issues in comprehensive pediatric nursingthomas/research/publications/nursing child... ·...

This article was downloaded by:[University of North Carolina]On: 17 October 2007Access Details: [subscription number 768500318]Publisher: Informa HealthcareInforma Ltd Registered in England and Wales Registered Number: 1072954Registered office: Mortimer House, 37-41 Mortimer Street, London W1T 3JH, UK

Issues in Comprehensive PediatricNursingPublication details, including instructions for authors and subscription information:http://www.informaworld.com/smpp/title~content=t713630242

Reliability of the Nursing Child Assessment FeedingScale During ToddlerhoodEric A. Hodges a; Gail M. Houck b; Thomas Kindermann ca Baylor College of Medicine, Houston, Texas, USAb Oregon Health & Science University, Portland, Oregon, USAc Portland State University, Portland, Oregon, USA

Online Publication Date: 01 July 2007To cite this Article: Hodges, Eric A., Houck, Gail M. and Kindermann, Thomas(2007) 'Reliability of the Nursing Child Assessment Feeding Scale DuringToddlerhood', Issues in Comprehensive Pediatric Nursing, 30:3, 109 - 130

To link to this article: DOI: 10.1080/01460860701525204URL: http://dx.doi.org/10.1080/01460860701525204

PLEASE SCROLL DOWN FOR ARTICLE

Full terms and conditions of use: http://www.informaworld.com/terms-and-conditions-of-access.pdf

This article maybe used for research, teaching and private study purposes. Any substantial or systematic reproduction,re-distribution, re-selling, loan or sub-licensing, systematic supply or distribution in any form to anyone is expresslyforbidden.

The publisher does not give any warranty express or implied or make any representation that the contents will becomplete or accurate or up to date. The accuracy of any instructions, formulae and drug doses should beindependently verified with primary sources. The publisher shall not be liable for any loss, actions, claims, proceedings,demand or costs or damages whatsoever or howsoever caused arising directly or indirectly in connection with orarising out of the use of this material.

http://www.informaworld.com/smpp/title~content=t713630242

http://dx.doi.org/10.1080/01460860701525204

http://www.informaworld.com/terms-and-conditions-of-access.pdf

Dow

nloa

ded

By:

[Uni

vers

ity o

f Nor

th C

arol

ina]

At:

15:0

0 17

Oct

ober

200

7

Issues in Comprehensive Pediatric Nursing, 30:109–130, 2007Copyright © Informa Healthcare USA, Inc.ISSN: 0146-0862 print / 1521-043X onlineDOI: 10.1080/01460860701525204

109

UCPN0146-08621521-043XIssues in Comprehensive Pediatric Nursing, Vol. 30, No. 3, July 2007: pp. 1–37Issues in Comprehensive Pediatric Nursing

RELIABILITY OF THE NURSING CHILD ASSESSMENT FEEDING SCALE DURING TODDLERHOOD

NCAFS Reliability During ToddlerhoodE. A. Hodges et al. Eric A. Hodges

Baylor College of Medicine,Houston, Texas, USA

Gail M. Houck

Oregon Health & Science University,Portland, Oregon, USA

Thomas Kindermann

Portland State University,Portland, Oregon, USA

The quality of the maternal-child feeding interaction has been proposedto be an important contributor to a child’s being overweight, yet assess-ment of this proposition has been hindered by a lack of age-appropriateinstrumentation. The primary aim of this study was to examine the reli-ability of the Nursing Child Assessment of Feeding Scale (NCAFS) ifextended to use during toddlerhood. A longitudinal design was used toassess NCAFS reliability at 12, 24, and 36 months. The NCAFS wasused to code videotaped feeding observations of 116 mother–toddlerdyads collected as part of a larger study examining mother–child interac-tions and adaptations of toddlers. Reliability was explored through theassessment of interrater reliability, internal consistency of the varioussubscales and the scale as a whole, and stability of the scale measure-ments over time. At each age, interrater reliability was generally quite goodwhereas the NCAFS’ internal consistency was low. Maternal contributions

Received 12 May 2007; 14 June 2007.Address correspondence to Eric A. Hodges, PhD, Baylor College of Medicine, USDA/ARS Children’s

Nutrition Research Center, 1100 Bates Street, Houston, TX 77030. E-mail: [email protected]

Dow

nloa

ded

By:

[Uni

vers

ity o

f Nor

th C

arol

ina]

At:

15:0

0 17

Oct

ober

200

7

110 E. A. Hodges et al.

to feeding interaction quality were stable over time but dyadic and childcontributions were not. The lower internal consistency estimates werelikely due to relatively low levels of variance among the dyads withineach age. Another probable contributor to diminished internal consis-tency was the attrition of several behavior items due to zero variance. Pos-sible explanations for this restriction of variance and several approachesfor improving the NCAFS internal consistency during toddlerhood areconsidered. With revision, the NCAFS could be useful in assessment offeeding interaction quality during the transition to toddlerhood whenissues of control and autonomy become increasingly prominent.

Helping young children establish healthy feeding patterns that will sup-port ongoing growth and development is a primary responsibility ofparenting. Unfortunately, significant numbers of children develop feedingproblems that lead to either undernutrition—primarily failure to thrive[FTT] (Benoit, 2000), or overnutrition—obesity (Kimbro, Brooks-Gunn, &McLanahan, 2007; Mei et al., 1998; Ogden et al., 1997). Although suchnutrition problems are responses to a combination of genetic, environ-mental, and interpersonal factors, early intervention and prevention ofthese problems requires focus on the quality of the parent/child feedinginteraction (Brewis & Gartin, 2006; Faith, Scanlon, Birch, Francis, &Sherry, 2004; Satter, 1990). The quality of feeding interactions during thetransition from relative dependence in infancy to emerging autonomy intoddlerhood has been proposed to contribute to the child’s ability to self-regulate feeding/eating that will support optimal growth and development(Birch & Fisher, 1998; Bruch, 1973; Costanzo & Woody, 1985).

While we cannot modify a child’s genetic inheritance, we can poten-tially intervene in parenting to optimize feeding environments and inter-actions. Observational assessments of interaction quality have revealedthat mothers of children with FTT were less sensitive (Hagekull, Bohlin,& Rydell, 1997), less responsive to their infant’s cues, and more interfer-ing and controlling (Chatoor, Egan, Getson, Menvielle, & O’Donnell,1987; Chatoor, Hirsch, Ganiban, Persinger, & Hamburger, 1998;Crittenden, 1987; Lucarelli, Ambruzzi, Cimino, D’Olimpio, & Finistrella,2003). Interest in the contributions of parent–child interaction quality tochildhood obesity has also grown (Faith et al., 2004), yet the contributionof interactions during infancy and toddlerhood have received relativelylittle attention to date. In a recent observational study, those infants withgreater weight gain during the first 6 months of life whose mothers exhib-ited greater feeding control (restriction of intake or pressure to eat) at 6months demonstrated greater weight gains from 6–12 months comparedto infants with less controlling mothers (Farrow & Blissett, 2006).

Dow

nloa

ded

By:

[Uni

vers

ity o

f Nor

th C

arol

ina]

At:

15:0

0 17

Oct

ober

200

7

NCAFS Reliability During Toddlerhood 111

A strong research base for understanding feeding interaction quality isessential for improving diagnostic, prevention, and intervention capabilitiesin a variety of pediatric practice settings and across an age range thatencompasses the child’s developmental a transition from relative depen-dence to independence in feeding. Research and research-based interven-tions in feeding interactions are limited by the lack of assessments offeeding/eating beyond the first year of life. The Nursing Child AssessmentFeeding Scale (NCAFS) (Sumner & Spietz, 1994) is an observationalchecklist used to assess the feeding interaction quality between caregiverand infant, but it was designed only for infants up to 12 months of age. Ithad not been established as valid for toddlerhood when parents mustaccommodate their children’s emerging autonomy and when control issuesbecome more salient. Other assessment tools are either focused on interac-tion involving children with feeding disorders (Chatoor et al., 1997) or havenot been widely disseminated (Spegman & Houck, 2005).

The purpose of this study was to examine the possible extension of theNCAFS to toddlerhood. It is well established and already used by over21,000 certified coders in research and clinical settings worldwide(D. Findlay, personal communication, November 12, 2002). Thus, the pri-mary aim of this paper is to report the reliability of the NCAFS duringtoddlerhood in order to enhance its further development and refinementfor potential use in research and early interventions with feeding/eatinginteractions and their outcomes.

METHODS

The research entailed an analysis of existing data collected in a larger lon-gitudinal, observational study that assessed mother–child interactions inwhich control and autonomy were important features at 12, 24, and 36months (Houck, 1999). The study was approved by the institutionalreview board of Oregon Health & Science University. Mothers providedconsent for themselves and their infants prior to entry into the study.

Sample

The original sample consisted of 162 mother–infant dyads recruited froma Family Practice Clinic and by word of mouth prior to the infant’s age of8 months. Complete self-report and feeding/eating observational data at12, 24, and 36 months were available for 116 of the original 162 mother–toddler dyads. All initial demographic data were collected via maternalself-report using a questionnaire at intake when the children were

Dow

nloa

ded

By:

[Uni

vers

ity o

f Nor

th C

arol

ina]

At:

15:0

0 17

Oct

ober

200

7


8 months old, with updates collected at each subsequent visit. Motherswere primarily Caucasian (n = 91); twenty-one mothers (18.1%) wereAfrican American and four were of other ethnicity. At baseline (8 monthsof age), mothers ranged in age from 17 to 47 years, with a mean of 28.3years (SD = 7.5). Nearly three fourths (n = 84) of the mothers were atleast high school graduates, of which 17.2% (n = 20) were college graduates.Fifty-four percent of the mothers (n = 63) were employed and by36 months, this proportion had risen to 60.3% (n = 70). Two thirds of themothers (63.8%; n = 74) were married, and 11 of the mothers (9.5%) wereeither living with a partner or in common-law marriages. The remainderhad either never married or were separated or divorced. Over one half ofmothers (50.9%; n = 59) reported income less than $17,000, one fifth(20.7%; n = 24) between $18,000 and $28,000, and the remainder (28.4%;n = 33) reported income greater than $29,000.

There were proportionately more male (65.5%, n = 76) than female(34.5%, n = 40) children in this sample. For nearly one half of mothers(n = 56), the child was their first, and most of the children (87.9%; n = 102)were full term. Only five reported some early difficulty with feeding dur-ing the first few months of life, including difficulty breastfeeding, diffi-culty adjusting to formula, or vomiting. Infants and their mothers wereexcluded from the study if they had a diagnosed physical or cognitive dis-ability. The majority of mothers reported themselves to be the primarycaregivers for their children at intake and throughout the study period(range: 80.2–92.2%). Most children were not in daycare over the courseof the study. More children were enrolled in daycare at 24 (42%, n = 49)and 36 (39%, n = 45) months compared to 12 months (30%, n = 35).

Observational Setting

Mothers and their children were videotaped through a one-way mirror in alaboratory setting. At 12 months of age, all children were placed in a highchair and the snack was brought in for the mother to facilitate feeding thechild as she normally would. At 24 and 36 months, a child-sized table andchairs were made available for the dyad and the snack tray was placed onthe table between them. There were no toys or other play items availableto the child during snack time. Instructions to mothers at every age ofobservation were that the snack would serve as a break for mother andchild, but the camera would continue recording. This setting is consistentwith those typically used for feeding observations, including NCAFS.The feeding session ended when the mother said so or after 10 minutes.The decision to end the recording at 10 minutes precluded observation of

Dow

nloa

ded

By:

[Uni

vers

ity o

f Nor

th C

arol

ina]

At:

15:0

0 17

Oct

ober

200

7


the natural termination of feeding for the majority of dyads at each age(12 months = 89 dyads; 24 months = 64 dyads; 36 months = 76 dyads).

Observational Measures

The Nursing Child Assessment Feeding Scale

The NCAFS (Sumner & Spietz, 1994) was used to assess the videotapedobservations of feeding between mothers and their infants at 12, 24, and36 months. The scale is currently considered reliable and valid for infantsfrom birth to one year. The NCAFS is an observational measure of inter-action quality between caregiver and child during feeding. The assess-ment can take place in any setting and its duration is determined by thenatural initiation and termination of feeding.

The NCAFS consists of 76 binary items that are scored as presence orabsence of behavior. These 76 items are organized into six subscales thatwere conceptually derived through a “combination of intuitive and experi-enced judgment” of a collaborative group of nurses familiar with childrenand parents and psychologists with developmental psychology and observa-tional methodology backgrounds (Barnard et al., 1989). The four caregiversubscales include: I) Sensitivity to Cues, II) Response to the Child’s Dis-tress, III) Social-Emotional Growth-Fostering, and IV) Cognitive GrowthFostering. There are two child subscales: V) Clarity of Cues and VI)Responsiveness to Caregiver. Scores for each of the subscales could rangefrom zero, indicating that none of the item behaviors were observed, to thetotal number of items in each subscale, indicating that all item behaviors inthe subscale were observed in the interaction. The four caregiver subscalescores combine to create a caregiver total with a possible maximum scoreof 50 and the two child subscales scores combine to create a child total witha possible maximum score of 26. These two scores combine to create adyadic total with a maximum score of 76. A high score represents moreoptimal interaction between caregiver and child. Cronbach’s alphas for thecaregiver, child, and dyadic total scores are reported as .83, .73, and .86respectively (Sumner & Spietz, 1994).

Procedures for Coding

All 116 videotaped observations were used for 12, 24, and 36 months.The first author was trained to accuracy on the NCAFS coding by a certi-fied NCAST instructor. A different NCAST instructor was hired to estab-lish interrater reliability and code half of the observations. For each ageperiod, the coders individually coded 58 tapes and cross-coded 9 tapes.Each tape was assigned an identification number and contained the

Dow

nloa

ded

By:

[Uni

vers

ity o

f Nor

th C

arol

ina]

At:

15:0

0 17

Oct

ober

200

7


observations for a single dyad. The 116 tapes were divided into two groupsaccording to numeric descending order. Of all tapes for 12-month-oldinfants, the researcher coded the first 58 tapes and the consultant coded theremaining 58 tapes. This coding order was reversed for coding the tapes ofthe 24-month-old infants, with the consultant coding the first 58 tapes, andreversed back to the original order for coding the tapes of the 36-month-oldinfants. After an initial 19 tapes were coded, the researcher and consultantdrew random numbers to determine which three tapes of the initial 19 wouldbe cross-coded to assess interrater reliability. This procedure was repeatedafter the next 19 tapes and after the final 20 tapes for each observationperiod. Meetings were held after each cross-coding (nine occasions) todetermine percentage agreement and to discuss items about which disagree-ment occurred until consensus was reached regarding an item’s score. Theconsensus coded sheets for dyads assessed by both coders were entered intothe database for analysis along with the individually coded dyads.

Analysis

Reliability was explored through the assessment of interrater reliability, inter-nal consistency of the various subscales and the scale as a whole, and stabilityof the scale measurements over time. Interrater reliability was assessedthrough percent agreement, Cohen’s kappa, and interrater correlations. Pairedsample t-tests to assess for differences between raters’ total scores for eachobservation period were also conducted. The internal consistency of theNCAFS individual subscales, as well as the broader dimensions of caregiver,child, and caregiver–child scales were assessed via the Kuder-Richardson for-mula (KR-20) (Kuder & Richardson, 1937) for 12-, 24-, and 36-month obser-vations. Equality of variance was tested via Chi-square between the studysample and the normative sample (Sumner & Spietz, 1994). Pearson’s R cor-relations were used to assess stability/test–retest of chosen NCAFS subscalescores (caregiver total, child total, and dyad total). All analyses were con-ducted using SPSS Graduate Pack Version 13.0 (Chicago, Illinois).

RESULTS

Interrater Agreement and Reliability for NCAFS at 12, 24, and 36 Months

Average percent agreement for each observation period was greater thanninety percent (Table 1). Interrater reliability assessment via Cohen’skappa was performed at the NCAFS subscale level (Table 2). The raters’

Dow

nloa

ded

By:

[Uni

vers

ity o

f Nor

th C

arol

ina]

At:

15:0

0 17

Oct

ober

200

7


total scores were significantly positively correlated at 12, 24, and 36months (r = .64, p < .01; r = .61, p < .01; and r = .75, p < .001, respec-tively). There were no significant differences between raters’ total scoresat any of the age periods (12 months, t(17) = 1.20; 24 months, t(17) = .28;36 months, t(17) = .62).

Internal Consistency of the NCAFS at 12, 24, and 36 Months

Table 3 provides the means and standard deviations for the individualsubscales, caregiver, child, and caregiver–child total scores at 12, 24, and36 months, as well as for the NCAFS normative sample at 12 months(Sumner & Spietz, 1994). Since there are no normative data at 24 and 36months, 12-month normative statistics were used in comparisons at eachtime point. Z scores were calculated to assess similarity between the studysample and the normative sample on caregiver, child, and caregiver–childtotal scores (Table 4). At each age, the study sample mean scores weresignificantly different from those of the normative sample, with the study

Table 1. NCAFS interrater reliability percent agreement

Observation Period n range (%) average (%)

12 months 18 82.9–97.4 90.8624 months 18 82.9–96.1 91.1636 months 18 90.8–100 94.57

Note: Overall average agreement = 92.2%.

Table 2. NCAFS interrater reliability assessed via Cohen’s kappa

Subscale12 mos

k (n = 18)24 mos

k (n = 18)36 mos

k (n = 18)

I. Sensitivity to cues .80 .75 .79II. Response to child’s distress .73 .79 .88III. Social–emotional growth fostering .65 .68 .83IV. Cognitive growth fostering .53 .85 1V. Clarity of cues .62 .66 .79VI. Responsiveness to caregiver .80 .75 .94

Note: Mean overall NCAFS kappa: k (12 months) = .69; k (24 months) = .75; k (36 months) = .86.

Dow

nloa

ded

By:

[Uni

vers

ity o

f Nor

th C

arol

ina]

At:

15:0

0 17

Oct

ober

200

7

116

Tab

le 3

.N

CA

FS:

Des

crip

tive

dat

a at

12,

24,

and

36

mon

ths

(N =

116

)

12 m

os24

mos

36 m

os

Scal

e or

Sub

scal

eP

ossi

ble

Ran

geA

ctua

lR

ange

MSD

Act

ual

Ran

geM

SDA

ctua

lR

ange

MSD

Nor

m*

Mea

nN

orm

*SD

I. S

ensi

tivity

to c

ues

0–16

6–15

11.3

41.

536–

1411

.05

1.65

8–15

11.0

81.

2213

.81

1.85

II. R

espo

nse

to c

hild

’s d

istr

ess

0–11

5–11

8.16

1.53

6–11

8.09

.89

6–11

8.22

.97

10.0

61.

31II

I. S

ocia

l–em

otio

nal g

row

th f

oste

ring

0–14

8–14

11.8

41.

217–

1411

.78

1.36

8–14

11.8

01.

6712

.25

1.78

IV. C

ogni

tive

grow

th f

oste

ring

0–9

7–9

8.63

.55

8–9

8.85

.36

7–9

8.91

.39

7.18

1.79

V. C

areg

iver

tota

l0–

5030

–47

39.8

13.

0534

–45

39.7

82.

6329

–46

40.0

22.

3843

.30

4.87

VI.

Cla

rity

of

cues

0–15

8–15

12.0

51.

469–

1512

.35

1.26

9–15

12.0

91.

3612

.93

1.90

VI.

Res

pons

iven

ess

to c

areg

iver

0–11

4–10

6.99

1.12

5–9

7.16

1.11

4–9

7.22

1.01

8.06

1.93

Chi

ld to

tal

0–26

14–2

419

.04

2.03

15–2

319

.52

1.87

14–2

319

.32

1.82

20.9

93.

40D

yad

tota

l0–

7649

–68

58.8

54.

0251

–66

59.2

93.

5648

–68

59.3

43.

1664

.29

7.27

Not

e: *

N =

791

, all

mot

hers

are

adu

lts a

ged

20–4

5 ye

ars;

chi

ldre

n 1–

12 m

onth

s of

age

. Sam

ple

is C

auca

sian

, con

trol

ling

for

mat

erna

l edu

catio

n an

d ch

ild a

ge.

Dow

nloa

ded

By:

[Uni

vers

ity o

f Nor

th C

arol

ina]

At:

15:0

0 17

Oct

ober

200

7


sample having lower mean scores. At each age, the study sample wasfound to have a significantly restricted variance compared to the norma-tive sample for caregiver, child, and caregiver–child total scores (Table 5).

The restriction in variance suggested that this study’s sample waslikely more homogenous than that of the normative sample. Therefore, tofurther explore similarities and differences between the samples, analyses

Table 4. NCAFS sample and normative data comparison: Z scores

Score 12 mos 24 mos 36 mos

Caregiver totalSamplea M(SD) 39.81 (3.05) 39.78 (2.63) 40.42 (2.38)Normb M(SD) 43.30 (4.87) 43.30 (4.87) 43.30 (4.87)Z Score −7.72*** −7.79*** −7.26***

Child totalSample M(SD) 19.04 (2.03) 19.52 (1.87) 19.32 (1.82)*Norm M(SD) 20.99 (3.40) 20.99 (3.40) 20.99 (3.40)Z Score −6.17*** −4.67*** −5.29***

Caregiver–child totalSample M(SD) 58.85 (4.02) 59.29 (3.56) 59.34 (3.16)*Norm M(SD) 64.29 (7.27) 64.29 (7.27) 64.29 (7.27)Z Score −8.05*** −7.40*** −7.34***

Note: aN = 116. bN = 791, all mothers are adults aged 20–45 years; children 1–12 months of age.Sample is Caucasian, controlling for maternal education and child age. Mean and SD values from nor-mative sample were used consistently for comparison at 12, 24, and 36 months.

***p < .001.

Table 5. Test of equality of variance between NCAFS studysample and normative sample

Scale or Subscale 12 mos 24 mos 36 mos

Caregiver totalSample SD 3.05 2.63 2.38Norm SD 4.87 4.87 4.87χ2 (df = 115) 45.04** 33.65** 27.49**

Child totalSample SD 2.03 1.87 1.82Norm SD 3.40 3.40 3.40χ2 (df = 115) 40.79** 34.72** 32.83**

Caregiver–Child TotalSample SD 4.02 3.56 3.16Norm SD 7.27 7.27 7.27χ2 (df = 115) 35.16** 27.59** 21.76**

**p < .01 (Critical values for χ2 (df = 100) = 70.07–135.81).

Dow

nloa

ded

By:

[Uni

vers

ity o

f Nor

th C

arol

ina]

At:

15:0

0 17

Oct

ober

200

7


of mean score differences for specific subgroups of subjects were per-formed at 12 months (see Tables 6 and 7). Caucasian dyads in whichmothers were classified as adolescents, low-education adults, or high-education adults did not significantly differ from similarly classifiednormative sample dyads in regard to caregiver, child, and caregiver–childtotal scores. African-American dyads classified according to the sameeducation categories also did not significantly differ from similarly classifiednormative sample dyads in regard to caregiver, child, and caregiver–childtotal scores except for caregiver–child total scores among high-educationadults (z = −2.01, p < .05). Thus, at the subgroup level, the study’s sampleand the normative sample were more similar, though given the smallsample size in these analyses the results must be interpreted with somecaution.

Internal Consistency

Alpha estimates obtained were well below conventional levels of accept-ability (typically .70) (Schmitt, 1996) and below those reported in theNCAFS manual (Table 8). Items from the majority of subscales weredropped out of the analysis due to zero variance. At the level of caregiver,child, and total scales, this item attrition generally tended to worsen withincreasing infant age. When individual items were included only if they

Table 6. 12-month NCAFS caucasian sub-sample and normative datacomparison: Z scores

Score Adolescent (n = 5) Low Ed. Adult (n = 8) High Ed. Adult (n = 24)

Caregiver totalSample M(SD) 39.40 (3.13) 40.38 (2.56) 39.04 (2.88)Norma M(SD) 37.76 (6.87) 38.40 (7.59) 41.18 (5.95)Z Score .53 .74 −1.77

Child totalSample M (SD) 19.20 (2.59) 19.00 (2.07) 18.88 (1.57)Normb M(SD) 18.95 (3.87) 19.82 (3.88) 20.20 (3.86)Z Score .14 .32 −1.67

Caregiver–child totalSample M (SD) 58.60 (5.03) 59.38 (4.10) 57.92 (3.46)Normc M(SD) 56.71 (9.46) 58.22 (10.12) 61.38 (8.74)Z Score .45 .32 −1.94

Note: aN = 236, adolescents are 13–18 years old, children are 1–12 months old. bN = 125, low-education adults are 19–25 years old with less than 12 years of education. cN = 430, high-educationadults are 19–25 years old with 12 or more years of education. Children are 1–12 months of age foreach of the normative groups.

Dow

nloa

ded

By:

[Uni

vers

ity o

f Nor

th C

arol

ina]

At:

15:0

0 17

Oct

ober

200

7


Table 7. 12-month NCAFS African-American sub-sample and normativedata comparison

Z Scores

Score Adolescent (n = 1) Low Ed. Adult (n = 2) High Ed. Adult (n = 10)

Caregiver totalSample M(SD) 38.00 (0.00) 36.00 (8.49) 37.70 (1.89)Norma M(SD) 37.76 (6.87) 38.40 (7.59) 41.18 (5.95)Z Score .03 −.45 −1.85

Child totalSample M(SD) 19.00 (0.00) 18.00 (1.41) 18.10 (2.23)Normb M(SD) 18.95 (3.87) 19.82 (3.88) 20.20 (3.86)Z Score .01 −.66 −1.72

Caregiver–child totalSample M(SD) 57.00 (0.00) 54.00 (7.07) 55.80 (3.26)Normc M(SD) 56.71 (9.46) 58.22 (10.12) 61.38 (8.74)Z Score .11 −.59 −2.01*

Note: aN = 236, adolescents are 13–18 years old, children are 1–12 months old. bN = 125, low-education adults are 19–25 years old with less than 12 years of education. cN = 430, high-educationadults are 19–25 years old with 12 or more years of education. Children are 1–12 months of age foreach of the normative groups.

*p < .05.

Table 8. Internal consistency of NCAFS (N = 116)

12 Months

K-R 20 (*)

24 Months

K-R 20 (*)

36 Months

K-R 20 (*)

NCAFS normativesample

aa (†)Subscales

I. Sensitivity to cues .36 (11) .31 (13) −.02 (13) .60 (16)II. Response to child’s distress .43 (11) .01 (10) .41 (8) .69 (11)III. Social–emotional growth fostering .31 (10) .41 (9) .22 (10) .63 (14)IV. Cognitive growth fostering .11 (4) −.06 (3) .58 (4) .69 (9)V. Clarity of cues .43 (7) .28 (9) .37 (9) .56 (15)VI. Responsiveness to caregiver .09 (11) .15 (8) .21 (4) .58 (11)Caregiver .57 (36) .44 (35) .43 (35) .83 (50)Child .41 (18) .37 (17) .40 (16) .73 (26)Total .60 (54) .52 (52) .47 (51) .86 (76)

Note: *number of items remaining in subscale/scale after items with zero variance removed.aalpha estimates for NCAFS subscales reported for the normative sample in the NCAFS manual

(Sumner & Spietz, 1994).†original number of items in subscales/scales.

Dow

nloa

ded

By:

[Uni

vers

ity o

f Nor

th C

arol

ina]

At:

15:0

0 17

Oct

ober

200

7


had a variance of greater than or equal to 10%, a relatively small but sta-ble set of items emerged (Table 9). Given the poor internal consistencyobtained initially, a decision was made to assess NCAFS item–total corre-lations with caregiver, child, and caregiver–child total scores. If an itemhad an item–total correlation of r = .10 or higher at one or more of theinfant ages, the item was maintained for calculation of new alpha coeffi-cients using KR-20; see Table 10. For the caregiver scale, 28 of 50 possi-ble items remained. For the child scale, nine of 26 possible itemsremained. For the caregiver–child scale, a loss of over one half the itemsfrom the original scale resulted in 37 of 76 possible items remaining.Alpha coefficients improved consistently at each infant age (Table 10).

In light of these results, we decided to assess whether the internal con-sistency of the subscales and the scale overall may be lower due to devel-opmental homogeneity of the study sample compared to the sample in theNCAST database. By developmental homogeneity we mean, for example,one would expect less variability in the behaviors of interest amonginfants of the same age (12 months), such as in our sample, than across

Table 9. Internal consistency: NCAFS items with 10% or greater variancein score

Subscales 12 months 24 months 36 months

I. Sensitivity to cues 5, 7, 8, 10, 1, 5, 7, 8, 10 1, 5, 7, 8, 11,11, 14, 15, 16 11, 14, 15, 16 14, 15

II. Response to child’s distress 18, 19, 21, 22, 23 18, 21, 22, 23 21, 22, 23III. Social–emotional growth fostering 29, 32, 34, 37, 41 29, 32, 34, 37, 29, 32, 34, 37,

39, 40, 41 39, 41IV. Cognitive growth fostering 46 46 –V. Clarity of cues 52, 53, 58, 61, 52, 53, 58, 61, 52, 53, 58, 61,

63, 64 63, 64 63, 64VI. Responsiveness to caregiver 67, 71, 72, 66, 67, 71 66, 71, 72, 73

73, 74 72, 73

Table 10. Internal consistency of the modified NCAFS (N = 116)

Subscales12mosKR-20

24mosKR-20

36mosKR-20

Caregiver .59 .47 .56Child .57 .50 .50Total .63 .57 .57

Dow

nloa

ded

By:

[Uni

vers

ity o

f Nor

th C

arol

ina]

At:

15:0

0 17

Oct

ober

200

7


ages (1–12 months), such as in the NCAST database. In order to assessthis possibility with our data, we randomly selected equal thirds of dyadsfrom the 12-, 24-, and 36-month observations. These thirds from each agegroup were combined to create one group of 116 dyads, which wasassessed for internal consistency using the KR-20 statistic (Table 11).There was no significant improvement in the internal consistency of theNCAFS.

Stability

Stability of the NCAFS was assessed for caregiver, child, and caregiver–child total scores between 12 and 24 months, 24 and 36 months, and12 and 36 months. Caregiver scores were significantly correlated at eachage: r = .15, p = .05; r = .25, p < .01; and r = .30, p < .01, respectively.Child scores were significantly correlated only between 12 and 36months, r = .24, p < .01. Caregiver–child total scores were also signifi-cantly correlated only between 12 and 36 months, r = .32, p < .01.

DISCUSSION

This study was undertaken with the proposition that, during the develop-mental transition from relative dependence in infancy to emerging auton-omy in toddlerhood, quality of feeding interactions between caregiver andchild contributes to the child’s abilities to self-regulate feeding conducive

Table 11. Internal consistency of NCAFS forcombined sample of randomly selected 12-, 24-,and 36-month dyads (N = 116)

Subscales K-R 20 (*)

I. Sensitivity to cues .16 (12)II. Response to child’s distress .34 (9)III. Social–emotional growth fostering .32 (9)IV. Cognitive growth fostering .30 (3)V. Clarity of cues .45 (8)VI. Responsiveness to caregiver −.60 (10)Caregiver .44 (33)Child .38 (18)Total .44 (51)

Note: *number of items remaining in subscale/scale afteritems with zero variance removed.

Dow

nloa

ded

By:

[Uni

vers

ity o

f Nor

th C

arol

ina]

At:

15:0

0 17

Oct

ober

200

7


to optimal growth and development within the child and environment’scapacities. The purpose of this study was to facilitate the extension of anassessment tool for feeding interaction quality during toddlerhood thatwould have utility in both research and clinical practice. The specific aimof this study was exploration of the reliability of the NCAFS beyond 12months to 24 and 36 months.

Interrater Reliability

The interrater reliability in this study was generally quite good. The aver-age percentage agreement was over 90% at all time periods. Given thatthe raters were trained to accuracy in coding at 90% or higher agreementwith NCAFS standards, the average percent agreement between raters inthis study suggests that the NCAFS behavioral checklist was being usedas it was intended. Moreover, the scores should accurately reflect interac-tion qualities as measured by the NCAFS if other trained raters were toscore these dyads.

Interrater agreement correcting for chance agreement also ranged fromgood to excellent according to criteria set forth by Fleiss, Levin, and Park(2003). The median overall NCAFS kappa statistics improved withincreasing infant age. This most likely reflects a growing ease in observ-ing presence or absence of interactive behaviors in both mother and child,which may have arisen from several sources.

First, given the children’s increasing capacity for fine and gross motorcontrol and locomotion, caregivers were no longer as responsible forpositioning and actively feeding the child at 24 and 36 months. The tod-dlers were most often seen positioning themselves in their chairs and themajority actively explored the various snacks available to them, engagingin self-feeding with help from mothers in the opening and closing of con-tainers and holding cups when necessary. They were also able to stand,turn, and walk away from or toward caregivers, with greater clarity ofengagement and disengagement cues. Perhaps the most striking contribu-tion to the growing ease of coding presence/absence of behaviors seemsto have been due to the children’s growing developmental sophisticationin communication through language to express desires, needs, and dis-tress. One would expect to see more verbal communication in feeding/eating interactions and children verbally asserting their desires to feedthemselves between 18 and 36 months (Sander, 1975). This was certainlythe case among the toddlers in this sample. Increasingly intelligible dis-cussion between mother and child in addition to increasing clarity of thetoddler’s behavioral engagement and disengagement cues further clarified

Dow

nloa

ded

By:

[Uni

vers

ity o

f Nor

th C

arol

ina]

At:

15:0

0 17

Oct

ober

200

7


the presence/absence of maternal and child interactive behaviors for theraters.

Among the dyads in which interrater agreement and reliability wereassessed, caregiver–child total scores were positively correlated betweenthe two raters at 12, 24, and 36 months. Furthermore, there were no sig-nificant differences between these scores at any of the infant ages. Thus,in terms of assessment of overall dyadic feeding quality, the coders ratedconsistently with one another.

Internal Consistency

Internal consistency of the NCAFS at 12, 24, and 36 months was low incomparison to conventional limits (Schmitt, 1996). However, Pedhazurand Schmelkin (1991) contend that reference to “authoritative” sources isnot as important as the researcher’s determination of the level of error thatis tolerable in light of the study’s purpose. Schmitt (1996) echoes this,suggesting that if a measure has meaningful content validity for thedomain of interest and is relatively unidimensional, then low reliabilitymay not be a serious problem depending upon the research goal. Further-more, Schmitt (1996) suggests that alpha is not a measure of unidimen-sionality and that alpha is an underestimate of reliability formultidimensional measures. These issues deserve further exploration.

The KR-20 reliability statistic parallels the alpha coefficient and is theappropriate choice for dichotomous data. However, Pedhazur andSchmelkin (1991) pointed out that the underlying assumption of KR-20 isone of parallelism of items, meaning that the true scores for all items areequal, as are all of the errors; this is rarely found to be true in reality.Dichotomous items are also likely to have lower reliability compared toitems with more categories or distinctions by virtue of their relative lackof variability (Pedhazur & Schmelkin, 1991).

The NCAFS caregiver, child, and caregiver–child total scores gener-ally had higher internal consistency at each age than their component sub-scales. This, in part, may be due simply to the larger number of items inthe caregiver, child, and caregiver–child composite scores. This is consis-tent with the principle of aggregation, in which the combination of a set ofitems has greater stability and less bias than any single item in the set(Rushton, Brainerd, & Pressley, 1983). The NCAFS is constructed in thisway, such that the caregiver–child total scores comprised of all itemsreflect overall feeding interaction quality. Sumner and Spietz (1994) ech-oed this idea of aggregation in their discussion of the high alpha estimatefor the caregiver–child total score compared to the lower alpha statisticsfor the subscales reported in the training manual. They suggested that this

Dow

nloa

ded

By:

[Uni

vers

ity o

f Nor

th C

arol

ina]

At:

15:0

0 17

Oct

ober

200

7


total score is more reliable for group comparison than are the subscalescores. The multiple subscales of the NCAFS imply multidimensionalityfor feeding interaction quality. Thus, in light of Schmitt’s (1996) asser-tions regarding multidimensionality, the KR-20 reliability statistics in ourstudy may underestimate the reliability of the NCAFS.

The internal consistency reliability estimates in this study were consid-erably lower than those reported in the NCAFS manual (Sumner &Spietz, 1994). A plausible explanation for this lies in the lower levels ofvariability in the coded behaviors in this sample compared to the samplein the NCAST database. The period from 1–12 months of age is markedby several developmental stages of adaptation in caregiver–infant interac-tion (Sander, 1975). Thus, one would expect a greater degree of variabil-ity in presence and absence of NCAFS behaviors over one year than onewould expect at a single age, such as 12 months. For example, one wouldtypically expect greater variability between a 1-month-old and 9-month-old than between two 12-month-olds. The same would likely be true ofvariability between a mother interacting with a 1-month-old and a motherinteracting with a 9-month-old compared to the variability between twomothers interacting with their two 12-month-old children. The results ofour analysis of a group composed of randomly drawn equal thirds of 12-,24-, and 36-month-old children and their mothers did not lead to a signif-icant improvement in the internal consistency of the NCAFS. This wasmost likely due to the inability of the items to differentiate the older chil-dren and their mothers in our sample as the items were written foryounger children.

Another potential explanation for the lower internal consistency of theNCAFS in our study compared to the statistics reported in the NCAFStraining manual (Sumner & Spietz, 1994) lies in methodological differ-ences in data collection. Internal consistency alphas in the NCAFS man-ual were based on live observations by a group, primarily nurses, whowere trained to reliability and who scored the observations independently,checked agreement, discussed differences, and submitted mutually agreedupon scoring for the observations (K. E. Barnard, personal communica-tion, March 15, 2006). K. E. Barnard (personal communication, March15, 2006) suggested that the use of video, in which multiple reviews areavailable, tends to result in lower alphas and that this difference in meth-ods likely results from the bias observers build into live scoring whenthey cannot recall some item detail and tend to score it like the other itemsin the subscale. In the Early Childhood Longitudinal Study-Birth Cohort(Andreassen & Fletcher, 2005), in which coding was performed on video-taped feeding sessions, alphas for the NCAFS were also consistentlylower than those reported in the NCAFS training manual. This suggeststhat the reported alphas in the NCAFS manual may be somewhat inflated

Dow

nloa

ded

By:

[Uni

vers

ity o

f Nor

th C

arol

ina]

At:

15:0

0 17

Oct

ober

200

7


due to a halo effect or the possibility that during live-time, single opportu-nities for coding, salient behaviors may be more readily recalled andinfluence the scoring of less salient or unseen behaviors.

As previously stated, approximately one third of the total items in theNCAFS were dropped from assessment of internal consistency due tozero variance. In addition, a large number of items had very low variance.Thus, one would expect the reliability coefficients of the NCAFS in thisstudy to be lower due to a decrease in the ratio of true variance to errorvariance (Streiner & Norman, 1995). The NCAFS was designed to assessfeeding interaction quality for infants up to 12 months of age. As dis-cussed earlier, the significant number of items with little or no variance inthis study could be explained by the developmental capacities of olderchildren and caregiver–child adaptations. For example, item 65 states“child has less than three rapid state changes during the feeding” (Sumner& Spietz, 1994). While this item may differentiate the younger infant whomay have less state stability, older children without developmental delaysare much less likely to demonstrate such lability in states of sleep andconsciousness. Item 2, which states “caregiver positions child so that thechild’s head is higher than the hips” (Sumner & Spietz, 1994) is no longeras applicable for older children because they have greater strength andmotor skills and can position themselves. However, at least one of theitems may have had little or no variance due to the characteristics of theobservational setting. Item 28, “Caregiver pays more attention to the childduring the feeding than to other people or things in the environment”(Sumner & Spietz, 1994), was endorsed as present most of the time. Thisfinding may not have been as consistent across the sample had the obser-vation taken place in the home with other family members and tasks com-peting for the mother’s attention.

Attrition of items led to a decreased number of items effectively differ-entiating the dyads. It is important to note that the internal consistency ofthe caregiver, child, and caregiver–child total scores improved at eachobservation period once items with correlations lower than .10 weredropped from the calculation of internal consistency. This does not meanthat the NCAFS need be discarded as a measure of feeding interactionquality for dyads with older children. It does suggest, however, that theNCAFS will require modification in order to be a more internally consis-tent measure of feeding interaction quality.

Recommendations about potential additions or revisions to the NCAFSfor older age children include items or new item descriptions that wouldreflect whether adaptation tasks are being met by the dyad. For example,whereas the current version has the caregiver achieving higher scoresthrough actively positioning and actively feeding the child, new itemsmight retain the importance of positioning and feeding but have wording

Dow

nloa

ded

By:

[Uni

vers

ity o

f Nor

th C

arol

ina]

At:

15:0

0 17

Oct

ober

200

7


that reflects the caregiver’s adaptive role of providing structured, safe,and appropriate meal contexts and foods that allow for the child’s devel-oping motor and cognitive skills in relation to eating.

Items related to social-emotional and cognitive growth fostering bycaregivers would also need to reflect the older child’s capacities andneeds for stimulation that are different from those of the younger infant.Item description for raters would need to reflect expected developmentalcapacities and needs of older children and ways that caregivers accommo-date these. Theoretically, items for both child and caregiver might also bechosen to reflect adaptive issues that arise during the second and thirdyear of life, such as the child’s increased initiative, increased capacity forlanguage and subsequent communication of inner experiences/feelings/intentions, and occasional purposeful disruption of previously facilitativeand synchronized interactions with the caregiver (Sander, 1975).

Stability of the NCAFS over Time

Although the correlations were weak to moderate, overall maternal contri-butions to dyadic feeding quality were relatively stable over the 1- and2-year intervals in this study. Child and caregiver–child total scores, how-ever, were significantly correlated only across 12 to 36 months. Interpre-tation of these data is somewhat problematic. There appeared to be lessstability in the child scores at 24 months and this impacted the stability ofthe caregiver–child total scores, as well. One of the central issues iswhether one would expect stability of child scores over time. Given thecurrent form of the NCAFS, some items might be endorsed as presentmore frequently in older children whereas others might more frequentlybe endorsed as absent, reflecting the developmental capacities of thechild. Thus, one may see a change in the pattern of endorsement of itemsas the child ages, but not necessarily in the overall score. The importanceor confidence one can place in the finding of a significant, though weak tomoderate, correlation between 12- and 36-month scores for both child andcaregiver–child total scores is unclear and requires further investigationgiven that test–retest correlations may be influenced by the reliability ofmeasures, true changes in individuals, or some combination of the two(Pedhazur and Schmelkin, 1991). However, both the possible disruptionof previously harmonious interactions between caregiver and child due todyadic adaptation tasks between 12 and 36 months (Sander, 1975) and thetypically successful negotiation of these tasks and return of harmoniousinteractions by around 36 months provide a theoretical explanation forthese findings.

Dow

nloa

ded

By:

[Uni

vers

ity o

f Nor

th C

arol

ina]

At:

15:0

0 17

Oct

ober

200

7


LIMITATIONS

This study drew from an existing data set that was collected from 1993–1997, as part of a larger investigation that assessed mother–child control-salient interactions at 12, 24, and 36 months (Houck, 1999). The qualityof feeding interaction was assessed as one of four mother–toddler interac-tions that were the focus of this larger study. Thus, the original study didnot include the measurement of variables that may have been of interest inrelation to the current study, such as anthropometric measures and feed-ing histories. The observational data in the current study consisted of vid-eotaped feeding interactions. Due to the relatively fixed position of thecamera, there is the potential for some facial expressions or other moresubtle interactive behaviors to have been missed, particularly as older tod-dlers became more mobile. However, the high chair, chairs, and tablewere positioned in such a way that the majority, if not all, of the physicalinteractive behaviors could be viewed. Furthermore, the interactivebehaviors were viewed in context with the concurrent sound recording ofverbal exchanges between mothers and their toddlers. Future observa-tional research of feeding interactions may be enhanced through the useof multiple cameras.

The duration of the videotaped feeding observation was generally set at10 minutes. For the majority of dyads at each age this precluded observa-tion of the natural termination of feeding. Consequently, a few behaviorswere scored as absent that may have been present had the feeding reachedits natural terminus. This scoring bias likely contributed to some dyads,mothers, and children obtaining lower scores than they may have other-wise. In future feeding studies, it will be important to measure the feedinginteraction from its natural onset to its natural termination, which wouldbe determined by each individual dyad.

Another potential limitation of the study may arise from reactivity ofthe mothers, children, or both to the observation context. Mothers andchildren may have interacted differently due to an awareness of beingobserved or the novelty of the observational setting in comparison tomore familiar contexts, such as the home. While mothers knew theywere being videotaped, children most often did not or did not expresssuch knowledge verbally during observation. Videotaping through aone-way mirror minimized intrusiveness of the observation. In addi-tion, the feeding interaction occurred after other mother–toddler activi-ties in the room, thereby decreasing the effect of novelty of context.Nonetheless, bias due to observational method and context remainspossible.

Dow

nloa

ded

By:

[Uni

vers

ity o

f Nor

th C

arol

ina]

At:

15:0

0 17

Oct

ober

200

7


SUMMARY AND RECOMMENDATIONS

In order to facilitate better understanding of feeding interaction quality,one of the primary aims of this study was to examine the reliability of theNCAFS during toddlerhood in order to enhance its utility in measuringfeeding interaction quality beyond its current limit of 12 months of childage. Reliability of the NCAFS at 12, 24, and 36 months was exploredthrough assessment of interrater reliability, internal consistency of thevarious subscales and the scale as a whole, and stability of the scale mea-surements over time.

The interrater agreement and reliability for the NCAFS was generallyquite good, implying that the NCAFS was being used in a consistent andreliable manner by the two coders in this study. The internal consistencyof the NCAFS was low in comparison to conventional limits. This waslikely due to relatively low levels of variance among the dyads withineach age and the attrition of several behavior items due to zero variance.

There are several approaches that could be considered for improve-ment of the internal consistency of the NCAFS during toddlerhood. Theaddition of items capable of more clearly capturing the frequency of cer-tain behaviors in which frequency may be important may capture greatervariability. Likert scaling of certain items may also achieve this. Replace-ment of items with zero or little variance that may no longer be develop-mentally relevant for older children or refinement of behavioraldescriptions for current items in order to assess whether dyadic adaptationtasks are being met and thereby better capture differences in interactionquality may also boost internal consistency.

Maternal contributions to feeding interaction quality remained rela-tively stable over time, but child and dyadic contributions did not. Thisappears to have been due to less stability of child scores at 24 months.Analysis of stability of interaction quality should be assessed in futurestudies with an instrument that is capable of capturing dyadic interactionquality in light of child development over time. A refined NCAFS maydemonstrate such a capacity.

ACKNOWLEDGMENTS

The primary author would like to acknowledge funding support for thisstudy from National Institute of Nursing Research grant numbers T32NR07061 and 1F31 NR08651-01, The Northwest Health Foundation,Nurses Educational Funds, Inc., and the Lindeman and Pearson Scholar-ship Awards from Oregon Health & Science University. The primaryauthor would also like to acknowledge Drs. Kathryn Barnard, Elizabeth

Dow

nloa

ded

By:

[Uni

vers

ity o

f Nor

th C

arol

ina]

At:

15:0

0 17

Oct

ober

200

7


Lecuyer, Sheila Kodadek, and Jennifer O. Fisher for comments on earlierdrafts of this manuscript and Laura Tomanka for assistance in observa-tional coding.

REFERENCES

Andreassen, C., & Fletcher, P. (2005). Early Childhood Longitudinal Study, Birth Cohort(ECLS-B), Psychometric Characteristics. In ECLS-B Methodology Report for theNine-Month Data Collection (NCES 2005-100) (Vol. 1). U.S. Department of Educa-tion. Washington, DC: National Center for Education Statistics.

Barnard, K. E., Hammond, M. A., Booth, C. L., Bee, H. L., Mitchell, S. K., & Spieker, S. J.(1989). Measurement and meaning of parent-child interaction. In F. J. Morrison,C. E. Lord & D. P. Keating (Eds.), Applied Developmental Psychology (Vol. 3, pp.39–80). New York: Academic Press.

Benoit, D. (2000). Feeding disorders, failure to thrive, and obesity. In C. H. Zeanah (Jr.)(Ed.), Handbook of infant mental health (2nd ed., pp. 339–352). New York: TheGuilford Press.

Birch, L. L., & Fisher, J. O. (1998). Development of eating behaviors among children andadolescents. Pediatrics, 101, 539–549.

Brewis, A., & Gartin, M. (2006). Biocultural construction of obesogenic ecologies ofchildhood: Parent-feeding versus child-eating strategies. American Journal of HumanBiology, 18, 203–213.

Bruch, H. (1973). Eating disorders: Obesity, anorexia nervosa, and the person within.New York: Basic Books, Inc.

Chatoor, I., Egan, J., Getson, P., Menvielle, E., & O’Donnell, R. (1987). Mother-infantinteractions in infantile anorexia nervosa. Journal of the American Academy of Childand Adolescent Psychiatry, 26, 535–540.

Chatoor, I., Getson, P., Menvielle, E., Brasseaux, C., O’Donnell, R., Rivera, Y.,et al. (1997). A feeding scale for research and clinical practice to assess mother-infant interactions in the first three years of life. Infant Mental Health Journal,18, 76–91.

Chatoor, I., Hirsch, R., Ganiban, J., Persinger, M., & Hamburger, E. (1998). Diagnosinginfantile anorexia: The observation of mother-infant interactions. Journal of the Amer-ican Academy of Child and Adolescent Psychiatry, 37, 959–967.

Costanzo, P. R., & Woody, E. Z. (1985). Domain-specific parenting styles and theirimpact on the child’s development of particular deviance: The example of obesityproneness. Journal of Social & Clinical Psychology, 3, 425–445.

Crittenden, P. (1987). Non-organic failure to thrive: Deprivation or distortion? InfantMental Health Journal, 8, 51–64.

Faith, M. S., Scanlon, K. S., Birch, L. L., Francis, L. A., & Sherry, B. (2004). Parent-ChildFeeding Strategies and Their Relationships to Child Eating and Weight Status. ObesityResearch, 12, 1711–1722.

Farrow, C., & Blissett, J. (2006). Does maternal control during feeding moderate earlyinfant weight gain? Pediatrics, 118, 293–298.

Fleiss, J. L., Levin, B., & Park, M. C. (2003). Statistical Methods for rates and proportions(3r ed.). Hoboken, NJ: John Wiley & Sons, Inc.

Dow

nloa

ded

By:

[Uni

vers

ity o

f Nor

th C

arol

ina]

At:

15:0

0 17

Oct

ober

200

7


Hagekull, B., Bohlin, G., & Rydell, A.-M. (1997). Maternal sensitivity, infant tempera-ment, and the development of early feeding problems. Infant Mental Health Journal,18, 92–106.

Houck, G. M. (1999). The measurement of child characteristics from infancy to toddler-hood: Temperament, developmental competence, self-concept, and social competence.Issues in Comprehensive Pediatric Nursing, 22, 101–127.

Kimbro, R. T., Brooks-Gunn, J., & McLanahan, S. (2007). Racial and ethnic differentialsin children’s overweight and obesity among 3-year-olds. American Journal of PublicHealth, 97, 1–8.

Kuder, G. F., & Richardson, M. W. (1937). The theory of estimation of test reliability.Psychometrika, 2, 151–160.

Lucarelli, L., Ambruzzi, A. M., Cimino, S., D’Olimpio, F., & Finistrella, V. (2003). Feed-ing disorders in infancy: an empirical study on mother-infant interactions. MinervaPediatrica, 55, 243–253.

Mei, Z., Scanlon, K. S., Grummer-Strawn, L. M., Freedman, D. S., Yip, R., &Trowbridge, F. L. (1998). Increasing prevalence of overweight among US low-incomepreschool children: The Centers for Disease Control and Prevention pediatric nutritionsurveillance, 1983 to 1995. Pediatrics, 101(1), e12.

Ogden, C. L., Troiano, R. P., Briefel, R. R., Kuczmarski, R. J., Flegal, K. M., & Johnson,C. L. (1997). Prevalence of overweight among preschool children in the United States,1971 through 1994. Pediatrics, 99(4), e1.

Pedhazur, E. J., & Schmelkin, L. P. (1991). Measurement, design, and analysis: An inte-grated approach. Hillsdale: Lawrence Erlbaum Associates, Inc.

Rushton, J. P., Brainerd, C. J., & Pressley, M. (1983). Behavioral development and con-struct validity: The principle of aggregation. Psychological Bulletin, 94(1), 18–38.

Sander, L. W. (1975). Infant and caretaking environment: Investigation and conceptual-ization of adaptive behavior in a system of increasing complexity. In E. J. Anthony(Ed.), Explorations in child psychiatry (pp. 129–166). New York: Plenum Press.

Satter, E. M. (1990). The feeding relationship: Problems and interventions. The Journal ofPediatrics, 117, S181–S189.

Schmitt, N. (1996). Uses and abuses of coefficient alpha. Psychological Assessment, 8,350–353.

Spegman, A. M., & Houck, G. M. (2005). Assessing the feeding/eating interaction as acontext for the development of social competence in toddlers. Issues in ComprehensivePediatric Nursing, 28, 213–236.

Streiner, D. L., & Norman, G. R. (1995). Health assessment scales: A practical guide totheir development and use (2nd ed.). Oxford: Oxford University Press.

Sumner, G., & Spietz, A. (1994). NCAST Caregiver/Parent-Child Interaction FeedingManual (pp. 1–176). Seattle: NCAST Publications, University of Washington, Schoolof Nursing.

issues in comprehensive pediatric nursingthomas/research/publications/nursing child... ·...

Documents