Children’s Health Care, 39:142–156, 2010
Copyright © Taylor & Francis Group, LLC
ISSN: 0273-9615 print/1532-6888 online
Assessing Children’s Mealtime
Problems With the Mealtime Behavior
Kristoffer S. Berlin
Department of Psychiatry and Human Behavior, Brown Medical School;
and Bradley Hasbro Children’s Research Center, Rhode Island Hospital,
W. Hobart Davies
Department of Psychology, University of Wisconsin–Milwaukee; and
Department of Pediatrics, Medical College of Wisconsin, Milwaukee, WI
Alan H. Silverman
Department of Pediatrics, Medical College of Wisconsin, Milwaukee, WI
Douglas W. Woods
Department of Psychology, University of Wisconsin–Milwaukee,
Elizabeth A. Fischer
Departments of Psychiatry & Behavioral Medicine, and Pediatrics,
Medical College of Wisconsin, Milwaukee, WI
Colin D. Rudolph
Department of Pediatrics, Medical College of Wisconsin, Milwaukee, WI
A caregiver questionnaire that assesses mealtime problems in children aged 2
to 6 years old was developed. Community caregivers .n D712/ completed the
Correspondence should be addressed to Kristoffer S. Berlin, PhD, Department of Psychology,
Ohio University, 200 Porter Hall, Athens, OH 45701. E-mail: email@example.com
MEALTIME BEHAVIOR QUESTIONNAIRE 143
Mealtime Behavior Questionnaire (MBQ) and measures of child behavior and
family mealtime behaviors and environment. Exploratory and conﬁrmatory factor
analyses revealed and validated the MBQ’s 4 subscales (food refusal/avoidance;
food manipulation; mealtime aggression/distress; and choking/, gagging/vomiting).
Mealtime problems occurred from “sometimes” to “always” for 1% to 61% of
the sample. The MBQ demonstrated excellent to fair internal consistencies, and
preliminary evidence for validity was found.
The survival, health, well-being, and development of all children depend on
eating a nutritionally balanced diet. Despite the seemingly instinctive ease with
which most children eat and feed, there are complex interactions among many
factors (genetic, biological, psychological, sociocultural, family, and environ-
mental) that may impede the development of healthy eating habits and behaviors.
There are alarmingly high prevalence rates of feeding problems in early child-
hood, with estimates ranging from 25% to 45% among typically developing
children (Bentovim, 1970; Linscheid, 1992; Manikam & Perman, 2000) and
reaching as high as 85% for children with developmental disabilities (Manikam
& Perman, 2000).
Behavioral feeding problems present across a range of behaviors and sever-
ities. They can include inappropriate mealtime behaviors, lack of self-feeding,
failure to advance texture from puree to table food texture, food refusal or
accepting only small quantities, picky eating, eating too slowly, frequent vom-
iting, eating too much, parental aversion to mealtimes, mealtime aggression
and distress, or poor dietary regimen adherence (Crist & Napier-Phillips, 2001;
Kerwin, 1999; Mackner, McGrath, & Stark, 2001).
Children with severe feeding difﬁculties are at risk for a variety of problems.
These include severe weight loss, malnutrition, lethargy, impaired intellectual–
emotional–academic development, growth retardation, aspiration, invasive medi-
cal procedures (e.g., placement of a nasogastric or gastrostomy tube), admission
to an inpatient unit for treatment of the feeding problem, or death (Budd et al.,
1992; Christopherson & Hall, 1978; Finney, 1986; Linscheid, 1978; Riordan,
Iwata, Wohl, & Finney, 1980; Sisson & Van Hasselt, 1989; Skuse, 1993; Whitten,
Pettit, & Fischhoff, 1969).
Eating and adequate nutrition also play key roles in the health and devel-
opment of children with medical conditions (Mackner et al., 2001). Among
the numerous medical conditions that require dietary modiﬁcation, disregulated
eating and impaired nutritional status are linked with signiﬁcant morbidity and
the exacerbation of the disease process (for a review, see Mackner et al., 2001).
Medical conditions in which nutrition plays a signiﬁcant role include cancers,
celiac disease, coronary heart disease, chronic renal disease, congenital heart
disease, cystic ﬁbrosis, encopresis, HIV infection, inﬂammatory bowel disease,
liver disease, osteoporosis, phenylketonuria, seizure disorders, short bowel syn-
drome, sickle cell disease, and type 1 and type 2 diabetes (Mackner et al., 2001).
144 BERLIN ET AL.
Previous work to identify, measure, and assess behavioral feeding problems
has found that differences in parental report of feeding difﬁculties between
healthy and clinical groups are reﬂected in the frequency in which children
engage in the problematic behaviors, rather than fundamental differences in the
types or groupings of behaviors exhibited during mealtimes (Crist & Napier-
Phillips, 2001). Existing measures include About Your Child’s Eating (AYCE;
Davies, Ackerman, Davies, Vannatta, & Noll, 2007), the Behavioral Pediatric
Feeding Assessment Scale (Crist & Napier-Phillps, 2001), and the Children’s
Eating Behavior Inventory (Archer, Rosenbaum, & Streiner, 1991). These mea-
sures may list both desirable (“eats fruit”) and problematic behaviors (“My child
hates eating”), and may ask parents to endorse whether particular behaviors
are problematic and what they do to address these problems (Archer et al.,
1991; Crist & Napier-Phillps, 2001). Although useful, these measures of feeding
problems unfortunately confound the frequency of the child’s behavior problems
at mealtime with the caregivers’ strategies used to address the problematic
feeding behaviors and more general ratings of the mealtime environment. This
has left researchers and clinicians without a measure of feeding problems based
only on the frequency of child behaviors. Furthermore, this has made it difﬁcult
to separate the child’s mealtime behavior problems from the strategies caregivers
use to address these problems. Therefore, this initial pilot study had four goals:
1. Develop a parent report questionnaire to assess the frequencies of mealtime
behavior problems in young children focused only on the frequency of
2. Ascertain the factor structure of this measure using exploratory factor
3. Examine preliminary evidence for reliability and validity.
4. Determine the inﬂuence of parent gender and child age and gender on
caregiver report of mealtime behavior problems.
Data were from a sample of community caregivers of children aged 2 to 6
years old. Descriptive statistics and demographic information for the community
sample can be found in Table 1. The ethnic composition was 8% African
American, 6% Asian American, 81% Caucasian, 3% Latino, and 2% mixed
race or “other.” The average level of education (as a proxy for socioeconomic
status) was some college (MD13:79 years, SD D3:02 years). Data from
community participants were collected by 90 students enrolled in an advanced
MEALTIME BEHAVIOR QUESTIONNAIRE 145
Sample Demographic Information and Descriptive Statistics
(% Female) M SD
EFA community Child age (years) 356 2.00 6.00 3.78 1.28
Parent age (years) 355 19.00 60.00 32.64 6.55
Child gender 355 nD179 (50.3%) nD176 (49.7%)
Parent gender 355 nD121 (34.1%) nD234 (65.9%)
CFA community Child age (years) 356 2.00 6.00 3.81 1.34
Parent age (years) 353 19.00 57.00 33.10 6.36
Child gender 356 nD189 (53.0%) nD167 (47.0%)
Parent gender 355 nD133 (37.6%) nD222 (62.4%)
Note. EFA Dexploratory factor analysis; CFA Dconﬁrmatory factor analysis.
undergraduate-graduate psychology laboratory course (offered during 3 consec-
utive summers). Each student obtained data from eight or more parents who
were known to them and who had a child between the ages of 2 and 6 years
old. To facilitate data entry and collection, participants completed surveys on
www.surveymonkey.com. Hard copies of the survey were made available to
participants without Internet access. The institutional review board approved the
study, and informed consent was indicated by the completion of the survey. The
only potential beneﬁt of participating was extra credit for the recruiting student
in a psychology course if so allowed. Several steps were taken to ensure the
authenticity of the data: (a) both students and their respondents were blind to
the study’s hypotheses and aims, (b) response patterns of data uploaded were
reviewed to check for duplicate IP addresses and to screen for multiple entries
within a short time period, (c) students were required to attend research ethics
seminars, and (d) students were required to review and abide by the university’s
policy on academic dishonesty.
Mealtime Behavior Questionnaire (MBQ). Given the absence of estab-
lished measures based solely on the problematic behaviors of children during
mealtimes, 33 items were generated by psychologists and advanced psychol-
ogy trainees working as members of a multidisciplinary feeding team (see the
appendix for the MBQ). Each team member was asked to generate a range of
speciﬁc feeding problems likely to be encountered in the clinic or in the homes
of typically developing children. Items were placed on a 5-point frequency scale
ranging from 1 (never), 3 (sometimes), to 5 (always). Participants were asked,
“Please rate each behavior in terms of how often or frequently it happens.
Please rate each behavior as it occurred during mealtimes or feeding over the
146 BERLIN ET AL.
past week.” The ﬁrst round of data collection also asked parents to rate the level
of difﬁculty they experienced around each behavior. The difﬁculty ratings were
judged to be redundant and dropped (and excluded from further data collection)
when initial correlations between frequency and difﬁculty consistently exceeded
rD:85 for all factors.
AYCE (Davies et al., 2007). The AYCE is a valid and reliable 25-item
parent report measure that asks about parent beliefs and concerns regarding their
child’s eating, by asking about the frequency of child eating behaviors, parents’
mealtime interactions with the child, and their feeling about mealtimes. The
AYCE has three scales: Child Resistance to Eating (CRE), Positive Mealtime
Environment (PME), and Parent Aversion to Mealtime (PAM). The internal
consistency of the AYCE factors is satisfactory across various demographic
subgroups, with the alphas from the total normative sample being good for
CRE (.88) and PAM (.80), and acceptable for PME (.70). The AYCE has
demonstrated convergent validity, as all scales relate (in the expected direction)
with established measures of family functioning (Davies et al., 2007).
To determine the underlying factor structure of the MBQ items, EFA was
conducted following the suggestions of Russell (2002) and Fabrigar, Wegener,
MacCallum, and Strahan (1999) using a randomly selected half of community
data (SPSS select cases function). The other half of the community data was
utilized to conduct conﬁrmatory factor analysis (CFA) to validate this factor
structure with an additional sub-sample and to determine the goodness of ﬁt
based on a variety of indexes.
Because the traditional chi-square statistic as a test of absolute ﬁt is sensitive
to sample size (Hu & Bentler, 1995), additional measures less inﬂuenced by
sample size were used, including the non-normed ﬁt index (NNFI) the com-
parative ﬁt index (CFI), and standardized root mean square residual (SRMR).
SRMR values of .08 demonstrate an adequate model ﬁt (Hu & Bentler, 1999).
NNFI and CFI values range from 0 to 1, and values above .90 represent a
good model ﬁt (Bollen, 1989; Hoyle & Panter, 1995). The root mean square
error of approximation (RMSEA; MacCallum, Browne, & Sugawara, 1996)
statistic was supplied as an indication of the population error variance (Browne
& Cudeck, 1993). Interpretation of RMSEA values in terms of ﬁt are as follows:
good (<.05), acceptable (.05–.08), marginal (.08–.10), and poor (>.10; Browne
& Cudeck, 1993; Fabrigar et al., 1999; Hu & Bentler, 1999). Error terms
MEALTIME BEHAVIOR QUESTIONNAIRE 147
(unexplained variance) among observed variables from the same latent construct
were allowed to correlate if model ﬁt was improved.
To gather preliminary evidence for construct validity, bivariate correlations
were examined (Messick, 1995). It was anticipated that the MBQ would have
signiﬁcant positive relations with other measures of mealtime problems and
inverse relations with positive ratings of the mealtime environment and child age.
Internal consistency (alpha) coefﬁcients were calculated for one index of relia-
bility. To estimate the prevalence of speciﬁc mealtime behavior problems over a
1-week period, the frequency and cumulative percentage of each MBQ item rated
3 (sometimes) or higher was calculated. Multivariate analyses of variance were
used to determine the inﬂuence of parent and child gender (and their interaction)
on caregiver report of mealtime behavior problems (the MBQ total score and
4 subscales) using an alpha level of p < :017 to control type 1 error (.05 of 3
effects: child gender, parent gender, and Child Gender Parent Gender).
EFA. To determine the number of factors to extract, a parallel analysis
(Reise, Waller, & Comrey, 2000) was conducted using MacParallel Analysis
(Watkins, 2000), a software program that contains tables of eigenvalues produced
by Monte Carlo simulations. Parallel analyses are a variant on the scree test,
where one plots the eigenvalues from actual data and data derived from factoring
a completely random set of data involving the same number of items and research
participants. The point at which the eigenvalues for the actual data dropped below
eigenvalues for the random data indicated the number of factors to be extracted.
A beneﬁt of parallel analysis is that it eliminates the subjectivity associated
with visual inspection of the scree plots (Kaiser, 1970). The parallel analysis
conducted for this study indicated that a four-factor solution was most appro-
priate. Table 2 presents the eigenvalues and percentage of variance accounted
for by each factor initially and upon extraction and the randomly generated
eigenvalues used for the parallel analyses. Based on the suggestions of Russell
Eigenvalues and Percentage of Variance Accounted for by the Initial Four Factors
Extraction Sums of
Squared Loadings Rotation Sums
Food refusal/avoidance 9.11 27.60 1.70 8.52 25.82 6.39
Food manipulation 2.58 7.82 1.60 2.00 6.06 6.00
Mealtime aggression/distress 1.87 5.65 1.54 1.38 4.19 5.66
Choking/gagging/vomiting 1.82 5.53 1.47 1.26 3.80 3.20
148 BERLIN ET AL.
(2002), EFA was then conducted using the correlation matrix, principal axis
factoring (to extract 4 factors), and promax rotation. Items with multiple factor
loadings (Items 4, 7, 21, and 25) were retained on the factor with the highest
loading. Items 15 and 29 were excluded from the factor scores given the item
content diverged from other factor items. Based on the content of the items, the
four factors were named food refusal/avoidance, food manipulation, mealtime
aggression/distress, and choking/gagging/vomiting. Table 3 shows the items and
pattern factor item loadings.
CFA model evaluation. An important assumption to satisfy the use of CFA
is that the measured variables have a multivariate normal distribution. Initial data
screening indicated departures from a normal distribution, with several items
(Items 5, 11, 14, 15, 29, 30, 31, and 32) having moderate skewness (>2)
and kurtosis (>5; West, Finch, & Curran, 1995). To address the violations
to distributional assumptions of CFAs, two suggested strategies were utilized
(Fabrigar et al., 1999; West et al., 1995). First, unweighted item sums of factor
subscales were used as item parcels-indicators of the latent variable total score
(rather than use individual items), as they (a) have more normal distributions,
(b) more closely approximate normally distributed continuous variables, (c) have
been found to produce path estimates identical to those obtained using individ-
ual items (Sass & Smith, 2006), and (d) decreased the number of parameters
estimated. The second strategy was to perform CFA using a robust maximum
likelihood estimation method in LISREL 8.54 (Jöreskog & Sörbom, 2003). This
method allowed for the computation of a Satorra–Bentler scaled (SB) and
robust standard errors, which adjust for multivariate kurtosis (Satorra & Bentler,
CFA was then employed to evaluate the factor structure of the MBQ for the
second half of the community sample. The initial results were mixed for the
community sample: Sattora-Bentler, .2/ D15:26; CFI D.97; NNFI D.91;
RMSEA D.140; and SRMR D.039. However, the model ﬁt improved signiﬁ-
cantly (p < :0001) and resulted in an excellent ﬁt by allowing the error variance
to correlate between the food manipulation and choking/gagging/vomiting sub-
scales: Sattora-Bentler, .1/ D1:25; CFI D1.00; NNFI D1.00; RMSEA D
.027; and SRMR D.010. All subscales served as statistically signiﬁcant (p <
:05) indicators of the MBQ total score, a latent variable reﬂecting various
typographies of mealtime behavior problems. Figure 1 presents the ﬁnal com-
pletely standardized factor solution and associated errors. Unstandardized factor
loadings, the covariance matrices, and additional details of the CFA can be
obtained from Kristoffer S. Berlin.
Reliability and prevalence rates. The internal consistencies of the MBQ
total score (sum of 31 items, excluding 2 items) and subscales were good, on
MEALTIME BEHAVIOR QUESTIONNAIRE 149
Items and Item Factor Loadings From Factor Pattern Matrix
Variable 1 2 3 4
24. Demanding alternative foods/forms of foods .665 .018 .058 .031
23. Eating too slowly .664 .201 .226 .031
22. Only eating a few foods .657 .123 .015 .076
17. Deal making (negotiation) .643 .208 .167 .118
16. Talking to keep from eating .618 .095 .044 .079
28. Verbally refusing to eat .601 .071 .146 .095
25. Playing with food .455 .380 .116 .054
4. Pushing spoon/food away .430 .306 .033 .104
33. Not sitting in chair .415 .275 .008 .046
21. Pushing away food from table .403 .309 .213 .101
3. Leaving the table .338 .007 .294 .018
27. Playing with toys rather than eating .268 .139 .212 .130
12. Letting food drop out of mouth .044 .712 .085 .000
5. Throwing food .242 .675 .190 .089
13. Spitting out food .186 .541 .078 .039
14. Hiding food .135 .516 .061 .005
1. Hands in front of face .109 .460 .036 .010
11. Spitting at a person .077 .404 .185 .094
2. Packing food in the mouth .203 .392 .088 .009
8. Screaming .079 .084 .63 4 .103
9. Hitting others or objects .141 .246 .569 .006
10. Kicking others or objects .042 .116 .563 .003
7. Crying .318 .035 .549 .012
18. Reporting physical pain .181 .189 .509 .110
6. Refusing to come to the table .288 .018 .43 6 .125
19. Asking for comfort or assurance .253 .157 .424 .238
20. Flailing arms/legs .135 .226 .408 .054
30. Biting others .279 .295 .361 .076
31. Gagging .132 .033 .171 .890
32. Vomiting .176 .016 .155 .716
26. Choking or coughing on food or liquid .000 .143 .039 .608
29. Biting self .193 .446 .165 .044
15. Hitting self .114 .415 .170 .029
Note. Items retained on factors are indicated in bold.
150 BERLIN ET AL.
FIGURE 1 Final measurement models with completely standardized estimates. Note.
MBQ DMealtime Behavior Questionnaire.
average (mean ˛D:83,SD D:06), and ranged from fair to excellent. Table 4
presents the total and subscale internal consistencies and descriptive statistics
for each sample (and combined community sample). Table 4 also presents the
percentage of respondents who endorsed an MBQ item at 3 or higher, reﬂecting
the prevalence of each mealtime behavior that occurred between “sometimes”
and “always” in the past week. The average prevalence rate across the 31 MBQ
items was 24.6% (SD D16:9%).
Validity. Construct validity was established through signiﬁcant correlations
(see Table 5) in the expected directions using the MBQ total score with es-
tablished measures (Cohen & Swerdlik, 1999). For solely descriptive purposes,
MBQ subscales have also been included in this table. The MBQ total score
was signiﬁcantly related to an established measure of family mealtime problems
(AYCE scales: CRE and PAM) and inversely to the PME rating and child age.
The goodness-of-ﬁt statistics present in the CFA section provided evidence for
structural and content validity (Messick, 1995).
Parent and child gender differences. Using an alpha of p < :017,
no effect of parent gender, F .5; 675/ D1:86,pD:10 (partial 2D:014);
child gender, F .5; 675/ D2:309,pD:043 (partial 2D:017); or the Parent
Gender Child Gender interaction, F .5; 675/ D1:18,pD:32 (partial 2D
:009), was found on caregiver report of child mealtime behavior problems.
MEALTIME BEHAVIOR QUESTIONNAIRE 151
MBQ Total Score, Subscale, and Item Descriptive Statistics and Internal Consistencies
for Clinical and Combined Community Sample
Variable M SD ˛
in Past Week
MBQ total score (31 items) 55.29 14.65 0.91 MD23:2%
Food refusal/avoidance 26.53 8.06 0.89 MD40:6%
24. Demanding alternative foods/forms of foods 2.26 1.14 41.2%
23. Eating too slowly 2.37 1.14 44.4%
22. Only eating a few foods 2.73 1.13 60.9%
17. Deal making (negotiation) 2.34 1.14 46.0%
16. Talking to keep from eating 1.90 1.01 30.4%
28. Verbally refusing to eat 2.04 1.01 34.0%
25. Playing with food 2.26 1.00 42.7%
4. Pushing spoon/food away 2.12 1.03 39.3%
33. Not sitting in chair 2.56 1.07 52.9%
21. Pushing away food from table 1.81 0.91 24.4%
3. Leaving the table 2.52 1.05 52.3%
27. Playing with toys rather than eating 1.62 0.89 17.6%
Food manipulation 10.78 3.38 0.73 MD15:9%
12. Letting food drop out of mouth 1.53 0.78 14.0%
5. Throwing food 1.36 0.73 8.8%
13. Spitting out food 1.60 0.78 16.3%
14. Hiding food 1.27 0.65 8.8%
1. Hands in front of face 1.83 0.93 25.8%
11. Spitting at a person 1.14 0.44 3.2%
2. Packing food in the mouth 2.04 1.01 34.5%
Mealtime aggression/distress 14.09 4.71 0.81 MD16:2%
8. Screaming 1.53 0.84 14.7%
9. Hitting others or objects 1.48 0.79 13.5%
10. Kicking others or objects 1.35 0.67 8.6%
7. Crying 1.88 0.94 27.2%
18. Reporting physical pain 1.63 0.92 17.3%
6. Refusing to come to the table 1.87 0.97 26.4%
19. Asking for comfort or assurance 1.87 1.00 27.4%
20. Flailing arms/legs 1.37 0.69 9.1%
30. Biting others 1.09 0.37 1.5%
Choking/gagging/vomiting 3.87 1.59 0.76 MD6:3%
31. Gagging 1.30 0.66 6.5%
32. Vomiting 1.14 0.49 2.8%
26. Choking or coughing on food or liquid 1.43 0.75 9.7%
15. Hitting self 1.12 0.41 2.2%
29. Biting self 1.05 0.29 1.0%
152 BERLIN ET AL.
MBQ Total Score and Subscale Correlations With the
About Your Child’s Eating Questionnaire
Score MBQ 1 MBQ 2 MBQ 3 MBQ 4
MBQ 1: Food refusal/avoidance 0.912 1.000
MBQ 2: Food manipulation 0.757 0.544 1.000
MBQ 3: Mealtime aggression/distress 0.834 0.628 0.557 1.000
MBQ 4: Choking/gagging/vomiting 0.495 0.318 0.425 0.350 1.000
Child age 0.187 0.104 0.268 0.156 0.157
Child resistance to eating 0.708 0.758 0.394 0.539 0.302
Positive mealtime environment 0.450 0.438 0.296 0.401 0.144
Parent aversion to mealtime 0.684 0.673 0.441 0.576 0.277
Note. All correlations are signiﬁcant at p < :001.
The MBQ was developed to provide researchers and clinicians with a validated
measure of the topographies of feeding difﬁculties in young children focused
solely on the frequency of child mealtime behaviors. Preliminary results of this
initial pilot study identiﬁed and validated a factor structure of the MBQ that
contained a total score and four subscales. These subscales reﬂect a variety of
problematic mealtime behaviors including food refusal/avoidance, food manipu-
lation, mealtime aggression/distress, and choking/gagging/vomiting. Overall, the
reliability of the MBQ and subscales was good and ranged from fair to excellent.
Preliminary analyses also suggest that the MBQ is valid in the assessment of
mealtime problems, as the MBQ relates (as anticipated) to the affective quality
of the mealtime environment, parental aversion to mealtimes, and parent report
of their child’s resistance to eating.
With regard to demographic inﬂuences, no effect of child or parent gender was
found on caregivers’ reports of their child’s mealtime behavior problems. Consis-
tent with previous research (Sanders, Patel, Grice, & Shepherd, 1993), younger
child age was associated with greater overall mealtime problems, food refusal
and avoidance, food manipulation, and mealtime aggression or distress. Child
age was not associated with parent report of choking, gagging, and vomiting.
With regard to the speciﬁc subscales, there is an interesting pattern of perfor-
mance for the choking/gagging/vomiting factor. This scale’s internal consistency
was good; however, it shows the weakest relations to other MBQ subscales and
to other measures of mealtime problems. One possibility for these ﬁnding is that
this scale may be tapping into feeding problems of a medical nature, more so
than behavioral feeding problems. For example, choking, gagging, and vomiting
MEALTIME BEHAVIOR QUESTIONNAIRE 153
can be signs of swallowing problems and aspiration (which are common in
children with feeding problems). This may account for the low correlations with
more behavioral measures. In light of this, it may be tempting to drop these
items for those wanting a more pure measure of behavioral feeding problems.
Substantial variability in the prevalence rates of feeding problems was ac-
counted for by the speciﬁc type of problem identiﬁed. The prevalence rates
of feeding problems across the MBQ subscales ranged from 6% to 41%, with
the most frequent types of mealtime behavior problems being food refusal and
avoidance. The single most common mealtime problem reported by parents
(60%) was that their child was “only eating a few foods.” These ﬁndings may
potentially suggest a developmental progression in the severity of pediatric
feeding problems that may begin with the relatively common food selectivity
and passive refusals, to more overt and aggressive types of behaviors.
This article contributes to the literature in a variety of novel ways. First, it
suggests that prevalence of mealtime problem rates vary according to the speciﬁc
behavioral topography and severity of the problem. Second, it provides a measure
of young children’s feeding difﬁculties that are not confounded by the caregiver’s
strategies for managing these problems. Third, this measure was developed and
validated using rigorous and advanced statistical techniques based on modern
quantitative recommendations (Fabrigar et al., 1999; Reise et al., 2000; Russell,
2002). No known measures of similar content have these advantages.
There are several important limitations of this study. These limitations include
the lack of racial diversity and the lack of data linking the MBQ to observed
meals. An additional limitation is that the sampling strategy used to collect
data did not provide estimations of participation rates. Therefore, it is unclear
if this sample was different from the large population in systematic ways. A
third limitation is that the eigenvalues from the EFA (e.g., amount of variance
explained from the factors) were rather low, suggesting higher than desired
error variance in the measure. Last, despite the many safeguards in place to
ensure data authenticity, a third limitation is that data may have been entered
from participants who were not parents of children aged 2 to 6 years old. The
coherence of the factor structure, the validity and reliability data, and the overall
pattern of results suggest that some preliminary conﬁdence can be placed on
these data. It will be critical, however, for future studies to replicate the factor
structure using CFAs with independent clinical and community samples and to
expand the evidence for validity (cross-cultural, ecological, etc.) and clinical
utility (e.g., the sensitivity and speciﬁcity).
Implications for Practice
The MBQ appears to be a promising measure of mealtime behavior problems in
young children. Given that this study represents a pilot investigation, great care
154 BERLIN ET AL.
must be taken to decide when or if to use this measure in practice. If used, the
MBQ could contribute to a multimethod, multidisciplinary assessment of young
children’s feeding problems, which could augment interview and observation
data by providing caregiver report information on potentially low base-rate
behaviors and a snapshot of a general 1-week pattern of behavior. This measure
could also be used prior to intake to help identify speciﬁc behaviors that would
require further assessment during a clinic visit. The descriptive data of the
MBQ can also be used to calculate zor tscores so that an individual case
presenting for treatment can be compared to a nonclinical sample to determine
“how discrepant” their factor scores are from the norms. A last potential use
would be to track clinical outcomes by re-administering the MBQ over the
course of treatment. Additional studies, however, are needed to further explicate
the clinical utility of the MBQ.
We thank the members of the Milwaukee Area Child Health Research Initiative
for their comments on an earlier version of this article.
Archer, L. A., Rosenbaum, P. L., & Streiner, D. L. (1991). The Children’s Eating Behavior Inventory:
Reliability and validity results. Journal of Pediatric Psychology, 16, 629–642.
Bentovim, A. (1970). The clinical approach to feeding disorders of childhood. Journal of Psycho-
somatic Research, 14, 267–276.
Bollen, K. A. (1989). Structural equations with latent variables. Oxford, UK: John Wiley and Sons.
Browne, M. W., & Cudeck, R. (1993). Alternative ways of assessing model ﬁt. In K. A. Bollen &
J. S. Long (Eds.), Testing structural equation models (pp. 136–161). Newbury Park, CA: Sage.
Budd, K. S., McGraw, T. E., Farbisz, R., Murphy, T. B., Hawkins, D., Heilman, N., et al. (1992).
Psychosocial concomitants of children’s feeding disorders. Journal of Pediatric Psychology, 17,
Christopherson, E. R., & Hall, C. L. (1978). Eating patterns and associated problems encountered
in normal children. Issues in Comprehensive Pediatric Nursing, 3, 1–16.
Cohen, R. J., & Swerdlik, M. E. (1999). Psychological testing and assessment: An introduction to
tests and measurement (4th ed.). Mountain View, CA: Mayﬁeld.
Crist, W., & Napier-Phillips, A. (2001). Mealtime behaviors of young children: A comparison of
normative and clinical data. Journal of Developmental and Behavioral Pediatrics, 22, 279–286.
Davies, W. H., Ackerman, L. K., Davies, C. M., Vannatta, K., & Noll, R. B. (2007). About your
child’s eating: Factor structure and psychometric properties of a feeding relationship measure.
Eating Behaviors, 8, 457–463.
Fabrigar, L. R., Wegener, D. T., MacCallum, R. C., & Strahan, E. J. (1999). Evaluating the use of
exploratory factor analysis in psychological research. Psychological Methods, 4, 272–299.
Finney, J. W. (1986). Preventing common feeding problems in infants and young children. Pediatric
Clinics of North America, 33, 775–788.
MEALTIME BEHAVIOR QUESTIONNAIRE 155
Hoyle, R. H., & Panter, A. T. (1995). Writing about structural equation modeling. In R. H. Hoyle
(Ed.), Structural equation modeling: Concepts, issues, and applications (pp. 158–176). Thousand
Oaks, CA: Sage.
Hu, L., & Bentler, P. M. (1995). Evaluating model ﬁt. In R. H. Hoyle (Ed.), Structural equation
modeling: Concepts, issues, and applications (pp. 76–99). Thousand Oaks, CA: Sage.
Hu, L., & Bentler, P. M. (1999). Cutoff criteria for ﬁt indexes in covariance structure analysis:
Conventional criteria versus new alternatives. Structural Equation Modeling, 6, 1–55.
Jöreskog, K. G., & Sörbom, D. (2003). LISREL 8.54: User’s reference guide. Chicago: Scientiﬁc
Kaiser, H. F. (1970). A second generation Little Jiffy. Psychometrika, 35, 401–415.
Kerwin, M. E. (1999). Empirically supported treatment in pediatric psychology: Severe feeding
problems. Journal of Pediatric Psychology, 24, 193–214.
Linscheid, T. (1978). Disturbances of eating and feeding. In P. R. Magrab (Ed.), Psychological
management of pediatric problems: Early life conditions and chronic disease (Vol. 1, pp. 191–
218). Baltimore: University Park Press.
Linscheid T. (1992). Eating problems in children. In C. E. Walker & M. C. Roberts (Eds.), Handbook
of clinical child psychology (pp. 451–473). New York: Wiley.
MacCullum, R. C., Browne, M. W., & Sugawara, H. M. (1996). Power analysis and determination
of sample size for covariance structure modeling. Psychological Methods, 1, 130–149.
Mackner, L. M., McGrath, A., & Stark, L. J. (2001). Dietary recommendations to prevent and manage
chronic pediatric health conditions: Adherence, intervention, and future directions. Journal of
Developmental and Behavioral Pediatrics, 22, 130–143.
Manikam, R., & Perman, J. A. (2000). Pediatric feeding disorders. Journal of Clinical Gastroen-
terology, 30, 34–46.
Messick, S. (1995). Validity of psychological assessment: Validation of inferences from persons’
responses and performances as scientiﬁc inquiry into score meaning. American Psychologist, 50,
Reise, S. P., Waller, N. G., & Comrey, A. L. (2000). Factor analysis and scale revision. Psychological
Assessment, 12, 287–297.
Riordan, M. M., Iwata, B. A., Wohl, M. K., & Finney, J. W. (1980). Behavioral treatment of
food refusal and selectivity in developmentally disabled children. Applied Research in Mental
Retardation, 1, 95–112.
Russell, D. W. (2002). In search of underlying dimensions: The use (and abuse) of factor analysis
in personality and social psychology bulletin. Personality and Social Psychology Bulletin, 28,
Sanders, M. R., Patel, R. K., LeGrice, B., & Shepherd, R. W. (1993). Children with persistent
feeding difﬁculties: An observational analysis of the feeding interactions of problem and non-
problem eaters. Health Psychology, 12, 64–73.
Sass, D. A., & Smith, P. L. (2006). The effects of parceling unidimensional scales on structural
parameter estimates in structural equation modeling. Structural Equation Modeling: A Multidis-
ciplinary Journal, 13, 566–586.
Satorra, A., & Bentler, P. M. (1994). Corrections to test statistics and standard errors in covariance
structure analysis. In A. von Eye & C. C. Clogg (Eds.), Latent variable analysis: Applications
for developmental research (pp. 399–419). Thousand Oaks, CA: Sage.
Sisson, L. A., & Van Hasselt, V. B. (1989). Feeding disorders. In J. K. Luiselli (Ed.), Behavioral
medicine and developmental disabilities (pp. 45–75). New York: Springer-Verlag.
Skuse, D. (1993). Epidemiological and deﬁnitional issues in failure to thrive. Child and Adolescent
Psychiatric Clinics of North America: Eating and Growth Disorders, 2, 37–59.
Watkins, M. A. (2000). MacParallel Analysis [Computer software]. Available from http://www.
156 BERLIN ET AL.
West, S. G., Finch, J. F., & Curran, P. J. (1995). Structural equation models with nonnormal variables:
Problems and remedies. In R. H. Hoyle (Ed.), Structural equation modeling: Concepts, issues,
and applications (pp. 56–75). Newbury Park, CA: Sage.
Whitten, C. F., Pettit, M. G., & Fischhoff, J. (1969). Evidence that growth failure from maternal de-
privation is seco ndary to undereating. Journal of American Medical Association, 209, 1675–1682.
Mealtime Behavior Questionnaire
Please rate each behavior in terms of how often or frequently it happens. Please
rate each behavior as it occurred during mealtimes or feeding over the past week.
Never Sometimes Always
1. Hands in front of face 1 2 3 4 5
2. Packing food in the mouth 1 2 3 4 5
3. Leaving the table 1 2 3 4 5
4. Pushing spoon/food away 1 2 3 4 5
5. Throwing food 1 2 3 4 5
6. Refusing to come to the table 1 2 3 4 5
7. Crying 1 2 3 4 5
8. Screaming 1 2 3 4 5
9. Hitting others or objects 1 2 3 4 5
10. Kicking others or objects 1 2 3 4 5
11. Spitting at a person 1 2 3 4 5
12. Letting food drop out of mouth 1 2 3 4 5
13. Spitting out food 1 2 3 4 5
14. Hiding food 1 2 3 4 5
15. Hitting self 1 2 3 4 5
16. Talking to keep from eating 1 2 3 4 5
17. Deal making (negotiation) 1 2 3 4 5
18. Reporting physical pain 1 2 3 4 5
19. Asking for comfort or assurance 1 2 3 4 5
20. Flailing arms/legs 1 2 3 4 5
21. Pushing away food from table 1 2 3 4 5
22. Only eating a few foods 1 2 3 4 5
23. Eating too slowly 1 2 3 4 5
24. Demanding alternative foods/forms of foods 1 2 3 4 5
25. Playing with food 1 2 3 4 5
26. Choking or coughing on food or liquid 1 2 3 4 5
27. Playing with toys rather than eating 1 2 3 4 5
28. Verbally refusing to eat 1 2 3 4 5
29. Biting self 1 2 3 4 5
30. Biting others 1 2 3 4 5
31. Gagging 1 2 3 4 5
32. Vomiting 1 2 3 4 5
33. Not sitting in chair 1 2 3 4 5
Copyright of Children's Health Care is the property of Taylor & Francis Ltd and its content may not be copied
or emailed to multiple sites or posted to a listserv without the copyright holder's express written permission.
However, users may print, download, or email articles for individual use.