CREDI Users Guide 20Apr2017 - cdn2.sph.harvard.edu · The CREDI has been tested in more than 15...
Transcript of CREDI Users Guide 20Apr2017 - cdn2.sph.harvard.edu · The CREDI has been tested in more than 15...
1
User’s Guide
April 20th, 2017
Dana Charles McCoy, PhD Harvard Graduate School of Education
Günther Fink, PhD Harvard T. H. Chan School of Public Health
Caregiver-Reported Early Development Index
2
Table of Contents
Executive Summary P 3
1. Background 4
2. Conceptual Framework 5
3. Purpose & Aims 7
A. General Objectives 7
B. Summary of the Short Form 7
C. Summary of the Long Form 7
4. Accessing Information 8
5. Item Format 9
A. Response Scale 9
B. Item Wording 9
C. Examples 9
D. Illustrations 9
6. Adaptation & Translation 10
A. Adaptation 10
B. Translation 10
7. Administration 11
A. General Guidelines 11
B. Choosing the Correct Items 11
C. Mode of Administration 12
D. Instruction Prompts 12
E. Protocol for Administration: In-Person Interview 13
F. Protocol for Administration: Online Survey 15
8. Scoring 16
A. Short Form 16
B. Long Form 17
C. On Track Development 17
9. Frequently Asked Questions 18
10. More Information on the CREDI 20
A. Academic Papers 20
B. Website 20
C. General Contact 20
D. CREDI Developers 20
11. Acknowledgements 21
3
Executive Summary
The Caregiver Reported Early Development Index (CREDI) was designed to serve as a population-level
measure of early childhood development (ECD) for children from birth to age three. As the name suggests, the
CREDI exclusively relies on caregiver reports, and thus primarily focuses on milestones and behaviors that are easy
for caregivers to understand, observe, and describe. The main objective of the CREDI is to assess the overall
developmental status of particular populations of interest. As such, the CREDI is not meant as a diagnostic or
screening tool, and should not be used to make claims about individuals or small groups of children.
In what follows below, we provide more detail on the background and conceptual ideas behind the CREDI, as well
as guidance on how to use the tool in practice. We strongly encourage all researchers and policy makers to keep
the following 5 principles in mind when using the CREDI:
1. The CREDI is an open-source tool developed for the global community. There are no fees or royalties
involved with using CREDI.
2. The CREDI has been tested in more than 15 high-, middle- and low-income countries, and is designed to be
culturally and linguistically neutral. Adjustments of the tool to local contexts should not be necessary.
In case some items seem too hard or not suitable, you should contact the CREDI research team before
making adaptations.
3. There are two versions of the CREDI: A Short Form, which has exactly 20 questions for each child, and a
more detailed Long Form which has up to 100 questions per child. For large-scale surveys and monitoring
efforts, we recommend the use of the Short Form. For research projects, the Long Form will provide more
domain-specific detail.
4. The CREDI Short Form creates a summary score for children’s overall developmental status; the Long
Form creates domain-specific developmental scores.
5. The CREDI was designed as a population-level developmental assessment, and was not designed as an
individual screening tool to detect early developmental delays or disorders.
For further questions, please kindly see the Frequently Asked Questions section at the end of this document or
contact us directly.
Sincerely,
The CREDI Team
4
1. Background
Growing evidence has confirmed the importance of investing in early childhood development (ECD) for
enhancing the economic, health, and educational status of individuals, communities, and nations (Heckman, 2006;
Nores & Barnett, 2010; Shonkoff, 2010). Previous research studies have used well-validated direct assessments of
children’s functioning (e.g., the Bayley Scales of Infant and Toddler Development, the Malawi Developmental
Assessment Tool) to evaluate local intervention effectiveness in this age group. In addition, a number of
developmental screeners (e.g., the Denver Developmental Screening Test) have been used to identify early
symptoms of developmental disability and delay in clinical settings. At the same time, the cost, time, and training
necessary for reliable implementation of these individual-level assessments precludes their use for examining
large numbers of children within population-representative samples. Furthermore, the cross-cultural
comparability of these tools remains relatively untested.
In recent years, several measures of children’s ECD status have been developed for large-scale use, including the
Early Childhood Development Index from UNICEF’s Multiple Indicator Cluster Survey (MICS), Save the Children’s
International Development and Early Learning Assessment (IDELA), and the Inter-American Development Bank’s
Regional Project on Child Development Indicators (PRIDI). Presently, however, no measures of population-level
ECD have been validated specifically for children ages 0 to 3, making cross-national comparisons of developmental
status and progress for the youngest – and potentially most vulnerable – children impossible. The primary aim of
the Caregiver-Reported Early Development Index (CREDI) is to address this gap by providing the first
population-level measure of ECD for children from birth to age three. In doing so, we hope to provide a tool
that allows us to quantify children’s skills and needs across countries with the ultimate objective of making more
informed decisions regarding ECD policies and resource allocation. We also hope the CREDI tool will be a useful
instrument to monitor progress in achieving global goals congruent with the post-2015 international development
agenda.
5
2. Conceptual Framework
From birth to age three, children’s brains are developing more rapidly than in any other developmental stage,
building between 700 and 1,000 neural connections per second (Harvard Center on the Developing Child, 2007).
The CREDI is designed to capture five inter-related domains of ECD that have not only been shown to develop most
rapidly within the 0 to 3 age range, but also to be integral in predicting later life success (See Figure 1).
First, the CREDI focuses on children’s motor development, or their ability to use fine and gross movements to
explore and engage with their environments. Second, the CREDI captures children’s language development, or
their ability to communicate their
needs and desires, and understand
what others are saying to them.
Third, the CREDI measures
children’s cognitive development,
including their ability to pay
attention, remember information,
perceive and discriminate between
objects and people in their
environment, solve problems, and
acquire basic knowledge. Fourth, the
CREDI considers children’s social-
emotional development, including
their ability to control their
behaviors and emotions, understand
their feelings, and get along well
with others. Finally, the CREDI
captures early symptoms of
children’s mental health, including
the absence of behaviors related to
aggression, anxiety, and distress.
In developing our conceptual framework for measuring ECD in the 0 to 3 age range, we point out several important
considerations that distinguish the CREDI from previous measurement approaches:
First, in addition to the motor, language, and cognitive skills that have been the traditional foci of individual ECD
assessments and screeners, we also consider an additional set of developmental competencies that prior
evidence has linked with later schooling and adult success (Blair & Razza, 2007; Caspi, Moffit, Newman, & Silva,
1996; Diamond, 2013; Mischel et al., 2010; Moffit et al., 2010). In particular, our cognitive and social-emotional
domains include items related to children’s early regulatory skills and executive function, including their ability to
focus and sustain attention, delay gratification, avoid impulsive reactions, and sooth themselves. In addition, our
social-emotional domain emphasizes children’s social and emotional competencies such as getting along with peers,
engaging in play, labelling and understanding emotions, theory of mind, and empathy. Finally, we also focus on
emerging mental health symptoms, including potential signs of anxiety, withdrawal, aggression, or emotional
lability.
Second, we draw explicit attention to both the achievement of developmental milestones (i.e., children’s
progressive acquisition of discrete skills like walking or talking), as well as the measurement of children’s typical
behaviors (i.e., the ways that they act and engage with their environments on a day-to-day basis). Whereas
milestones are skills that typically develop within a relatively narrow age window and then persist or advance with
time, behaviors may show slower growth or differential developmental patterns, with these patterns being heavily
influenced by a host of environmental inputs (e.g., risk and protective factors such as poverty and stimulation) in
addition to biological processes. The majority of items in our motor, language, cognitive, and social-emotional
6
domains represent developmental milestones, whereas many items in our mental health domain represent
behaviors.
Third, unlike previous developmental screeners that attempt to identify children with severe delays or disabilities,
the CREDI is designed to represent developmental milestones and behaviors across a normal range. In drawing
this distinction, we also make explicit the goal of population-level ECD interventions to improve the lives of all
children, rather than focusing solely on those with developmental delays (See Figure 2).
Figure 2. Conceptual distinctions between developmental screeners (left) & the CREDI (right)
Targets of measurement
Goals of interven on
Developmental
delay
Screeners CREDI
To shi the full developmental curve To eliminate
developmental delay
Screeners CREDI
The full developmental curve
7
3. Purpose & Aims
A. General Objectives
As noted above, the primary aim of the CREDI is to serve as a population-level measure of ECD for children from
birth to age three. With this broader aim in mind, the specific goals of the CREDI are to:
1. capture the five domains of ECD outlined above for children ages 0 to <3 years,
2. be clear and simple enough to be reportable by primary caregivers and implemented broadly with
minimal training (at low cost),
3. be culturally neutral in ideology and content to allow for cross-context comparison, and
4. be psychometrically valid and reliable.
The CREDI includes two different products – or forms – which are summarized below and detailed throughout the
rest of this document.
B. Summary of the Short Form
C. Summary of the Long Form
• Purpose: Measurement of population-level ECD status
• Potential Setting/Use: Nationally-representative household surveys and other global monitoring
efforts (e.g., the Multiple Indicator Cluster Surveys, the Demographic Health Surveys)
• Length & Format: 20 items per age band (0-5mo, 6-11mo, 12-17mo, 18-23mo, 24-29mo, & 30-35mo)
• Administration Time: <5 minutes
• Scoring: Continuous score of overall developmental status
• Considerations: As a population-level measure, the short form is not intended as a measure of
individual children (e.g., as a screener) or specific development within a particular domain (e.g.,
cognitive). It is not designed to have sensitivity/specificity for detecting intervention efficacy.
• Purpose: Measurement of ECD status, overall and by domain
• Potential Setting/Use: Large-scale evaluations of established intervention programs
• Length & Format: 117 items total, administered with start/stop rules by age and domain
• Administration Time: ~15 minutes
• Scoring: Continuous score of overall ECD status and domain-specific development
• Considerations: The long form provides greater detail than the short form, but still lacks the ability to
provide information about individual children. It may also not be sensitive enough to detect smaller
changes attributable to intervention, and therefore should be paired with more precise, direct-
assessment measures whenever possible.
8
4. Accessing Information
A key characteristic of the CREDI is that it is an open source tool. The CREDI forms and their supporting
documentation are freely available for researchers, agencies, and practitioners interested in measuring ECD at the
population level via the CREDI website (see For More Information section, below, for link). The CREDI should be
considered as a “living and breathing” set of items and products that will receive continual improvement and
updating. As such, users of the CREDI should closely monitor the website to find updates and new materials. As an
open source tool, we also hope that the users of the CREDI will contribute feedback and data to the CREDI website
on a regular basis to support the aim of continual improvement and expansion of knowledge.
9
5. Item Format
A. Response Scale
All items on the CREDI are administered directly to the primary caregiver of the target child using a yes/no
response scale. (If caregivers are unsure of their response, they may also choose to respond by saying Don’t Know).
B. Item wording
Children’s milestones are assessed using items beginning with the phrase “Can the child…”, whereas children’s
behaviors are assessed using items beginning with the phrase “Does the child…” Behavioral items also include
frequency anchors, which allow for a more fine-grained understanding of the frequency with which the identified
behavior occurs that can be calibrated to reflect developmental norms. Frequency anchors are important for the
behavioral items given that behaviors and traits that are non-discrete are typically measured using Likert-like
scales (e.g., Never/Sometimes/Often/Always), rather than dichotomously. This strategy allows us to distinguish,
for example, between a child who occasionally acts aggressively with objects (a developmentally normative
behavior for young children) versus a child who frequently acts aggressively (which may be a cause for concern,
even at a young age).
C. Examples
Simple examples are included in the CREDI items whenever possible to concretize skills and behaviors in ways
that are comparable cross-culturally. In particular, items asking caregivers about their children’s use of words
include examples of words such as “mama” and “ball,” and items asking caregivers about their children’s
manipulation of small objects include examples of objects such as “stones” or “sticks.” Although the vast majority
of examples are intended to function equally well across cultures, local teams should consult with the CREDI
developers to adapt or amend these examples, as needed, to facilitate interpretation, local relevance, and/or
comparability with the original intent of the item. If, for example, the word for “ball” in another language is
linguistically complex (e.g., multi-syllabic), difficult to say, or culturally inappropriate, the local team might decide
to replace this word with a more appropriate example matching the intent of the item. (See section below on
Adaptation and Translation for details of this process.)
D. Illustrations
In addition to written examples, the CREDI also includes
illustrations for items in the motor domain to help
caregivers to visualize the skill being assessed (see, for
example, Figure 3). These illustrations are designed to
be shown to every caregiver after reading the
corresponding item text. Illustrations are included in a
separate document that should be printed for in-person
interviews and are marked on the CREDI Short Form by
a * . (All motor items in the CREDI Long Form include an
illustration and are therefore not marked.) In order to
ensure consistency of implementation and cross-
contextual comparability, no additional information,
examples, or demonstrations beyond what has been
provided with the CREDI should be administered to the
caregivers.
Figure 3. Example illustration for motor item
“Can the child pick up a small object (e.g., a small
toy or stone) with just his/her thumb and a finger?”
10
6. Adaptation & Translation
A. Adaptation
As is noted above, the CREDI is designed to be “culturally neutral” in that it should only reference developmental
skills and behaviors that are universally relevant and important for children within the 0 to 3 age range.
Furthermore, the objects and ideas used as examples in the CREDI are designed to be available regardless of
culture, socioeconomic status, or urbanicity. Extensive piloting and field testing using cognitive interviews with
caregivers has been conducted to ensure that each item is clear and consistently understood across a number of
diverse cultural and linguistic contexts. As a result, CREDI items should not require adaptation within a specific
context in order to provide a useful picture of children’s ECD status. Adaptation of the CREDI items would in fact
impact their comparability, and is therefore not recommended. However, individual research teams or agencies are
encouraged to supplement the core set of CREDI items with additional items or measures that have been shown to
function well in their particular settings. Such supplementation allows for a customized approach to measurement
that can provide more detailed information regarding specific needs and strengths in a given setting.
One exception to this rule of maintaining exact comparability across settings is in the examples used to clarify an
item. As is noted above in the Item Format section, examples are often included in an item to make the item more
concrete and easily understood by a caregiver. Certain items may require the use of different examples than what
is suggested in the English (core) version of the CREDI. Some questions, for example, provide examples of simple
words such as “ball” that a child may say before he/she is able to say more complex words. If the word “ball” is
linguistically complex (e.g., multi-syllabic), difficult to say, or culturally inappropriate in a particular
language/setting, the local team should replace this word with a more appropriate example word.
Before amending an item, teams should consult the CREDI website to identify whether other users have already
adapted the CREDI for their local setting. Teams should also consult with the CREDI developers (see Contact
Information, below) before making substantive changes. Furthermore, all adaptations should be made with a
careful eye to the original intent of each item. The CREDI Item Guide provides detailed information on the intent
of each CREDI item and should be consulted closely to ensure that the original item’s meaning is retained.
B. Translation
Although cultural adaptation is not required, appropriate translation of the CREDI is critical to its utility and
comparability as a population-level measure. Users should consult the CREDI website to see whether an approved
translation has already been developed for their language/setting of interest. If not, we recommend that all
individual items and prompts are translated and back-translated by different native speakers of the particular
dialect used within the target population. Discrepancies between back-translated and original items must be
reviewed and resolved prior to implementation. Translations should be made such that a caregiver with minimal
formal education can understand the items easily. In other words, colloquial wording should be used instead of
formal or academic language. Likewise, literal translation is not necessary when translating the CREDI. Rather,
teams should use “meaningful translation” techniques that allow for the meaning or intent of an item to be
reflected in a translation. Once again, teams should refer closely to the CREDI Item Guide document, which
provides a detailed description of the meaning and intent behind each item.
As the CREDI is designed to be an open source tool, we encourage teams to keep detailed records of translation
difficulties to help us to refine items’ wording over time, and to communicate openly with the CREDI core team
about translation questions and issues. We also hope that teams who have completed translation of the CREDI will
share these versions with the broader ECD community using the CREDI website.
11
7. Administration
The information in this section provides guidance on how to use the CREDI Short Form or the CREDI Long Form in
the field. We recommend that CREDI users include relevant information from this section as part of their study’s
training manual.
A. General Guidelines
All items on the CREDI are addressed to the primary caregiver of the target child, and are not based on any direct
interactions with the child. The target child is the child who has been identified in the study protocol as being the
subject of the interview. Although study- or sample-specific criteria may apply when selecting the target children,
the CREDI is designed to be administered to all children in the 0 to <3 age range, regardless of disability status,
culture, language, etc. All CREDI items include text referring to “the child.” Assessors should feel free to replace the
text of “the child” with the target child’s first name or with “he” or “she” in order to reduce the formality of the
interview.
The primary caregiver is the person who cares for the target child most often and who knows the most about his
or her abilities and behavior. Although the primary caregiver is often the mother, this is not always the case. If the
father, sibling, grandparent, or neighbor cares for the child more than the mother, than this person should be
considered the primary caregiver and be interviewed.
For all CREDI items, there are three response options: Yes, No, and Don’t Know. All responses should be checked or
coded according to the protocols of the specific study, though it is recommended that Yes responses are coded as 1,
No responses are coded as 0, and Don’t Know responses are coded as 8 or 9. Whenever possible, caregivers should
be encouraged to respond with either Yes or No to avoid missing data. Don’t Know should only be used when the
caregiver is unfamiliar with the child’s behavior or skills in a particular area and cannot accurately respond using
Yes or No.
B. Choosing the Correct Items
To minimize the effort of administering and answering questions, both the CREDI Short Form and the CREDI Long
Form include different items depending on children’s age. To begin, data collectors should know the precise age of
the target child, in months. A child who is 8 months and 29 days is still considered to be 8 months old.
Below are directions on how to determine which items to administer to the target child based on his/her age.
Short Form
The CREDI Short Form includes different sets of items that are administered to children in 6 different age bands:
0-5 months, 6-11 months, 12-17 months, 18-23 months, 24-29 months, and 30-35 months. Each age band
includes a total of 20 items that cover different developmental domains. Although some items appear on more
than one set of items, many of the items in each age band are unique. Items were selected for each age band based
on developmental appropriateness. For example, because almost all children are able to sit up without support
before their first birthday, this item is not included for older children.
Only one set of items should be administered to each child receiving the CREDI Short Form. For example, if a child
is 4 months old, he/she should receive the set of items for 0-5 months.
Long Form
The CREDI Long Form includes four sets of items: one set for motor items, one set for cognitive and language items,
one set for social-emotional items, and one set for mental health items. Each set of items is ordered by items’
difficulty, where easier items (e.g., standing) come before more difficult items (e.g., walking, running). The CREDI
12
Long Form uses start/stop rules to determine which items should be administered to each child. Like the CREDI
Short Form, 6 different age bands are used: 0-5 months, 6-11 months, 12-17 months, 18-23 months, 24-29 months,
and 30-35 months. The start/stop rules differ slightly by domain.
Start rules: For the motor and cognitive/language domains in the CREDI Long Form, data collectors should identify
the first item corresponding to the target child’s age band. For example, if a child is 18 months old, the data
collector should start with the item next to the “18-23 month” age band marker. For the motor and
cognitive/language domains, if the caregiver responds No to any of the first three items asked, the data collector
should go back to the item corresponding to the next lowest age band. For example, if the caregiver of the 18-
month-old child reports that the child cannot yet complete at least one of the first three items, the data collector
should find the start point for the 12-17 month age band and administer those items. When all of the 12-17 month
age band items in the particular domain are complete, the data collector can then continue with the rest of the 18-
23 month age band items. For the social-emotional and mental health items, there are no age-dependent start
rules; data collectors should begin with the first item for all children.
Stop rules: Once a starting place has been established (i.e., caregivers have responded affirmatively to at least
three questions in a row, or the earliest questions have been administered), for the motor, cognitive/language, and
social-emotional domains, the data collector should continue to ask items on the CREDI Long Form until the
caregiver responds “No” to 5 items in a row or until all items have been asked. If, for example, the caregiver
responds Yes to all items in the 18-23 month age band, the data collector should continue by asking the items in the
24-29 month age band, etc. If a caregiver responds “No” to any of the easiest three items on the scale (items 1-3,
youngest age group), no further questions should be asked. For the mental health items, all items should be asked
to all children.
One important consideration is that some of the items in the social-emotional and mental health set are reverse
coded, where a “No” response is actually a positive indication of development and a Yes response is a negative
indication of development. For example, a “No” response on the item “Does the child become upset when you are not with him/her?” would indicate a high level of mental health. These items need to be recoded prior to scoring (see Scoring
section, below), and should not be used to compute simple averages in the raw data.
C. Mode of Administration
The CREDI Short Form and the CREDI Long Form can both be administered in different ways. To date, two primary
modes of administration have been used: in-person interviews and online surveys. Using the in-person
interview, a data collector asks each CREDI item out loud to the caregiver, who then responds verbally by saying
Yes, No, or Don’t Know. This verbal response is then recorded by the data collector. This mode of administration is
appropriate in contexts in which literacy rates and/or access to the internet are low.
The online survey option, on the other hand, allows the caregiver to respond directly to each CREDI item at his or
her own pace by clicking Yes, No, or Don’t Know for each item. Links to an online version of the CREDI can be found
on the CREDI website. Teams interested in administering their own online survey should follow a similar format.
Users should choose the mode of administration that best suits their needs and resources. Initial evidence from
the United States suggests that the interview and online modes of administration provide data that are relatively
comparable, although additional testing may be needed within a given context to confirm that this is true.
D. Instruction Prompts
Both the CREDI Short Form and the CREDI Long Form include prompts to explain the purpose of the items to the
caregiver. In the in-person interview format, these prompts should always be read out loud prior to asking the
items. In the online format, these prompts should come at the top of the page, before the CREDI questions are
listed. The instruction prompts are designed to help the caregiver understand what types of questions will be
asked and how to respond to the questions (i.e., by saying Yes, No, or Don’t Know). Prompts are also designed to help
13
caregivers feel comfortable and reduce the potential for social desirability (i.e., answering in a way that is
considered socially acceptable but not necessarily representative of the child’s true skill/behavior).
Instruction prompts vary slightly in wording depending on the mode of administration. Specific prompts to be
included at the start of the assessment for both the Short Form and the Long Form are listed below. For the Long
Form, this prompt will be listed just before beginning the motor items.
In-Person Interview
Online Survey
The CREDI Long Form also has several additional prompts that are included at the beginning of each domain to
remind caregivers about the rules of the data collection, and to signal the beginning of a new domain. These
prompts are the same regardless of assessment format (in-person versus online):
Cognitive/Language
Social-Emotional
Mental Health
E. Protocol for Administration: In-Person Interview
When the CREDI is administered using an in-person interview format, data collectors should follow standard
operating procedures to ensuring respondent comfort and research integrity. Although the specific approaches
that are used may vary based on culture or study protocol, we strongly recommend using the following general
strategies when administering the CREDI.
Preparing for the Interview
Now I am going to ask you about the types of things your child is currently able to do. Please answer "yes" or "no" to these questions. If you are unsure, you can also answer "don't know." Please keep in mind that children learn and grow at different rates, so it is fine if your child can't yet do these things. Some of these skills children only achieve at older ages. If there is any question you feel uncomfortable answering, please let me know and we can move to the next question.
Below [or on the next screen] you are going to see questions about the types of things your child is currently able to do. Please respond to these questions by clicking "yes" or "no." If you are unsure of the response, you can also click "don't know." Please keep in mind that children learn and grow at different rates, so it is fine if your child can't yet do these things. Some of these skills children only achieve at older ages. If there is any question you feel uncomfortable answering, please skip it and move to the next question.
Now we are going to ask a new set of questions about what your child can do. If you are unsure, you can still answer "don't know."
Now we are going to ask a new set of questions about your child's behavior. The rules for these questions are the same. If you are unsure, you can still answer "don't know." If there is any question you feel uncomfortable answering, feel free to move on to the next question.
We have just a few more questions to answer. The rules for these are the same. Please continue to answer with "yes", "no," or "don't know."
14
• Make a good first impression by dressing nicely, introducing yourself, and being polite. Act professionally,
and do not overstay your welcome or burden the caregivers in any way. Try to avoid visiting the caregiver
during meal times or other points of the day when he/she will be busy or distracted. Scheduling the
appointment in advance is a smart way to ensure that you will be welcomed.
• Interview the caregiver in a location that is private and free of distractions. If children are present, they
should be cared for by another responsible adult whenever possible during the interview. Other adults (e.g.,
fathers) should not be present during the interview. Only the responses of the primary caregiver should be
recorded.
• Before administering any CREDI, be sure to go through the study-specific consent process to ensure that the
caregiver understands the risks and advantages of answering the questions, whether the information will be
kept private, etc. Answer all questions from the caregiver before moving forward with the survey.
• Read the instruction prompt exactly as it is written. Provide clarifications on the instructions if the
caregiver has questions.
Administering the CREDI Items
• Ask all CREDI items exactly as they are written. Do not skip, replace, or add any words. Do not include any
additional examples, gestures, or actions to clarify the item. The only words that can be changed are “the child,”
which can be replaced with target child’s first name or with “he” or “she.”
• For the Long Form, be sure to follow the specific guidelines listed above regarding start and stop rules.
Remember that these rules differ depending on the domain being administered, with some domains (e.g.,
motor, language/cognition) requiring different start points based on age.
• Ask all items in exactly the same order that they are listed on the survey form. Do not change the order for
any reason.
• When items include an illustration (marked by a * on the Short Form, and all Motor items in the Long Form),
read the item first, and then show the corresponding illustration to the caregiver. Show only the
illustration that is relevant to the particular item. Do not show more than one illustration at a time, and do not
show the text of the item. Allow the caregiver enough time to see the illustration before requesting a response.
• Allow the caregiver time to think about his/her response. Do not rush him/her.
• Do not help caregivers by giving examples that are not listed in the questions themselves, by demonstrating
the behavior or skill, or providing any additional information other than what is stated in the item itself (or the
accompanying illustration).
• If the caregiver is unsure of how to respond to a item, politely but clearly ask him/her to make his/her best
guess.
• Based on the caregiver’s verbal response, mark the appropriate response (Yes, No, or Don’t Know). Do not
interpret ambiguous responses (e.g., “My child does that sometimes”) but rather ask the caregiver to respond
using one of the three response options. If a response is unclear, ask the caregiver to repeat him/herself.
• Do not leave any items blank. All items must be answered as either Yes, No, or Don’t Know.
Addressing Concerns
• If the caregiver is bored, distracted, or hesitant to respond, take a short break to engage him/her in unrelated
conversation to enhance attention and comfort levels.
• If the caregiver is uncomfortable or anxious about the questions or survey in general, remind him/her
intermittently that there are no right or wrong answers, that children grow and show behaviors at different
rates, and that what is normal for one individual child or age group may not be normal for another. You may
also let him/her know that the items are meant to be difficult, and that most children will not be able to do all
of these things because they are still very young.
• Remind caregivers intermittently that all responses are confidential and will not be shared with others.
• If caregivers exhibit continued or severe signs of distress, discomfort, or concern during the CREDI
administration, assessors should follow study-specific protocols for discontinuing the survey and/or
providing referrals for child-or family-related support services.
15
• Occasionally during the course of a survey a caregiver may express a specific concern about a child’s ability
that is either related or unrelated to the CREDI items (e.g., concerns about physical or mental disability,
malnutrition, infection, language skills, etc.). If this occurs, the data collector should tell the caregiver that
he/she is not a clinician or health/development professional and cannot make a specific diagnosis. The data
collector can also let the caregiver know that the questions are not designed to give a diagnosis to the child.
The data collector should then refer the caregiver to a specific clinic or service provider according to the
broader study’s protocol.
Ending the Interview
• Before ending the interview, quietly review all items and ensure that there are no missing responses. If there
are missing responses, ask the caregiver the item and fill in his/her response.
• End the interview by thanking the caregivers for their participation and asking if they have any remaining
questions or concerns. Do whatever is possible within the study protocol to address any issues.
Throughout the course of the interview, data collectors can choose to keep detailed notes on items that were
particularly difficult for caregivers to answer, as well as any concerns that caregivers had about the individual
items or interview as a whole. Because the CREDI is an open source tool, these notes may be submitted to the
CREDI team to help to provide additional information on how to improve the CREDI in the future.
F. Protocol for Administration: Online Survey
When the CREDI items are administered in an online survey format, the protocol for administration is relatively
straightforward. Caregivers are linked directly to the online survey via the CREDI website or another study-
specific recruitment source (e.g., email, listserv, etc.). The first several sets of screens for the online survey should
include screening and consent information that determine (1) whether the caregiver is literate in the given
language, (2) the age of the target child, and (3) whether the caregiver provides his/her consent to participate in
the survey. The script of the consent should be tailored to the specific study and may, for example, notify
caregivers of their ability to discontinue the survey at any point, the confidential and anonymous nature of the
survey, or the contact information should they wish to ask any questions or follow up about the survey.
Following these introductory screens, the caregiver should automatically be presented with the appropriate
instruction prompt (see Instruction Prompt section, above) and set of CREDI items for his/her child’s age group and
form (Short versus Long). The instruction prompt should be shown at the top of the screen, before the CREDI
items are listed. All motor items should include the appropriate Illustrations next to the text of the item.
Caregivers should then respond to each item by clicking on Yes, No, or Don’t Know. After completing all items, the
caregiver then clicks the “next” button at the bottom of the screen to finish the CREDI. The survey can conclude
with an additional set of items (e.g., demographic questions) and/or a “thank you” screen.
16
8. Scoring
A. Short Form
The CREDI Short Form is designed to produce a single age-specific standardized ECD score (value) for each child.
The CREDI Short Form does not provide domain-specific scores due to the small number of items. To generate
domain-specific scores (e.g., a score specific to children’s language abilities), users should implement the CREDI
Long Form.
Below is a summary of the steps necessary to generate age-specific standardized scores on the CREDI Short Form:
1. Administer all 20 age-specific items on the CREDI Short Form. For details on administration, see
Administration section, above.
2. Recode scores according to the following rules:
a. Recode all “9” or other non-0/1 values as missing.
b. Reverse code selected items. Specifically, recode the following items so that the original 1(Yes)
values are coded as a 0 and the original 0(No) values are coded as a 1:
Age group 24-29 Age group 30-35
CREDI_E11 CREDI_F16
CREDI_E13 CREDI_F17
CREDI_F18
3. Sum up the number of items that have been completed for each child. In other words, count how
many of the 20 items have valid (non-missing) answers for each child.
4. Sum up the total score for each child. In other words, add the number of 1s (“Yes” for most items, “No”
for reverse coded items) for each child.
5. Calculate the adjusted raw score for each child by correcting the total score for missingness. To do
so, use the following formula:
adjusted_raw_score = (total_score*20)/(items_completed)
6. Recode the final raw score as missing if the number of completed items is <15. Scores with high rates
of missing data are not trustworthy and therefore should not be analysed.
7. Based on the final raw score, look up the age-specific standardized score in the table below for each
child. For example, if the child is 14 months and had a final raw score of 5, his/her age-specific
standardized score is 44.7. This standardized score is preferable to the raw score because it allows for
comparison across age bands.
Table 1. CREDI Short Form age-specific standardized scores
CREDI Short
Form raw score
CREDI Short Form Standardized Score (by Age Band)
0-5 mo 6-11 mo 12-17 mo 18-23 mo 24-29 mo 30-35 mo
0 21.8 29.1 38.3 46.2 48.3 48.7
1 23.8 30.3 39.8 47.3 49.5 49.9
2 26.9 32.1 41.6 48.9 50.9 51.4
3 28.8 33.3 42.9 49.9 51.9 52.3
4 30.2 34.2 43.8 50.7 52.6 53
5 31.3 35.1 44.7 51.4 53.2 53.6
17
6 32.2 35.9 45.5 52 53.8 54.1
7 33 36.7 46.2 52.6 54.3 54.6
8 33.7 37.5 46.8 53.1 54.8 55.1
9 34.4 38.3 47.5 53.6 55.2 55.5
10 35.1 39.1 48.1 54.1 55.6 55.9
11 35.8 39.9 48.7 54.6 56 56.3
12 36.5 40.8 49.4 55.1 56.4 56.7
13 37.3 41.6 50 55.5 56.8 57
14 38.1 42.4 50.6 56 57.2 57.5
15 39 43.3 51.3 56.4 57.7 57.9
16 40 44.3 52 57 58.2 58.4
17 41.2 45.3 52.9 57.5 58.8 59.1
18 42.6 46.6 54 58.3 59.6 60
19 44.7 48.5 55.7 59.6 61 61.4
20 46.9 49.9 56.9 60.5 61.9 62.4
B. Long Form
The Long Form scoring system is still under development and should be released in the latter half of 2017.
C. “On Track” Development
In addition to the continuous scores provided by the current versions of the CREDI Short Form and the CREDI Long
Form, in the future we hope to provide cut-points for on versus off track status. These cut-points are
forthcoming.
18
9. Frequently Asked Questions
Q: Can the CREDI be used to diagnose or screen children for early signs of disability or developmental delay?
A: No. The CREDI is not designed to provide information on individual children and therefore should not be used
in this way.
Q: I am planning a study to evaluate the effects of an intervention program. Can I use the CREDI in my
evaluation?
A: Yes, with some caveats. The CREDI Long Form can be used for the purpose of intervention evaluation,
particularly if the use of more costly approaches (e.g., direct assessment) in the full sample is infeasible or
impractical. At the same time, we recommend that, whenever possible, the CREDI should be paired with a
more detailed and domain-focused measure that captures finer differences in the specific skill(s) that your
intervention is designed to improve. If, for example, your intervention is targeting children’s language
outcomes, we recommend that you pair the CREDI Long Form with a direct assessment of children’s language
skills in at least a subsample of your participants. Doing so will help to mitigate the possibility that you may fail
to detect small but potentially meaningful effects of your program in at least one domain. Furthermore,
triangulation of measurement (i.e., using multiple approaches to measure the same thing) is useful for
addressing potential weaknesses in one approach versus another. Given that the CREDI is a caregiver-reported
scale, using a direct assessment to address issues of social desirability (for example) may be useful.
Q: I am hoping to adapt some of the CREDI items so that they are more targeted to my specific population. Is
that ok?
A: No. The CREDI Short Form and the CREDI Long Form have been validated as scales that are designed to be use
exactly as they are written. You should not adapt the items or change them in any way unless absolutely
necessary (see Adaptation and Translation section, above). If you want to add more items that are specifically
relevant to your population, you may create your own items or borrow them from another tool. If you choose
to do this, you should still score the CREDI using only the “core” items based on the suggestions in the Scoring
section above. If you want to integrate your additional items, you will need to validate your new scale on your
own and be clear in publishing about the changes you have made.
Q: I want to use the CREDI, but I am only interested in using a subset of the questions. Can I pick and choose
which items I use?
A: No. Once again, the CREDI Short Form and the CREDI Long Form are designed to be used as-is. You should not
get rid of any items or you will compromise the properties of the scales. If, however, you are interested in
adding items, you may do so with the understanding that you will need to develop your own separate scoring
method and can no longer call your scale the CREDI.
Q: I recently translated the CREDI into a new language that is not currently on your website. Should I send
this to you?
A: Yes! We are always looking for new translations! Please feel free to follow the instructions listed in the
Adaptation and Translation section (above) and send us your new translation at [email protected]
Q: I have a suggestion for improving the CREDI’s wording/usefulness/cultural relevance. Who should I talk
to?
A: We are always eager to hear from users about their experiences with the CREDI and how we can improve it for
the future. Please contact us at any time at [email protected]
Q: I am looking for other measures of ECD that I might use instead of or in addition to the CREDI. Do you
have any suggestions?
A: There are many wonderful tools available for measuring ECD in diverse global settings. You should always
select your measures to best fit your specific needs, populations, and resources. We recommend several
resources for identifying and selecting measures:
• The World Bank Toolkit on ECD Measurement in LMICs: http://siteresources.worldbank.org/INTCY/Resources/3957661187899515414/Examining_ECD_Toolkit_FULL.pdf
19
• The Inter-American Dialogue Report on Measuring ECD in the Americas: http://www.thedialogue.org/resources/measuring-early-childhood-development-in-latin-america/
Q: What are your “next steps” for the CREDI?
A: The CREDI is an evolving tool that we hope to continue to update and refine. In particular, we are currently
collaborating with ECD experts from the WHO, UNICEF, and a number of academic institutions to plan a future
study using CREDI items and other ECD tools to develop ECD norms, standards, and benchmarks. Please visit
our website frequently for updates!
20
10. More Information on the CREDI
A. Academic Papers
McCoy, D. C., Sudfeld, C., Bellinger, D. C., Muhihi, A., Ashery, G., Weary, T. E., Fawzi, W., & Fink, G. (2017).
Development and validation of an early childhood development scale for use in low-resourced settings,
Population Health Metrics, 15(3).
McCoy, D. C., Black, M., Daelmans, B., & Dua, T. (2016). Measuring population-level development in 0-3. Early
childhood matters. The Hague, Netherlands: Bernard van Leer Foundation.
B. Website
sites.sph.harvard.edu/credi/
C. General Contact
D. CREDI Developers
Dana Charles McCoy, PhD
Assistant Professor
Harvard Graduate School of Education
Günther Fink, PhD
Associate Professor
Harvard School of Public Health
21
11. Acknowledgements
The authors of the CREDI would like to acknowledge the generous financial and intellectual support Grand
Challenges Canada through the Saving Brains initiative. We also acknowledge and thank the Harvard School of
Public Health, the Harvard Graduate School of Education, the Harvard Center on the Developing Child, UNICEF, and
the WHO for their organizational support, as well as the members of our Advisory Panel for their valuable input.
Finally, we would like to express our gratitude to the thousands of caregivers and children around the world who
participated in our validation effort.