lecture5.pptx

Data Sampling, Collection, and
Testing

Session
Contents
1. Unit of measurement
2. Types of data collection approaches
(structured)
3. Types of data collection approaches (semi-
structured)
4. Types of data collection approaches (mixed
methods)
5. Frequently Asked Questions (FAQs)

What is it?
 The unit that will be used to record,
measure and analyse observations/
information collected
 Examples?
 Individual
 Family
 Household
 Community/ group
 Town/ village
 Facility
 Cow

Remember…
 Unit will impact the time, resources needed to collect and analyse information
 Unit will define the depth of information possible and scope of analysis
Depth
of
information
Location level
Household level
Individual level
Community/Group level
Time / Cost / Access

Data
collection
approaches:
Structured

1. The structured survey approach
 Information collected through an interview, a discussion, a conversation
 Using structured, close-ended data collection tools
 Collection of quantifiable information
 Cross-sectional or longitudinal
Types of data collection methods?
Household (HH) survey – collecting data at HH level, to understand experiences and characteristics of HHs within population of interest
Individual survey – collecting data at individual level, to help understand situation and characteristics of individuals within the population of
interest  can include some HH level indicators if needed
Key informant interview – collecting data at community. location or group level from a key informant (KIs) i.e. an individual whose informal/
formal position gives him specific knowledge about other people, processes, or events that is more extensive, detailed, or privileged than
other individuals in their group/ community/ location
Group discussion – collecting data at community, location or group level from a group of representatives e.g. KIs

1. The structured survey approach- Applicability
When should you use this approach?
 To measure prevalence  provide a quantifiable, numeric description of
the trends, behaviours, experiences, attitudes or opinions of a population
 To generalize findings to a wider population  probability sample 
statistically representative information
 Need prevalence data, understanding of scale of crisis but probability
sampling not possible  non-probability sample  indicative information
Types of research cycles this approach is commonly used for?
Multi-sector needs assessments
In-depth thematic needs assessments e.g. WASH Cluster needs assessment
Longitudinal studies
Third party monitoring (impact evaluation, outcome monitoring, post-
distribution monitoring, etc.)

2. The structured experimental approach
What is it?
 Similar to survey approach
 But relies on experimental survey design 
control vs. treatment group
Household (HH) survey – collecting data at HH level, to understand experiences and
characteristics of HHs within population of interest
Individual survey – collecting data at individual level, to help understand situation and
characteristics of individuals within the population of interest  can include some HH level
indicators if needed

2. The structured experimental approach - Applicability
To measure prevalence and evaluate the outcomes or impact of a medium to large-scale intervention on
the population of interest
 Generalize findings to a wider population  probability sample  statistically representative
information
 Outcome monitoring
 Impact evaluations
 Etc.

3. The structured observation approach (Description)
What is it?
 Information collected through observation rather than
conversation
 Using structured, close-ended checklists to collect
quantifiable information
 Looking for specific object, behaviour or event against a
checklist e.g. Household using soap? Damage to health
center? Students participating in classroom?
 Can be used as part of experimental approach
 Participant observation – researcher participates in
context (e.g. anthropologists)
 Direct observation – researchers observes context (e.g.
psychologists or clinical research)


3. The structured observation approach - Applicability
 Serves similar purpose as survey approach
 Depends on research objectives  observation vs.
conversation?
Could be same as survey approach
Could be same as experimental approach

Data collection
approaches:
Semi-structured

4. The semi-structured discussion approach
What is it?
 Information collected through detailed, narrative interviews, group discussions
 Using semi-structured (NOT UNSTRUCTURED) data collection tools 
open-ended questions, probes
 Purposefully selected participants
Individual interview – collecting data at individual level, to help understand situation and characteristics of individuals within the population
of interest  can include some HH level indicators if needed
Key informant interview – collecting data at community. location or group level from a key informant (KIs) i.e. an individual whose informal/
formal position gives him specific knowledge about other people, processes, or events that is more extensive, detailed, or privileged than
other individuals in their group/ community/ location
Group discussion – collecting data at community, location or group level from a group of representatives e.g. Kis
Focus group discussion – bringing together people from similar backgrounds or experiences to discuss a specific topic of interest; data
collected at community, location or group level

4. The semi-structured discussion approach - Applicability
 To gather detailed insights about the
experiences, perspectives of specific population
group or location
 To provide a qualitative description of the
experiences, trends, attitudes or opinions of a
population
Types of research cycles this approach is
commonly used for?
 In-depth assessments where there is limited
prior understanding of a situation e.g. access to
cash among refugees & migrants in Libya
 Participatory mapping exercises (mapping FGDs
or KI interviews)

‘Most Significant Change’ data collection technique
 A very specific type of participatory,
discussion-based data collection
method used for monitoring &
evaluation
 Invites participants (through KI
interviews, individual interviews or
FGDs) to explain the most significant
changes brought about in their lives
by a project over a given period of time,
in key domains of change
 Useful for third party monitoring or
impact evaluation research cycles

5. The semi-structured observation approach
 Similar to structured observation approach
 But two key differences:
Structured observation Semi-structured observation
1. Differences in data
collection methods
Information collected using a
structured set of questions,
usually to identify specific object,
behaviour or event against a
checklist
Information collected based on a
short set of open-ended
questions for observations e.g.
movement patterns of refugees
in and out of camps during a
sustained period of time
2. Differences in purpose Provide a quantifiable, numeric
description of the trends,
behaviours, experiences, etc. of
a population
Gather detailed insights about
the behaviours, experiences of
a specific population group or
location, and to understand, by
observation, how things are
done and what issues exist

Data collection
approaches:
Mixed Methods

6. Sequential mixed methods data collection
 Method used to sequentially elaborate or expand on the findings of one type of
research method with another
• Identify coping
strategies
Qualitative
• Measure
prevalence of
identified coping
strategies
Quantitative
1. Exploratory sequential approach
• Measure
prevalence of
known coping
strategies
Quantitative
• Understand and
contextualize
observed trends
in prevalence
Qualitative
2. Explanatory sequential approach
• Identify coping
strategies
Qualitative
• Measure
prevalence of
identified coping
strategies
Quantitative
• Understand and
contextualize
observed trends
in prevalence
Qualitative
3. The “ideal” sequential approach

7. Concurrent mixed methods approach
 Method used to merge or converge the findings from different research methods collected at the
same time
 Alternative to sequential approach if time constraints  sequential better practice if time and
resources allow
 Concurrent mixed methods serves two key purposes:

Triangulation strategy
Convergence
of information
collected?
Divergence of
information
collected?
Embedded strategy
Primary
method:
quant
(What?
Where?)
Secondary
method:
qual (How?
Why?)
Key findings &
conclusions

Case study data collection technique
 Using a combination of different data
collection methods to zoom in to a
specific issue, area or group
 A component within a research
cycle, not a research cycle by itself
 Useful to collect detailed information
on an event, activity, process, group
e.g. zoom in to one specific type of
intervention in an area within a larger
DFID-funded humanitarian programme

Frequently
Asked
Questions
(FAQs)

FAQs (1)
 What is the difference between a key informant interview and an individual interview?
Isn’t the key informant also technically an individual?
 The differences lies in the unit of measurement  individual experiences (individual interview) vs.
community/ village/ institution experiences (KI interview)
 For semi-structured data collection, when is it recommended to use FGDs over KI or
individual interviews?
 This depends on two things
 Research objectives and type of information needed e.g. Variety of opinions and
experiences useful? Specific information needed from an expert? Topics sensitive to discuss
in group setting?
 Logistical constraints e.g. Large number of individuals to be reached within a short
timeframe?

FAQs (2)
 Is it possible to have two different units of measurement in the same questionnaire?
 Ideally, should be avoided, but there are some exceptions:
 Individual information within a household survey (e.g. child attendance roster)
 Household information within an individual survey (e.g. household size or income indicators)
 Individual information within a village/ community/ location level interview (e.g. KI’s displacement status and experiences,
if KI also part of the affected population)
 Household information within a village/ community/ location level interview (e.g. KI estimates # or % of households
affected by a specific situation in a village)
 What if my population of interest includes minors (i.e. individuals <18 years of age)? Can I collect data
from minors?
 Only if absolutely necessary to meet objectives of the research
 Only if required information cannot be collected from adult respondents e.g. parents or caregivers
 Ideally, only from respondents >15 years
 Only if the required protocols are being followed
 Will de discussed later in this training 

What methods
to use if you
don’t have
access to the
population of
interest?

What is remote data collection?
Remote data collection is a means of gathering data without a
physical presence in the data collection location and without
direct, in-person contact with the population of interest
When is it useful?
When it is not possible to conduct in-person visits to the
locations / populations of interest because of reasons such as:
 Disease outbreak (e.g. COVID-19)
 Time or resource constraints (e.g. not enough budget to hire
enumerators to cover all areas for face to face interviews)
 Access constraints due to:
 Security concerns
 COVID-19 travel restrictions
 Physical access barriers such as lack of infrastructure
 Severe weather conditions which limits travel
possibilities, etc.
 Etc.

Pros and cons of remote data collection
Pros Cons
Planning efficiency
More time and resource efficient; if
necessary logistics already in place,
could be fairly straightforward to
deploy
Challenging and time consuming to
set up correctly (e.g. identifying
respondents, organizing necessary
logistics, etc.), difficult to apply
stratification in sampling; challenging
to monitor progress
Implementation efficiency
Easier to implement even with
limited time, access and resources
(assuming planning and design is
done robustly)
Higher likelihood of low response
rates; limited means of verifying
responses/ data quality assurance;
more challenging to build trust with the
respondents; difficult to deploy long
or complicated questionnaires
Coverage
Ensures maximum possible
coverage of areas and population of
interest despite access constraints
Difficult to have the “full picture” as it
could introduce potential sampling
biases (e.g. based on phone network
coverage) and results in exclusions/
oversight of certain population
groups or areas

Some types of remote data collection methods (1)
1. Phone-based (individual, household, community level)
 Most relevant for: needs assessments, post distribution monitoring (PDMs),
humanitarian situation monitoring (HSM)
 Representative sampling could be possible
2. REACH “Area of Knowledge” methodology (face-to-
face data collection in alternate location)
 Most relevant for: community-level needs assessments or HSM
 Representative sampling not relevant (requires identifying the
respondent most likely to have the required knowledge)
3. Internet-based data collection
 Tools include: social media, web-based surveys, online discussion
platforms, chatbots (WFP mVAM), etc.
 Most relevant for: community-level needs assessments or HSM (KI
interviews or group discussions), PDMs (individual perception surveys)
 Representative sampling could be possible (but extremely difficult to
implement e.g. would need email address database and usually low
response rates)

Some types of remote data collection methods (2)
4. Remote sensing
 Only relevant if aim is to gain an understanding based on specific physical
characteristics of an area (e.g. agriculture and vegetation health analysis, shelter
damage assessment, flood impact assessment, etc.)
 Representative sampling or even census could be possible
5. Secondary data review and “expert” consultations
 Most relevant for: needs analysis or HSM
 Only feasible if relevant and «reliable» data sources already exist
6. Paper form submissions
 Only applicable if respondents have no movement restrictions and are able to
send paper forms back through required means
 Logistically difficult, not the most time and resource efficient
 Most relevant for: community-level needs assessments or HSM (KI
interviews), PDMs (individual perception surveys)
 Representative sampling could be possible (but extremely difficult to
implement e.g. would need postal address database and expect very low
response rates)

lecture5.pptx

More Related Content

Similar to lecture5.pptx (20)

Recently uploaded (20)

lecture5.pptx

Editor's Notes