? Should the data be analysed in an ad hoc or systematic fashion, that is, highlight the relevant information or code it?

Codes are words or short phrases that capture a “summative, salient, essence-capturing, and/or evocative attribute for […] language-based or visual data”.Saldana, 2009, p. 3. Coding is the process of labelling as “belonging to” or representing some type of phenomenon that can be a concept, belief, action, theme, cultural practice or relationship. Coding can be accomplished manually, using paper and highlighters, or by a computer in a Word document, an Excel spreadsheet or a qualitative data analysis software like NVivo.

To begin coding the data manually, gather the hard copies of all the data; mark up the text with pens, pencils, highlighters and markers; and finally cut, paste, hole-punch, pile and string together the data. It is preferable to leave wide margins and lots of white space for the markings.

As most M&E practitioners have access to word-processing and spreadsheet programs, these are quite popular in qualitative data analysis (see Figure 4.10 illustrating coding in a Word document).

For large amounts of data, such as 20 or more interview transcripts, a license for an existing qualitative data analysis software package like NVivo can be purchased. For an introduction to NVivo and an introductory guide to setting up and coding with this software, please see QSR International’s (2014) Getting Started Guide.

For guidelines on how to analyse qualitative data, see Annex 4.12. Steps for analysing qualitative data.

Once the quantitative data has been entered in a spreadsheet, it is ready to be used for creating information to answer the monitoring or evaluation questions posed. Statistics help transform quantitative data into useful information to help with decision-making, such as summarizing data and describing patterns, relationships and connections. Statistics can be descriptive or inferential. As its name already reveals, descriptive statistics give information that help describe data and help to summarize it. Other methods of descriptive statistics are graphical representations in form of histograms, pie charts and bar charts, to name a few. This provides a quick method of making comparisons between different sets of data and spotting smallest and largest values, trends or changes over a period of time. Inferential statistics use data drawn from a sample of the population to make generalizations about populations.

As most statistics used in M&E exercises are descriptive, the following discussion will provide tools and examples on how to calculate these types of statistics

As already mentioned in the measurement section, data is collected from units, which can be individuals, households, schools, communities and others. The different measurements, questions or pieces of information that are collected from/about these units are the variables. There are two types of variables, quantitative numerical (quantitative) and categorical. Whereas categorical variables are made up of a group of categories (such as sex, male/female), numerical variables are numbers such as the number of participants at a training.

Categorical data groups are all units in distinct categories, which can be summarized by determining how many times a category occurs. For example, the number of females in a community, which can be described as the frequency of females in the community. This information is presented using a frequency table. The frequency table shows how many individuals in the community fall into each category (male/ female). This can also then be represented as a percentage or proportion of the total.

Frequency tables can be used to present findings in a report or can be converted into a graph for a more visual presentation. A proportion describes the relative frequency of each category and is calculated by dividing each frequency by the total number.

Percentages are calculated by multiplying the proportion by 100. Proportions and percentages can be easier to understand and interpret than examining raw frequency data and are often added into a frequency table (see Table 4.15).

The centre and the spread of the data are two commonly used descriptive statistics. Whereas the centre describes a typical value, the spread describes the distance of a data point from the centre of the data.

The most common statistics used to describe the center are the mean (that is, the average) and the median. The median is the middle value in a data set; half the data are greater than the median and half are less. The mean is calculated by adding up all the values and then dividing by the total number of values.

Other statistics describing spread are the interquartile range and standard deviation.

• The interquartile range is the difference between the upper quartile and lower quartile of the data. A quarter (or 25%) of the data lie above the upper quartile and a quarter of the data lie below the lower quartile.
• The standard deviation shows the average difference between each individual data point (or age of youth in the above example) and the mean age. If all data points are close to the mean, then the standard deviation is low, showing that there is little difference between values. A large standard deviation shows that there is a larger spread of data.

For information on how to calculate descriptive statistics using Microsoft Excel, see Annex 4.13. Calculating descriptive statistics.

Triangulation is the process of comparing several different data sources and methods to corroborate findings and compensate for any weaknesses in the data by the strengths of other data.

Triangulation can enhance the validity and reliability of existing observations about a given issue. The ability to compare and contrast different findings and perspectives on the same situation and/or phenomenon is an effective way to find inconsistencies in data and identify areas for further investigation. When findings converge, this can lead to new, credible findings about an issue and create new ways of looking at it.

Although there are no fixed rules for analysing data for triangulation, there are several activities at the heart of the process:

• Critically assess the data. For example, prioritize those findings that are most relevant to the goal(s) of triangulation, identify ways the findings from different sources relate to one another and highlight any gaps in the data.
• Identify any trends and whether they are drawn from a single or from multiple data sources.
• Develop working hypotheses related to the goal(s) of data triangulation. For example, if the goal is to understand if certain behaviours are changing among beneficiaries and whether any changes can be linked directly to the intervention, hypotheses from the available data that are linked to this goal should be developed. Hypotheses can be in support of the goal; for example, a supportive hypothesis could be “providing psychosocial support has reduced signs of post-traumatic stress disorder (PTSD) symptoms among beneficiaries”.
• Confirm or refute hypotheses. This is a critical point in triangulation when new ideas, perspectives and explanations are likely to emerge. It is also a point when gaps in data are identified, which could lead to a search for additional data. If no additional data is available, a hypothesis may need to be modified or dropped. Any modified hypotheses should then be reconfirmed.
• Use the convergence of data supporting or not supporting the hypothesis to draw reasoned conclusions from the triangulation exercise. The conclusions should be linked as closely as possible to the stated goal(s) of triangulation. The key to this process is to make the strongest case for a hypothesis/ goal given the evidence. Questions that may be helpful to consider during the process include the following:
  • Which hypotheses are supported by the most rigourous data?
  • Which hypotheses are supported by independent sources?
  • Which hypotheses are supported by both quantitative and qualitative data?
  • Are there important biases or limitations in the available data?
  • Are there any other possible explanations not covered by the hypotheses?
  • How confident are you in the conclusion?
  • Is the conclusion actionable (that is, does it lead to a specific improvement in the response)?
• Carefully and thoroughly document conclusions before disseminating them.

Step 1: Identify the data visualization goal

Before M&E practitioners start designing any data visualization, the following questions should be asked:

• What is the data trying to communicate?
• How will it engage or persuade the audience to act upon the information being presented?
• What is the takeaway message the audience should be left with?

It is important to be clear about the goal(s) of presenting data visually in order to design it correctly. Defining the message is a crucial step in the process, and the graphic should reinforce who the organization or intervention is and what it does.

Step 2: Know the audience

Knowing the audience means asking what the audience already knows, what additional information they wish to have to learn and how much detail they require to understand the message being conveyed.

Step 3: Think about how to visualize the story

Once the data collected is cleaned and has been analysed, a more precise idea about what findings to present should emerge. Table 4.15 provides an overview of the key visualization techniques to use depending on what the data reveals.

Source: Carrington and Handley, 2017.

For additional information regarding each type of visualization, see Annex 4.14. Types of visualizations.

What is it?

Data governance represents the framework used by IOM to manage the organizational structures, policies, fundamentals and quality that will ensure access to accurate information. It establishes standards, accountabilities and responsibilities and ensures that migration data and information usage achieves maximum value to IOM, while managing the cost and quality of information handling. Data governance enforces the consistent, integrated and disciplined use of migration data by IOM.

How is it relevant to IOM work?

Data governance allows to view data as an asset in every IOM intervention and, most importantly, it is the foundation upon which all IOM initiatives can rest. Evidence-based programming only becomes a reality when data can prove what the problem is and how to solve it. This means being able to measure what is not known and knowing what is available and what is possible to work with.

The migration data life cycle throughout the whole project cycle includes planning and designing, capturing and developing, organizing, storing and protecting, using, monitoring and reviewing and eventually improving the data or disposing it.

Things to look out for:

(a) Data steward: At the intervention implementation phase, the data to collect has a clear data steward. If the intervention is a project implemented at the mission level, the chief of mission will be the data steward. If it includes more than one country, the data steward is most likely to be the regional director. If it is a global project, it should be ascertained which thematic area the project belongs to and, as such, the division head would be the data steward. Where the data is cross-cutting, it is likely that the data steward would be the department head.

(b) Roles and responsibilities: All project staff should be aware of their roles and responsibilities regarding the data and only have access to the data that they need to do their work.

(c) Data quality: Data needs to be accurate, valid, reliable, timely, relevant and complete.

(d) Data classification for security and privacy: The level of risk for the collection of data should be determined and classified accordingly, so that it can be stored with accurate access controls.

(e) Data processing including collection and use: Tools that allow to collect only the data needed for the purpose of its use should be developed.

• For a list of relevant resources, please refer to the text box IOM migration data governance in this chapter

Steps to conduct a desk review

Step 1: Identify all possible sources.

Step 2: Categorize documents.

Because some documents will be more pertinent than others, not all documents should be given equal weight or attention. To facilitate the desk review, available documents can be categorized into tiers:

• Tier I are documents specifically on the subject of the monitoring/evaluation exercise, such as situation, progress and monitoring reports and project proposals;
• Tier II are background documents such as media coverage and other agency reports;
• Tier III are non-project-related documents.

Step 3: Decide on the approach (Structured/Unstructured).

Often times, due to time constraints, monitoring and evaluation activities limit themselves to first-tier and partially second-tier documents for which an unstructured approach is suitable. To include a full second- and third-tiers documents in the desk review, a structured approach will be required, such as the following:

• Structured review form to record comments read through the documents;
• Rubric to rate parts of the documents, for example, using a four-point scale to divide documents into the following:
  • Those that do not address the topic at all.
  • Those that address the topic in a minor way.
  • Those that address the topic in some significant way.
  • Those focused principally on the topic.
• Indexing and searching documents for content analysis.

The accuracy of the collected data and conclusion drawn depend on the M&E practitioner and the respondents and how they address and comply with the different steps in the data collection, analysis and reporting processes. No study is ever entirely free from bias. Therefore, it is important to be transparent about any bias in the data collected in monitoring/evaluation reports. A statement relating to the potential biases and the steps that were taken to control such biases should be included in all monitoring/evaluation reports.

Respondent bias

Non-response bias: This bias occurs when individuals selected refuse to, or are unable to, participate in the survey. As a result, the data that is collected will differ in meaningful ways from the target population. To avoid this, M&E practitioners should ensure that the selected sample is representative of the target population or adjust the sample if this bias is becoming too important.

Acquiescence bias: Acquiescence bias occurs when a respondent tends to agree with and be positive about whatever the interviewer asks. To avoid this, questions that suggest such an answer should be revised to gage the respondent’s true point of view on the issue(s) of interest.32

Social desirability bias: This bias involves respondents answering questions in a way that they think will lead to being accepted and liked. To avoid this, indirect questioning may be used, which entails asking about what a third party thinks, feels and how they will behave. Such approaches allow respondents to project their own feelings onto others, while still providing honest, representative answers.33

Habituation bias: For habituation bias, the respondent provides the same answer to all those questions that are worded in similar ways. To avoid this, questions should be worded/reworded differently and have varying response options.33

Sponsor bias: As respondents generally know who is the organization funding the intervention, their feelings and opinions about that organization may bias their answers. For instance, respondents may present a dire situation in the hope of obtaining additional or future funding from the organization. This bias may be more difficult to address, but the same approach for acquiescence bias may be used.33

Attrition/Mortality bias: When respondents drop out of the study, the sample selected may no longer be representative of the target population. The sample needs to be adjusted.

Selection bias: Selection bias is the distortion of the data because of the way in which it was collected. Self-selection is one common form of selection bias whereby people volunteer to participate in the study.The source of bias here is that the participants may respond differently from those who do not volunteer to participate in the study.

Recall bias: This bias arises when the respondents have difficulties remembering certain information resulting in the collection of inaccurate information. One way to minimize this bias is to anchor questions to key events that respondents are familiar with that can help them recall the relevant information.

Evaluator/Researcher bias 

Confirmation bias: This type of bias occurs when an M&E practitioner forms a hypothesis or belief about the intervention being monitored/evaluated and uses respondents’ information to confirm that hypothesis or belief. Confirmation bias can also extend into the analysis stage, with evaluators/researchers tending to remember points that support their hypothesis and points that disprove other hypotheses. To minimize confirmation bias, M&E practitioners should regularly re-evaluate impressions of respondents and challenge pre-existing assumptions and hypotheses.

Question-order bias: This bias arises when one question influences respondents’ answers to subsequent questions. The words and ideas presented in questions prime respondents, thereby impacting their thoughts, feelings and attitudes on subsequent questions. While this type of bias is sometimes unavoidable, it can be reduced by asking general questions before specific, unaided before aided and positive before negative.

Leading questions and wording bias: This type of bias arises when M&E practitioners elaborate on a respondent’s answer in an effort to confirm a hypothesis, build rapport or overestimate their understanding of the respondent. To minimize this bias, practitioners should ask questions that use the respondents’ language and avoid summarizing what the respondents said in their own words.

Publication bias: This bias occurs when negative results are less likely to be submitted and/or published than positive ones. “Most evaluations are commissioned by agencies or donors with a vested interest in the result, so it is possible that the incentive structure tends toward more positive findings even when external consultants are hired to carry out the evaluation.”33

• 32Sarniak, 2015
• 33 a b c d Ibid.

Random sampling

Simple random sampling

Random samples are samples in which each unit in the target population for the monitoring/evaluation exercise has an equal chance of being selected. This approach is fair and reduces selection bias, which undermines the accuracy of the predictions being made about the target population (see section on “Bias”).

Ideally, a sample should be representative of the entire target population. In order to select a random sample, a sampling frame is required. Each unit is assigned a unique identification number and then using a random number table or generator, a certain number of units is randomly selected.

Example: At site X in country Y, there is a total of 1,536 IDPs, and 1,536 is a four-digit number, so every individual in the population is assigned a four-digit number beginning with 0001, 0002, 0003, 0004 and so on. Then, starting at any point in the random number table, choose successive four-digit numbers until 300 distinct numbers between 0001 and 1,536 are obtained or generate 300 random numbers between 0001 and 1,536 using a software such as Microsoft Excel (see example below).

How to select a random sample of 300 IDPs from a total population of 1,536 IDPs using Microsoft Excel

Step 1: Click on cell A1 and type RANDBETWEEN(0001,1536) and press Enter

Step 2: To generate, for example, a list of 300 random numbers, select cell A1, click on the lower right corner of cell A1 and drag it down to cell A300.

Systematic random sampling

Systematic random sampling is a technique that randomly selects a number near the beginning of the sampling frame list, skips several numbers, and selects another number, skips several more numbers, and selects the next name and so forth. The number of names skipped at each stage depends on the desired sample size.

How to select a systematic random sampling

Step 1: Estimate the number of units in the population (for example, 1,536 IDPs at site X).

Step 2: Determine the sample size (for example, 300 IDPs).

Step 3: Divide step 1 by step 2 (k=N/n) to get the skip number. Example: k = 1,536/300 = 5.12

Step 4: Select a subject at random from the first Kth number in the sampling frame (for example, fifth number).

Step 5: Select every Kth number listed after that one until the required sample is selected.

Because of the luck of the draw using simple random sampling, a good representation of subgroups in a population may not be obtained. Several other random sampling techniques exist to address this issue.

Stratified random sampling

Stratified random sampling, also referred to as proportional or quota random sampling, is a technique that divides the sampling frame in two or more strata (subpopulations) according to meaningful characteristics, such as type of migrant or gender from which participants are then randomly selected. Then, a simple random sample from each strata is taken. When using the same sampling fraction within the strata, proportionate stratified random sampling is conducted. When using different sampling fractions in the strata, disproportionate stratified random sampling is used. This technique is useful when the project, programme or policy is targeting several groups to compare.

Example: Among the people at the IDP site in country Y, how many are children, youth, young adults, adults and some elderly? If children, youth and elderly represent only a small proportion of the total IDP site population, a simple random sample may not include enough of them to allow for a meaningful analysis to be conducted.

How to select a stratified random sample

Step 1: Divide the population into the strata of interest. For example, of the 1,536 IDPs, there are 142 children (0–12), 157 youth (13–25), 413 young adults (26–34), 701 adults (35–60) and 123 elderly (60+).

Step 2: Select a simple random sample from each stratum. Example: 142/1,536 = .092 * 142 = 13.12

Select a simple random sample of 13 from the children stratum.

• Note:   The number of units that is selected from each stratum should be equivalent to the stratum’s proportion of the total population.

Simple, systematic and stratified random sampling techniques require a sampling frame, which is very difficult to have for individuals or families. When a sampling frame is not available or the units on the list are so widely dispersed that it would be too time-consuming and expensive to conduct a simple random sample, cluster sampling is a useful alternative.

Cluster random sampling

Cluster random sampling divides the population into several clusters and then a simple random sample is selected from the clusters. The units in the selected clusters constitute the sample. Unlike stratified random sampling, cluster random sampling uses the clusters to identify units not to compare them. A drawback of this approach is that the clusters may differ in important characteristics from the ones not included in the sample, thereby biasing the accuracy of the inferences made to the target population.

How to select a cluster random sample

Step 1: Identify the population of interest (for example, 1,536 IDPs at site X in country Y).

Step 2: Divide the population into a large number of clusters (there are 10 zones in the IDP camp of approximately 150 IDPs each, of which two are randomly sampled).

Step 3: Select a simple random sample of the clusters (for example, randomly sample 2 of the 10 zones, yielding a total sample of about 300).

Multistage random sampling

Multistage random sampling is a technique that combines two or more random sampling methods sequentially. The process usually begins by taking a cluster random sample, followed by a simple random sample or a stratified random sample. Multistage random sampling can also combine random and nonrandom sampling techniques.

Example: In country Z, there are 7 IDP sites. In order to assess the support being provided by IOM to the IDP populations in these locations, purposefully select 2 of the 7 sites according to a set of criteria. Within each of the two sites, there are 8 zones with about 60 IDPs each. Randomly select 2 zones from each of the two IDP sites, yielding a total sample of 240 IDPs.

Non-random sampling

Purposeful sampling

Purposeful sampling is when a sample is selected according to set of predetermined criteria that are believed to provide the data necessary for monitoring/evaluating the project, programme or policy under review. Unlike random sampling, this sampling technique is mainly used with a limited number of persons with the required information and limited time and resources to collect it. In emergency settings such as conflict-affected societies, this approach may also be more appropriate, as taking a random sample may face the risk of aggravating tensions.

Unlike random sampling, purposeful sampling is deliberately biased in order to select the most appropriate cases for the monitoring/evaluation questions posed. Thus, if this sampling technique is used, it is necessary to be transparent and rigourous when selecting a sample to control for and identify any potential bias in the data that need to be considered in the interpretation of results. For a discussion on how to select purposeful samples, see Buchanan-Smith et al.'s Evaluation of Humanitarian Action Guide (2016).

Snowball sampling

A snowball sample is another form of purposeful sampling that is used when it is not known who, what or how many perspectives to include. Begin with an interview and then ask the interviewee to identify other potential additional respondents to talk to. This process is continued until having reached a point of saturation. Saturation is the point in the data collection process where no new or relevant information emerges that addressed the monitoring/evaluation questions proposed. Snowball sampling is particularly useful when trying to reach populations that are inaccessible or difficult to locate.

When using purposive sampling, a variety of different perspectives should be included in the sample to ensure the credibility of the findings, that is, data source triangulation. For instance, accounting for roles, gender, ethnicity, religion, geographic location and other factors important to the problem being addressed by the project, programme or policy under review should be done to capture diverse perspectives in the sample. When this is not possible, it is required to be transparent about the perspectives included and those that were not in the monitoring/evaluation report.

Quota sampling

Quota sampling, also a purposeful sampling technique, consists of selecting a specific number of different types of units from a population non-randomly according to some fixed quota. The quota sampling can be proportional (to represent the major characteristics of the target population by sampling a proportional amount of each) or non-proportional (to specify the minimum number of sampled units in each category without considerations for having numbers that match the proportions in the target population). For example, in the IDP site, it is possible to select 150 women and 150 men to interview (n = 300, proportional) or 200 IDPs, 50 returnees and 50 refugees (n = 300, non-proportional).

Convenience sampling

Convenience samples are selected based on the availability or self-selection of participants (that is, volunteers) and/or the researcher’s convenience. While this technique is neither purposeful nor strategic, it remains widely used because it is inexpensive, simple and convenient. For instance, arriving in a project target location and interviewing the project staff available on the day of the visit and beneficiaries that are encountered while walking through different parts of the location. However, this sampling technique is also the least reliable of the non-random sampling approaches presented above, as those most available are over-represented, thereby underestimating the variability that exists within the target population. If this technique is used, using quotas can help ensure that the sample is more inclusive. For instance, ensure sampling an equal number of men and women, and every second interview can be conducted with a female.

Question format, types and response options

The format of survey questions can be closed-ended or open-ended. Many surveys tend to include a mixture of both. Closed-ended questions can be answered with a simple response or a selection of response options. Common formats for close-ended questions include response options that are:

• Dichotomous

  Example: Do your children go to school? □ Yes □ No

• Ordered

  Example: What is the frequency of food (or cash/vouchers) distribution at the site?

  □ Every day □ Twice a week □ Once a week □ Every two weeks □ Once a month

  □ Irregular □ Never □ Unknown □ No answer, why? _______________

• Unordered

  Example: Reason for displacement

  □ Conflict □ Flood □ Fire □ Drought □ Landslide □ Other

• Fill in the blank

  Example: How much money do you earn per day? ____

For more information and examples on the different formats for closed-ended questions, see David and Sutton’s chapter on “Survey design” in Social Research (2011).

In contrast, open-ended questions enable respondents to answer in their own words. For example, “Why did you specifically choose this destination country?”

There are six different types of questions that can be used to design a survey.37

(a) Behaviour/experience questions explore what the respondents do/aim to do, how they behave/act, what they experience/encounter, how they respond or just what happens or occurs in the context in which the project, programme or policy is being implemented. Example: Do you intend to move to a different location?

(b) Opinion/value questions explore the respondents’ thoughts, reactions, impressions, attitudes and outlook on the activities/issues being monitored/evaluated.

Example:

How would you describe your household’s access to public services, such as education, shelter, health and other services in the area in which you currently reside?

□ Excellent: We experience no problems whatsoever.

□ Good: Access is good but we experience minor delays.

□ Neutral.

□ Bad: We experience delays and problems.

□ Very bad: There are delays and denial of access from local community and authorities.

(d) Feeling questions explore the respondents’ emotions or emotional reactions about a particular experience, thought or issue.

Example:

Do you consider yourself locally integrated?

□ Yes □ Partially integrated □ No □ I do not know

(e) Knowledge questions inquire about the respondent’s knowledge about a particular issue. Example: How many members of your household suffer from a chronic illness?

(d) Sensory questions explore what the respondent sees, hears, touches, tastes and smells. Examples: Is there enough lighting in the site at night?

(e) Background/demographic questions elicit biographical or historical information from the respondents.

Example:

Sex: □ Male □ Female

When developing survey questions, it is important to avoid making them ambiguous, doubled barreled, leading and with double negatives. An ambiguous question is when the respondent can interpret the question in several different ways because it uses a vague term(s) or phrase. A double-barreled question is when you ask about two different issues at the same time. A question that contains “and” or “or” is an indication that the question may be double barreled. A leading question is when you suggest to the respondent a certain answer. Double negatives in a question introduces unnecessary confusion, thereby increasing the risk of gathering unreliable data. By avoiding these mistakes, it is ensured that the data collected is valid and reliable.

Example of ambiguous questions:

Example of double-barreled questions:

Example of leading questions:

Example of double negatives questions:

Sequencing questions

The order in which the survey questions are asked can affect the quality of the responses received, as well as whether or not the survey will be completed by the respondents. For example, if a survey is started with highly sensitive questions, the participants may provide inaccurate information, may skip the questions or drop out of the survey altogether. The following are a few tips on sequencing survey questions:

• Begin with questions that are of interest to the respondents.
• Ask questions about the present before questions about the past or future.
• Spread out fact-based questions throughout the survey, as these tend to disengage respondents.
• Begin with easier questions and place more difficult or sensitive questions near the end.
• Place personal/demographic questions at the end, as some respondents may feel uncomfortable continuing with the survey if they are asked certain background questions in the beginning.
• Group similar or related questions together.
• Ensure there is a logical flow to the questions.
• Ask the most important questions by two thirds of the way through.
• Prioritize questions, dropping those which are of low priority.

• 37Patton, 1987. There are other types/categories of questions that are used for interviews and that are further developed in Annex 4.7.

Layout 

Similar to the sequencing of survey questions, the layout of the survey is also important for gathering high-quality data (especially for self-administered surveys). The following are some of the main points to include in a survey:

• Title, date and respondent identification number;
• Contact and return information; 
• Introductory statement/cover letter that indicates the purpose of the survey and explains how it will be used;
• Limit the number of headings for the topics being covered;
• Keep questions simple, short and concise;
• Questions and response choices should be formatted and worded consistently;
• Each question should be numbered;
• Provide instructions on how to answer each question (for example, choose one or all options that apply, tick a box, circle the answer or write a short answer);
• Order responses from a lower level to a higher level, moving from left to right (for example, 1 – Not at all; 2 – Sometimes; 3 – Often; 4 – Always).

Introductory statement/cover letter 

The introductory statement/cover letter should be brief and easy to understand. The following are the main points that are generally included:

• Purpose and importance of the survey;
• Identification of the organization conducting the survey;
• Importance of the respondent’s participation;
• Explanation of why the respondent was selected to participate in the survey;
• Approximate time it will take to complete the survey;
• Assurance that the information will remain anonymous and confidential;
• Appreciation for the respondent’s time and effort;
• A person’s name and contact details for further enquiries;
• An offer for feedback of the survey results.

Building rapport 

At the beginning of a survey (or an interview, see section on Interviews), it is important to establish a good rapport with the respondent. Rapport is the ability to relate to the respondent in a way that creates a level of trust and understanding. How to build good rapport with your respondent?

(a) Make a good first impression.

Upon arriving at the meeting place to carry out the survey, do the best to make the respondent feel at ease. With a few well-chosen words, it is possible to put the respondent in the right frame of mind for participating in the survey. Begin with a smile and greeting such as “Good morning” or “Good afternoon” and then proceed with the introduction.

(b) Obtain the respondent’s consent.

(c) Answer any of the respondent’s questions frankly.

(d) Assure anonymity and confidentiality of responses.

If the respondent is hesitant about responding to a question or asks what the data will be used for, explain that the information collected will remain confidential, no individual names will be used for any purpose, and all information will be grouped together when writing any reports. Also, it is advisable not to mention other surveys conducted or show completed survey forms to other enumerators in front of a respondent or any other person.

(e) Always have a positive approach.

(f) Interview the respondent alone.

The presence of a third person during the survey can avoid obtaining frank, honest answers from the respondent. Therefore, it is very important that the survey be conducted privately and that all questions be answered by the respondent. If other people are present, explain to the respondent that some of the questions are private and ask to interview the person in the best place for talking alone. Sometimes, asking for privacy will make others more curious, so they will want to listen. Establishing privacy from the beginning will allow the respondent to be franker and more attentive to questions.

If it is impossible to get privacy, it is possible to carry out the interview with the other people present. In this case, separate the respondent from the others as much as possible.

(g) Use participants as experts.

Individuals are invited to participate in a study because they are viewed as possessing important knowledge required for monitoring/evaluating a specific project, programme or policy. In that case, it is advised to let participants know that the survey will learn from them. Expressing this to participants helps to establish a respectful appreciation for valuable contributions that they will make to the monitoring/evaluation exercise.
 

Reviewing, translating, pretesting and piloting the survey

Once the initial version of a survey is drafted, it is recommended to engage the key stakeholders and local experts in reviewing the draft closely and amend it based on the feedback received. This process may have to be repeated several times before the survey is ready for pretesting. If the survey will be conducted in the local language, the survey needs to be translated prior to the pretest. The translator(s) should be fluent in both languages and, to the extent possible, be familiar with the issues being covered in the survey. Once the survey is translated into the local language, a second translator should translate it back into the original language. This process ensures an accurate translation. Any gaps or misunderstanding need to be addressed before the survey can be pretested.

When conducting a pretest, it is important to test the survey among a group of people from diverse backgrounds relevant to the issues being monitored/evaluated. The goal of a pretest is to ensure that the survey is collecting the information it is aimed to collect. A good pretest should look at the survey at three levels:

• As a whole: Are all sections of the survey consistent? Are there any sections that ask the same question?
• Each section: If the survey has more than one section, does each section collect the intended information? Are all major activities/issues being monitored/evaluated accounted for? Are there any questions that are not relevant?
• Individual questions: Is the wording clear? Is the translation correct? Does the question allow ambiguous responses? Are there alternative interpretations?

One approach to assessing the survey on these three levels is to sit down with a small number of respondents as they complete the survey and ask them to reason aloud as they fill it out. This process, although time intensive, can yield important insights about how potential respondents will interpret the questions being asked. Any question that is misunderstood should be revised and tested again. This process may need to be repeated several times, especially if the survey has been translated into another language, which can result in gaps or misunderstandings regarding the accurate translation.

Once the survey has been finalized, it should be piloted with an appropriate sample of potential respondents to participate in the survey. This will provide a good indication of the validity and reliability of the questions in the survey. The pilot is also an opportunity to practice implementing the survey, which can help identify any potential challenges that may be encountered during the actual data collection.

Excerpts from the IOM South Sudan post-distribution monitoring of non-food items, Wau, September 2017

A mobile phone-based post-distribution monitoring household questionnaire was used to carry out household data collection and included visual observation and recording (photos) of the usage and state of distributed items. From 3,507 households, an initial sample size of 88 households was calculated using the sample size calculator referred to in the South Sudan Shelter-Non-food Item (S-NFI) Cluster Guidelines. Moreover, as per recommendation stipulated in the guidelines, 20 per cent of the initial sample size was added to account for any spoiled surveys or improper data entry. Therefore, 19 households (20% of the initial sample size) was added, totaling a target sample size of 105 households. Confidence was 96 per cent with 10 per cent margin of error. With three days to survey households, each enumerator collected data from seven households per day over three days. One enumerator interviewed an additional two households, resulting in a total sample size of 107 households.

Introduction

• Please introduce yourself and the purpose of the visit to the interviewee clearly.
• Please confirm the interviewee did receive the shelter and NFI items (blankets, mosquito nets, collapsible jerrycans and kangas) in the IOM distribution in May 2017.
• Please seek the consent of the interviewee before proceeding with the questionnaire, telling him/her that the interview will take about 30 minutes of their time.
• Please explain that you are not going to provide any additional items but that the information provided is only to help improve distributions in the future.
• Please try to keep the interview as confidential as possible to avoid bias. This may mean asking bystanders politely to move away, and/or finding a space where people are not able to overhear.
• Please confirm that the head of household and/or the individual who was registered and who collected the items at distribution time is the person you are interviewing.

The following are many possible examples and show only a short segment of a survey.

Did you receive NFI from IOM distribution during May 2017?

□ Yes □ No

1a. Select the State:

□ Abyei Administrative Area □ Central Equatoria □ Eastern Equatoria □ Jonglei

1b. Location of interview: ____________________________

2. Name of enumerator: ___________________________

3a. Name of respondent (Beneficiary name):___________________

3b. Please mark sex of the respondent: □ Male □ Female

3c. What is your household size?

□ 1 □ 2 □ 3 □ 4 □ 5 □ 6 □ 7 □ 8 □ 9 □ More than 10

4a. Did you feel you received the items in time to respond to your needs?

□ Yes, timely □ Delayed □ Too late

4b. What was the quality of the sleeping mat?

□ Good □ Average □ Not good

4c. Was the distribution well organized?

□ Excellent □ Good □ Averagely organized □ Poorly organized □ Badly organized

4d. What changes have you experienced since the items were distributed to you? (that is, protection from malaria, more comfortable sleep)

Parts of the interview and types of questions

Table A4.1. Overview of interview structure and types of questions

The opening

During the first few minutes of an interview, it is advisable to inform the participant about the interview by using an introductory statement similar to that for a survey. The length of the interview can be confirmed by saying something like this:

• The interview should last about an hour. Does that still work for you? While I will encourage you to elaborate on your answers to the questions that I will ask, there may be times when I redirect, so that we may be sure to cover all the issues within the hour.

The first questions should then focus on building rapport, helping the participants feel comfortable and knowledgeable. Therefore, questions should be non-intimidating, open-ended, easy and inviting, such as open-ended experience questions that will prompt the respondents to tell stories (for example, “What can you tell me about Project X?”) or factual questions about the issues being monitored/ evaluated (for example, “What basic services are available in your community?”).

The body

After the opening, you can begin to ask generative questions, which are non-directive, non-intimidating questions aimed at generating frameworks for talk. Tour questions ask the interviewee to share familiar descriptive knowledge or memories about an activity or event; for example, “Can you describe a typical day for me?” Tour questions can be followed with probes by asking for examples or timeline questions; for example, “What were the events that led up to you leaving your home?” Hypothetical questions ask interviewees to imagine their behaviours, actions, feelings or thoughts in certain situations; for example, “Imagine you receive X amount of money; how would you use it?” Behaviour and action questions can also be asked, as well as compare–contrast questions; for example, “How is your daily routine similar to or different from the daily routine you had before leaving your home?” Finally, questions can be about motives (of the interviewee and/or those of another person). Such questions include asking about feelings, actions or behaviours, worded in the format of “why” or “how” questions. After asking about past and present experiences, future prediction questions can be asked to obtain further related information, for example, “Where do you see yourself living in the near future?”

To obtain specific information, directive questions are used. The simplest type of directive question is the closed-ended question, which has predetermined single, dichotomous or multiple response options. There are also typology questions where respondents are asked to organize their knowledge into different types or categories; for example, “What types of recreational activities do you engage in on a regular basis?” Prompts can be used to encourage participants to articulate a range of categories or types. If interested to ask the participant to comment on the data collected thus far, member reflection questions can be used; for example, “On the basis of my fieldwork so far, it seems that one reason for … is … What do you think about my interpretation here?” If potentially threatening/intimidating questions need to be asked (such as personal or political), these should be left for the end of the interview as these may be less problematic if good rapport is already built with the participants.

The closing

Several types of questions exist for closing an interview. Catch-all questions can be used to capture and tie together loose ends or unfinished stories; for example, “What question did I not ask that you think I should have asked?” The end of an interview is also the time to ask identity-enhancing questions, which encourage the interviewee to leave the discussion feeling appreciated and an expert. For instance, “What did you feel was the most important thing we talked about today and why?” Answers to these questions can also guide future interviews. For demographic questions, when and how to ask them remains debated. Whereas some researchers and practitioners believe they should be asked at the beginning, in case the participant terminates the interview prematurely, others find they can interfere with building rapport. At the end of the interview, remember to thank the interviewees and reassure them of anonymity and confidentiality.

Formulating interview questions

Good quality interview questions should have the following characteristics:

• Simple and clear (no acronyms, abbreviations or jargon);
• Not double barreled;
• Promote open-ended and elaborate answers;
• Note: If it is decided to include yes/no questions, these should be followed by “Why?” or “In what ways?” or they should be reworded to encourage a more fine-grained answer (for example, “To what extent is…”).
• Straightforward (no double negatives), neutral and non-leading;
• Non-threatening to the interviewee;
• Accompanied by appropriate probes.

Probes

Probes are responsive questions asked to clarify what has been raised by the respondent. The aim is to obtain more clarity, detail or in-depth understanding from the respondent on the issue(s) being monitored/evaluated.

Examples of clarifying probes:

• Did I understand you correctly that…?
• When you say … what exactly do you mean?

Examples of detail probes:

• How did you deal with …?
• Can you tell me more about …?

Examples of analytical probes:

• How would you characterize …?
• What is/was important about …?

Examples of variations and counterfactual probes:

• Would you deal with X in the same way the next time?
• Some of the people I have talked to said that … What is your take on this?

Reviewing, translating, pretesting and piloting the interview

Similar to surveys, interviews too should be reviewed, translated (if necessary), pretested and piloted. For a review on how to proceed, refer to the subsection and annex on “Surveys”.

Tips for conducting interviews

• Let the interviewee know about the purpose and timing of the interview, the reason for being interviewed, how they were selected, how the data will be used (anonymity and confidentiality), how long the interview will take, whether they will receive a copy of the report, and that summary notes of the interview will be available if desired.
• Pick a time and location that is safe, easily accessible, quiet and free from distractions.
• Have a note-taker and/or record the interview, when feasible.
• If taking notes, make sure not to be distracted from the conversation:
  • Maintain eye contact as much as possible;
  • Write key words and phrases (not verbatim);
  • To capture a certain response, ask the interviewee to repeat to have sufficient time to capture it;
  • For important responses, ask the interviewee if their exact words can be used/quoted.
• Stick to the selected interview approach (structured, semi-structured, unstructured).
• Establish rapport and avoid asking sensitive questions until the interviewee appears comfortable.
• Give the interviewee sufficient time to respond.
• Be aware of cultural norms with eye contact, gender issues and asking direct questions.
• Write up the interview notes as soon as possible after the interview.

Evaluation: IOM Timor-Leste Counter Trafficking Project – Interview protocol for stakeholders and 
project

Relevant questions were drawn from the comprehensive list below, depending on the respondent’s role (such as project implementer or stakeholder) and areas of competency.

I have been requested to conduct an evaluation of the IOM project titled “Strengthening government and service provider responses to human trafficking in Timor-Leste: A capacity-building initiative”. The objectives of the evaluation are as follows: (a) measure the progress made by the project; (b) identify any challenges faced in the implementation of this project; and (c) identify lessons learned, best practices and potential areas for future project design and implementation. The evaluation focuses on the activities conducted under this project specifically, and not on IOM’s entire programme of activities in the country. The key respondents in this evaluation are IOM staff involved in project implementation, IOM’s implementing partner organizations, beneficiaries of the project’s activities, and government and civil society stakeholders. Individual responses will be kept confidential, and we will only share generalized findings and anonymous comments.

Thank you for your time and cooperation in this process!

Background information

(a) What is your title, role and your responsibilities in relation to the IOM project?

(b) How long have you been in this position?

Relevance

(c) To your knowledge, what does the IOM project aim to achieve?

(d) To what extent are the objectives of the programme still valid?

(e) To your knowledge, what are the main activities and outputs of the programme?

(f) Are the activities and outputs of the programme consistent with the outcomes and attainment of its objective?

(g) In your view, what are the assumptions linking the activities, outputs, outcomes and objectives?

Effectiveness

(h) Has your organization been involved in the implementation of any of the activities of this project?

(i) If yes, which ones?

(ii) If no, have you heard anything about the implementation of these activities? If yes, which ones?

(iii) What have been key achievements of these activities?

(iv) In your experience, what types of activities have been most of least successful? Why?

(v) What have been the key factors that have positively or negatively affected your work (or other’s work) in this project?

Sustainability

(i) What factors will contribute to or impede the continuation of the project’s achievements after the end of the project?

(j) To what extent have project outputs been institutionalized? For example, have any guidelines, terms of reference, policy documents, legislation or other items been adopted by national institutions?

Project progress (for implementers)

(k) What do you consider as the project’s key achievements to date?

(l) What have been key disappointments?

(m) To what extent is the project’s implementation on schedule? Why?

(n) What have been strengths and weakness in the implementation of the pilot initiative? (such as in terms of timeliness, managing risks, partners and resource allocation)

(o) What key lessons have been learned to date from implementing the pilot initiative? What recommendations or suggestions can you make with regard to the remaining implementation period of the project? Beyond the project?

Preparing for a focus group discussion

An average focus group discussion involves 6 to 8 people and a maximum of 15 and lasts between one to two hours. When selecting the participants for a focus group discussion, it is important that the group is homogeneous so that participants feel more comfortable expressing their opinions. To select homogenous groups of participants, consider, among other things, the following:

• Gender: Will men and women feel comfortable discussing this topic in a mixed-gender group? For example, women might feel uncomfortable discussing issue X if men are in the group.
• Age: Will age affect the way that people react to this topic? For example, a young person might feel uncomfortable talking about issue X if older people from his community are in the group.
• Hierarchy: Will people of different hierarchical positions be able to discuss this topic equally? For example, a resident from village A might feel uncomfortable discussing issue X if the local administrator is in the group.38

Other considerations can also include the participation of officials of the government among participants who may influence the answers or cultural differences that may affect the answers (mixing participation of indigenous people with other non-indigenous communities).

• 38Humans of Data, 2017

Once the participants are selected, obtaining their informed consent is required, verbally or on a written form. For the location, it is important to select a quiet and secure place, and easily accessible by all the participants. If not doing any survey and collecting demographic data is needed, a short form can be designed and administered to the participants at the end of the focus group discussion.

How to conduct a focus group discussion

Introduction

At the start of the focus group, the moderator should present the aim of the discussion and an overview of the topics that will be covered, assure participants that their responses will remain confidential and anonymous, and lay down the ground rules. The following is an example of an introduction to read at the outset of the focus group to all the participants:

"Thank you for agreeing to participate in this focus group discussion. We will discuss … Let me remind you that the information given will be treated confidentially and that you may refuse to answer any of the questions and end your participation in the discussion at any time.

For the group discussion to proceed smoothly and respectfully for all participants, the following ground rules should be respected at all times by everyone:

• One person speaks at a time.
• What is shared in the room stays in the room.
• Everyone has a chance to share their ideas, experience and opinions.
• There are no right or wrong answers.
• Everyone actively listens to and respects each other.

In addition to these ground rules, I would like to ask you if you have any other ground rules that you would like to add to the list.”

Warm-up

Before starting with the focus group questionnaire or topic guide, a warm-up time can be spent to make the participants feel comfortable around each other and safe to open up and share their ideas, experience and opinions. To do so, begin the discussion by asking the participants to introduce themselves (for example: “First, I’d like everyone to introduce themselves. Can you tell us your name?”) or conduct an icebreaker exercise (see Example box).

Questionnaire/Topic guide

Similar to interviews, focus groups will vary in the extent to which discussions are structured. If having a strong sense of the issues to be explored, consider developing a questionnaire (the number of questions will depend on the length and number of participants, but should not exceed 10 questions). If not having a thorough understanding of the issue(s) to be explored, consider developing a topic guide that will allow the group itself to shape the agenda and the flow of the discussion.

Wrap-up

• “What is one thing that you heard here that you thought was really important?”
• “Thank you for your time and participation. This has been a very successful discussion.”
• “Your opinions will be a valuable asset to… We hope you have found the discussion interesting.”

After the focus group discussion

If the discussion is recorded, having specified it at the beginning, transcribe the conversation as soon as possible, so the specific nuances of the dialogue are not lost.

Tips for moderating a focus group discussion

(a) Actively listen to the participants and remain neutral.

Active listening involves hearing what each person is saying and observing the body posture and facial gestures, which can provide insights regarding the appropriate ways to engage participants. It is important to remain as impartial as possible, even if having a strong opinion about something. If participants sense about you having an opinion, they may want to change their responses so that they will seem more socially desirable, rather than reflect what they truly believe or feel about a topic.

(b) Show the participants that they are listened to what they are saying.

Some of the common signs that indicate about paying attention include leaning forward slightly, looking directly at the participants while they are speaking and nodding at appropriate times. Frequently looking away and/or at a watch, or even worse yawning, can risk making participants feel that they are not listened to or boring, which can result in them becoming disengaged from the discussion.

(c) Use silence to encourage elaboration. 

Allowing silence at times during the discussion can encourage participants to elaborate upon what they are saying.

(d) Use probes.

When participants give incomplete or irrelevant answers, it is possible to probe for more developed, clearer responses. Some suggested probing techniques are as follows:

• Repeat the question;
• Pause for the answer;
• Repeat the reply;
• Ask when, what, where, which and how questions;
• Use neutral comments such as “Anything else?”

Introduction

Using probes for clarification also reinforces the impression that the participants have expert knowledge, based on their direct experiences with the topic(s) or issue(s) being monitored/evaluated. Good probes let the participants know that they are listened to and that it is worth to know more about what they have to say. It is important to avoid asking leading questions, as these can convey to participants an opinion and that it is not about learning from them as an unbiased listener. This type of questioning can also lead participants to answer questions in a socially desirable manner.

(e) Keep the participants talking.

To avoid possibly interrupting the participants if there is a need to follow-up with something, make a mental note of it and ask them about it once they have finished their thought.

(f) Keep track of time.

Some individuals have a tendency to talk at length about their ideas, experience and opinions. The moderator has to structure the focus group discussion in such a way that it is possible to elicit complete responses from the participants without rushing them, while still respecting the time constraints.

(g) Keep the discussion moving.

When the participants are sharing less pertinent or off-topic information, it is possible to politely move the focus group discussions forward, for instance by highlighting something that the respondent talks about that is relevant to another question or set of questions prepared for the discussion. Another way is to acknowledge that time together is waning, and there are some other aspects to have time to discuss, and for this reason, invite to move on.

(h) Reduce the pressure to conform to a dominant point of view.

When an idea is being adopted without any general discussion or disagreement, it is likely that group pressure to conform to a dominant viewpoint has occurred. To minimize this group dynamic, it is suggested to probe for alternative views; for example, “We have had an interesting discussion, but let’s explore other ideas or points of view. Has anyone had a different idea/experience/idea that they wish to share?”

(i) Note-taking.

Handwritten notes should be extensive and accurately reflect the content of the focus group discussion, as well as any salient observations of nonverbal behaviour, such as facial expressions, hand movements and group dynamics.

Focus group discussion guide: Excerpts from the IOM South Sudan post-distribution monitoring of nonfood items (Wau, September 2017). 

Two focus group discussion were carried out, one with women (11 participants) and one with men (9 participants). Community mobilizers assisted in gathering the participants. One local interpreter was present to facilitate discussions from Wau Arabic to English and vice versa.

Effectiveness

• What items did you receive?
• How did you hear about the registration taking place?

Protection

• Have you seen changes after receiving the shelter/NFIs? (Specify changes in their relationship with the community, family and others. Did jealousy arise? Did this have any implications on their sense of security? Were there thieves?)

Appropriateness (items) 

• Did the items you receive meet your needs? (Can you rank the items that you needed the most at the time of distribution?)

Coverage

• Were all those displaced registered?

Quality of intervention (services provided by the organization)

• Were the community, local authorities and beneficiaries involved in the criteria and needs identification of the NFI materials?

Accountability

• Was there a complaint desk?

Examples of observations

Structured observations are conducted using a recording sheet or checklist to list targeted observations.

Semi-structured observations can be conducted using observation guides

Unstructured observations are usually conducted when en route to or while walking around the observation site (see example in Table A4.4).

Planning for observations

When planning for observations, it is advised to consider the following steps:

Step 1:Determine what is being monitored/evaluated, that is, identify the indicators being monitored, evaluation criteria and questions being explored.

Step 2:Determine the specific items for which to collect data on and how to collect the information needed (recording sheets and checklists, observation guides and/or field notes).

Step 3: Select the sites for conducting the observations.

Step 4: Select and train the observers.

Step 5: Pilot observation data collection procedure(s).

Step 6: Schedule observations appropriately to ensure observing the components of the activity that will answer the evaluation questions.39

The observation data collection procedure should be piloted before it is used for monitoring/evaluation. To do this, a minimum of two observers should go to the same location and complete the observation sheet. Once completed, the sheets should be compared. If there are large differences in terms of the observations made, the observers may require more training and clarification. If there is little difference, the procedure can be used for monitoring/evaluation.

Tips for conducting observations

Do

a) Enter the observation process without preconceived notions and fixed expectations.

(b) Note observations made and information volunteered that are related to subjects beyond formal assessment concerns.

(c) Record information that is contradictory or surprising to expectations.

(d) Stay focused to make useful comparisons.

(e) Be active and curious in the observation process, which is about seeing, hearing, smelling, tasting, feeling and touching.

(f) Be aware of what was not seen, such as the absence of people, services and infrastructure.

(g) Respect the local culture.

Do not 

(a) Begin the observation process with a set of expectations or seek to record data primarily to prove a pre-existing hypothesis.

(b) Rely on remembering information. Record observations on the observation sheet.

(c) Focus solely on misery and destitution. Be aware of capacities, opportunities and social capital within the affected community.

(d) Be intrusive. Take steps to be as sensitive and respectful as possible.

(e) Take a photograph without asking prior permission.

• 39CDC, 2018

Step 1: Get to know the data 

Before beginning to analyse the data, M&E practitioners need to familiarize themselves with them. This process requires reading and rereading the notes taken (the data) in their entirety. As they go through the data, it is important to take notes of any thoughts that come to mind and summarize each transcript and piece of data that will be analysed. The goal at this stage is to absorb and think about the data that has been collected, jotting down reflections and hunches but reserving judgement. Some open-ended questions that can be asked include: “What is happening here?” or “What strikes you?”40

Step 2: Initial round of coding

Once the content of the data is known, practitioners can begin coding the material to condense the information into key themes and topics that can help answer the M&E questions posed.

There are two main approaches to coding qualitative data. The first approach consists of creating a framework that reflects the monitoring or evaluation aims and objectives, which is then used to assess the data gathered. This is a deductive approach, as the concepts and issues of interest are first identified, which allows one to focus on particular answers of respondents and disregard the rest.

The initial round of coding begins with an examination of the data and assignment of words or phrases that capture their essence. Those who use a manual approach could write the code in the margin, and those who use a word-processing software could type the code in the Comment function or in another column.

The codes assigned in this first round of coding are usually, but not always, also first-level codes. Firstlevel codes focus on “what” is present in the data. They are descriptive, showing the basic activities and processes in the data such as LAUGHING), thereby requiring little interpretation, which is done in the second round of coding.

Throughout the coding process, it is important to continuously compare the data applicable to each code and modify the code definitions to fit new data (or the codes and create a new code). Both lumping data into large bins and fracturing them into smaller slices have advantages and disadvantages.41 Those who first fracture the data into small pieces, each with its own code, usually connect these bits into larger categories during later coding cycles. In contrast, those who lump first usually make finer distinctions later.

What data should be coded first? Many qualitative experts suggest first coding the data that are typical or interesting in some way, and then moving on to contrastive data. The initial data texts coded will influence the resulting coding scheme, so it is advised to choose texts in these early stages that represent a range of the data available. Also, an iterative approach does not require that the entire corpus of data be put through a fractured and detailed primary coding cycle. After having read through all the data a few times, and having conducted line-by-line initial coding on a portion, it is possible to consider some focusing activities.

As practitioners engage in the initial round of coding, it is helpful to create a list of codes and a brief definition or representative example of each, especially if the codes are not self-explanatory. As the coding becomes more focused, it is wise to develop a systematic codebook – a data display that lists key codes, definitions and examples that are going to be used in the analysis. Codebooks are like “legends” for the data, helping to meet the challenge of going through pages of transcripts, highlighting and scrawling.
 

• 40Tracy, 2013
• 41 Bazeley and Jackson, 2013.

In addition to creating a codebook, it is important to frequently return to the monitoring or evaluation questions posed. As most M&E practitioners face various time and resource constrains, many pursue analysis directions that align not only with themes emerging in the initial coding, but also with ones that mesh well with monitoring/evaluation goals, experience and deadlines.

Throughout the analysis, revisiting research questions and other sensitizing concepts helps to ensure they are still relevant and interesting. Original interests are merely points of departure, and other more salient issues may emerge in the data analysis.

Step 3: Second round of coding 

The second round of coding is about beginning to critically examine the codes identified in the initial round of coding and organize, synthesize and categorize them into interpretive concepts. This second round aims to explain, theorize and synthesize the codes from the first round by identifying patterns, rules or cause–effect progressions and making interpretations.

For instance, if codes that continually reappear in the data are identified, M&E practitioners may decide to link them together in a specific way that responds to the monitoring/evaluation question posed.

Accordingly, at this point, a better understanding of which data will be most important for the analysis will emerge. Certain data, even if they are already collected, may only tangentially relate to the questions being explored, and therefore, they will not be included in the analysis at hand. It is also at this point that M&E practitioners will see whether additional data needs to be collected to flesh out an emerging code or explanation of what is being observed in the data collected. One way to identify whether additional data is required is to ask this question: “Does the emerging analysis address the monitoring/evaluation question posed in an interesting and significant way?” If not, this may suggest the need for more data. It might also suggest the need for additional synthesizing activities.

Throughout the coding process, it is important to record the emerging thoughts and ideas systematically. First, create a document that records all the analysis activities chronologically (date and discussion of what was accomplished in terms of analysis). Second, write analytic memos, both as a part of the analysis process and as an analysis outcome. Analytic memos are sites of conversation with oneself about our data. Analytic memos help M&E practitioners figure out the fundamental stories in the data and serve as a key intermediary step between coding and writing a draft of the analysis. Although they can take many forms, analytic memos are often characterized by one or more of the following features:

• Define the code as carefully as possible;
• Explicate its properties;
• Provide examples of raw data that illustrate the code;
• Specify conditions under which it arises, is maintained and changes;
• Describe its consequences;
• Show how it relates to other codes;
• Develop hypotheses about the code.44

• Analytic memos are very helpful for thinking through how codes relate to each other.

Practitioners should also play devil’s advocate with themselves through the process of negative case analysis. Such a practice asks them to actively seek out deviant data that do not appear to support the emerging hypothesis, and then revise arguments so that they better fit all the emerging data. Negative-case analysis discourages the practice of cherry-picking data examples that only fit early explanations and ignoring discrepant evidence. As such, negative case analysis helps to ensure the fidelity and credibility of emerging explanations.

In addition to the analytic memos, M&E practitioners should create a loose analysis outline that notes the questions posed and the potential ways the emerging codes are attending to them.

Once the data is coded, it is time to abstract themes from the codes. At this stage, practitioners must review the codes and group them together to represent common, salient and significant themes. A useful way of doing this is to write the code headings on small pieces of paper and spread them out on a table; this process will give an overview of the various codes and also allow them to be moved around and clustered together into themes. Look for underlying patterns and structures – including differences between types of respondents (such as adults versus children and men versus women) if analysed together. Label these clusters of codes (and perhaps even single codes) with a more interpretative and “basic theme”. Take a new piece of paper, write the basic theme label and place it next to the cluster of codes. In the final step, examine the basic themes and cluster them together into higher order and more interpretative “organizing themes”.

• 44Tracy, 2013.

Calculating these descriptive statistics can be easily done with Microsoft Excel. To do this, install the Data Analysis Tool pack. Open Excel, go to FILE | OPTION| ADD INS and add the Analysis Tool. Once this is done, Data Analysis should appear on the far right of the tool bar. The advantage of the data analysis tool is that it can do several things at once. If a quick overview of the data is needed, it will provide a list of descriptive statistics that explain your data.

There are multiple ways and tool for visualization of data, and here are some of the most common samples and types.

Summary table

Summary tables are useful for displaying data in simple, digestible ways. The use of a summary table allows the reader to assess data and note significant values or relationships. Figure A4.1 depicts a summary table of the types of sites as hosting IDPs displaced due to the ongoing conflict in South Sudan.

Facts and figures 

Infographics are a useful way to draw attention to important facts and figures in the data. Icons and images, as well as different font sizes, can be used to present the data values in an appealing way that is easily digestible (see Figure A4.2 for an example).

Comparison, rank and distribution  

Bar charts and heat maps can be used to compare, rank and show the distribution of data values. Bar charts use a horizontal (X) axis and a vertical (Y) axis to plot categorical data or longitudinal data. Bar charts compare or rank variables by grouping data by bars. The lengths of the bars are proportional to the values the group represents. Bar charts can be plotted vertically or horizontally. In the vertical column chart, the categories being compared are on the horizontal axis, and on the horizontal bar chart (see Figure A4.3), the countries being compared are on the vertical axis. Bar charts are useful for ranking categorical data by examining how two or more values or groups compare to each other in relative magnitude at a given point in time.

Histograms are a graphical representation of the distribution and frequency of numerical data. They show how often each different value occurs in a quantitative, continuous data set. Histograms group data into bins or ranges to show the distribution and frequency of each value.

Figure A4.4 shows the proportion of survey respondents reporting exploitation for any type of the Central Mediterranean route by age group. Here, the age of the respondents are grouped into “bins”, rather than displaying each individual age.

Another approach to visualize the distribution of the data is to use heat maps. Figure A4.5 shows the concentration of returnees from Pakistan and Islamic Republic of Iran to three provinces in Afghanistan (Laghman, Nangarhar and Kunar).

Note:   This map is for illustration purposes only. The boundaries and names shown and the designations used on this map do not imply official endorsement or acceptance by the International Organization for Migration

Proportion or part-to-whole

Pie charts or donuts are circular charts divided into slices, with the size of each slice showing the relative value, typically out of 100 per cent. Pie charts and donuts are useful for providing an overview of categories at a single point in time (see Figures A4.6 and A4.7).  

If deciding to use a pie chart, make sure to limit the number of pie slices to five, as too many risk distracting the reader from the main point. Also, it can happen that the value of some of the slices are relatively the same, which makes it hard to compare their contribution to relate to one another. In this case, a horizontal bar chart may be more appropriate if the values of the slices are relatively the same to clearly see the difference between them.

Change over time

Bar charts can also be used to represent longitudinal data repeated over time to help identify temporal trends and patterns (see Figure A4.8). Similarly, line charts are another great way for displaying trends (see Figure A4.9).

Relationships and trends

Scatter charts are commonly used to show the relationship among the variables where both the horizontal and vertical axes are value axes, not categorical axes. For instance, the United Nations has released a new study that finds a causal relation between long-lasting droughts in El Salvador, Guatemala and Honduras and the increase in irregular migration from these countries to the United States. “Members of families affected by the drought are 1.5 per cent more likely to emigrate than similar households elsewhere. Although this is a low value, the significance lies in the fact that the correlation between drought occurrence and emigration is positive and the probability of emigrating is higher than that of families who are not from the Dry Corridor”.45 The scatter plot can be used to illustrate this positive relationship (as the length of drought increase, irregular migration increases). In Figure A4.10, the scatter plot demonstrates a positive relationship between the number of units sold by product family and revenue. The more units sold, the greater the revenue.

A bubble chart is a variation of a scatter chart in which the data points are replaced with bubbles, and an additional dimension of the data is represented in the size of the bubbles. Just like a scatter chart, a bubble chart does not use a category axis; both horizontal and vertical axes are value axes.

• 45 World Food Programme, 2017, p. 16.

Text analysis

Text analysis refers to various processes by which qualitative data can be modified so that they can be organized and described in a clear and intuitive manner. To summarize text gathered such as focus group discussion notes, basic text summaries and analyses can be conducted. Some of the most common ways of achieving this is using word frequencies (lists of words and their frequencies) and word clouds (see Figure A4.11).

Table A4.5. Evaluating data visuals checklist

 For more information and examples of the points items listed in the checklist for evaluating the data visuals (Figure A4.2), see chapter 13: Evaluating real data visualizations in Data Visualization for Dummies by Yuk and Diamond (2014).