Measuring Feeding Practices among Early Care and Education Teachers and Examining Relations with Food Insecurity

Abstract

Early care and education teachers’ (ECETs) dietary and feeding behaviors have the potential to influence children’s health outcomes. This study sought to (1) gather data on the properties and performance of the Child Feeding Questionnaire and Child Feeding Style Questionnaire in an ECET sample and compare properties to published parent samples, (2) document rates of teacher food insecurity over the last 12 months in the largest sample of ECETs to date, and (3) examine relations between FI experiences by ECETs and reported feeding practices. ECETs completed 506 cross-sectional surveys. Mean patterns, ranges, and internal consistency values on the adapted instruments for ECETs were consistent with those published for parents. Significant mean differences between parents and ECETs on established scales using one-sample t-tests were prevalent with medium to large effect sizes despite small, relative differences. The majority of ECETs were authoritarian (35.6%), followed by indulgent (29.2%), authoritative (17.9%), and uninvolved (17.3%). T-tests indicated that ECETs who were currently food insecure were significantly higher than teachers who were currently food secure on scales of perceived responsibility, concern about child weight, restriction, pressure to eat, monitoring, demandingness, and responsiveness (all p < .001). Chi-square tests found that food insecurity was not independent from ECET feeding style, with a greater occurrence of authoritarian and less of indulgent feeding styles for ECETs who were food insecure. Overall, analyses support that two popular measures of feeding practices function similarly in ECETs as they do in parents. Additionally, results demonstrate associations between food insecurity and ECETs’ feeding practices.

Introduction

Over 15 million children under age 5 attend center-based care in the United States where they spend an average of 36 hours per week (Child Care Aware of America, 2017). Early care and education teachers (ECETs) provide care to as many as 20 children in a classroom per day and provide oversight for 2 to 3 meals and snacks per day for children in their care (Adminsistration for Children and Families, 2017). In fact, children may eat as much as two-thirds of their daily nutrients in this setting (Radcliffe, Cameron, & Baad, 2002). ECETs have many opportunities to influence children with their behaviors and feeding practices (Benjamin-Neelon, 2018). Studies confirm that ECETs’ enthusiastic verbal role modeling (Hendy & Raudenbush, 2000), making positive comments about healthy foods (Gibson et al., 2012), developing children’s interest through conversations about food (Nicklas et al., 2009), and encouraging children to assist with meal preparation all support children in learning positive food behaviors and attitudes (Gubbels, Gerards, & Kremers, 2015). Thus, early childhood education is a critical setting for prevention of excess weight and promotion of positive nutritional habits and preferences. In light of the potential influence of ECETs on children’s health outcomes, the field requires studies that help our understanding of predictors of variation in ECET feeding and modeling behaviors.

Food insecurity among the early care and education workforce has been documented in prior studies. In a 2013 Michigan Head Start sample, ECETs indicated food insecurity at a rate of 28% using the United States Department of Agriculture (USDA) household food security module survey (Song et al., 2016). A 2012/2013 sample of ECETs across public and private childcare found that nearly half of participants (48%) worried about having enough money to provide food for their families (Whitebook, Phillips, & Howes, 2014). More recent studies of ECETs in Arkansas have documented overall food insecurity rates at 34.5% (Swindle et al., 2018) and 40% (McKelvey, Forsman, & Morrison-Ward, 2018) using screeners based on the US food security module. Sub-groups in these studies experienced even higher rates, such as assistant teachers at 44.6% (Swindle et al., 2018) and ECETs in infant-toddler classrooms at 50% (McKelvey et al., 2018). In every case, rates of food insecurity are strikingly high – between double and quadruple the US national estimate of 11.8% (United States Department of Agriculture Economic Research Service, 2017). These studies illustrate high rates of food insecurity among ECETs but do not explore the consequences of this experience in the classroom.

Children are embedded within a complex system of family, school, and community (Bronfenbrenner, 2005). They both influence and are influenced by all of the factors within these systems. As reviewed above, prior work has demonstrated that ECET feeding practices influence children (e.g., Gubbels et al., 2015). Therefore, ECETS are a natural target for interventions aimed at other systems beyond the family for influencing children’s health outcomes. However, personal factors of ECETs that could moderate their behavior in the school environment need to be understood as programs may need to focus on understanding and addressing the influence of those factors before they are able to effect change with their child health intervention program. For example, prior study has identified that ECETs who were on a diet or of minority ethnicity were more likely to use restrictive feeding practices (Dev et al., 2014).

Another potential personal factor that may influence ECETs’ behavior around feeding practices in the classroom are their own experiences with food insecurity, especially among low-resource school districts. If ECETs both know what food insecurity feels like personally and know that many children in their classrooms also experience food insecurity, they may be more likely to disregard training, school policies, and evidence-based practices related to portion sizes, pressuring children to eat, etc. (Swindle, Patterson, & Boden, 2017). The influence of food insecurity on feeding practices has been demonstrated in research with parents such that food insecurity decreases positive parenting practices, which, in turn, predict use of feeding practices (Bronte-Tinkew et al., 2007). This suggests that outcomes of school-based nutrition interventions or obesity prevention efforts may be moderated by ECET experiences with food insecurity.

A few studies have explored the psychometric properties of tools designed to assess feeding practices for ECETs. For example, the Environmental Policy Assessment and Observation Self-Report contains subscales which assess ECET feeding practices and has demonstrated test-retest reliability (Fallon et al., 2018; Ward et al., 2015) and predictive validity (Anundson et al., 2018). The About Feeding Children (AFC) survey was developed with extensive input from childcare stakeholders to reflect the unique aspects of feeding multiple children at once in the childcare setting and has demonstrated a strong psychometric structure and appropriate internal consistency (Swindle, Sigman-Grant, Branen, Fletcher, & Johnson, 2018). Other tools are available to measure environmental factors and policies that may impact feeding practices of ECETs (Benjamin et al., 2007; Falbe, Kenney, Henderson, & Schwartz, 2011; Henderson et al., 2011; Seward et al., 2017; Ward et al., 2015). Each of these are useful tools with distinct purposes and have the value of being designed for the childcare setting. However, none of the available tools designed with early care and education in mind facilitate comparison with feeding practices children experience at home.

Several researchers have used the measures available for parenting to measure feeding practices of ECET, the Child Feeding Questionnaire (CFQ; Dev & McBride, 2013; Dev, Speirs, McBride, Donovan, & Chapman-Novakofski, 2014; Elford & Brown, 2014) and the Child Feeding Styles Questionnaire (CFSQ; Dev & McBride, 2013; Freedman & Alvarez, 2010; Musher-Eizenman & Holub, 2007). The CFQ is designed to measure parents’ perceptions, attitudes, and practices related to feeding their children. The CFSQ is designed to group parents into feeding styles of authoritative, authoritarian, indulgent, and uninvolved. Both the CFQ and CFSQ can be modified to be relevant for ECETs who are not parents of the children in their care. Such use of these measures is appealing to facilitate comparisons with the wealth of previous studies of parent feeding practices impacts on children. These types of comparisons would be valuable in studies targeting multiple settings of children’s dietary intake or to understand differential impacts of parent and ECET influence on children’s eating behavior. That is, researchers would be able to observe the aspects of parents and ECETs feeding that are most influential on children, both inside and outside of intervention settings and across diverse groups. However, no studies have endeavored to determine properties of these tools among ECETs.

The objectives of the current study were to (1) gather data on the properties and performance of the CFQ and CFSQ in an ECET sample and compare properties to published parent samples and (2) examine relations between FI experiences by ECETs and reported feeding practices. We expected the following for each of these objectives:

Hypothesis 1: The adapted CFQ and CFSQ will show comparable psychometric properties in ECETs as they did for parents in prior studies.

Hypothesis 2: Food insecurity will be significantly associated with feeding practices among ECETs.

In accomplishing these objectives, the current study will provide important information on the utility of tools to compare parent and ECET feeding practices. Further, this study will provide insights into the relation between ECET food insecurity and feeding practices that are not yet available in the literature.

Method

Study Design

Participants for the surveys were recruited at state-funded training opportunities and state conferences in early childhood education across two school years (Fall 2015 to Spring 2017) in two Southern states of the United States. Study staff were present at the conferences in public areas (e.g., lobbies, exhibition halls) and invited ECETs to participate (e.g., Would you like to complete a survey and select a small item for your classroom?”); there was a 61.2% participation rate among invited ECETs. Additionally, some surveys were filled out by ECETs attending training for a classroom-based nutrition program prior to the beginning of training. In both settings, ECETs who agreed to participate received surveys with instructions that explained the purpose of the study (i.e., “a research project that intends to develop tools to assess the strategies educators use to feed children in early childcare”), the voluntary nature of the survey, the anonymity of the survey, and the availability of research staff to address their questions.

Surveys had no personally identifying information, and all surveys were completed with pen and paper. All ECETs employed in any type of preschool setting were eligible to fill out the surveys. ECETs received an incentive with a value of up to a $5.00 (e.g., notebook, pen set) value for participation.

Measures

The Child Feeding Questionnaire (CFQ; Birch et al., 2001) and Child Feeding Style Questionnaire (CFSQ; Hughes et al., 2012) are validated instruments that assess parental feeding practices. The CFQ includes seven scales defined as following: (1) perceived responsibility is the “perceptions of responsibility for child feeding,” (2) perceived adult weight is the “perceptions of own weight status,” (3) perceived child weight is the “perceptions of the child’s weight status,” (4) concern about child weight is the “concern about child’s risk of being overweight”, (5) pressure to eat is the “tendency to pressure children to eat more food,” (6) restriction is “restricting access to food,” and (7) monitoring is the “extent to which adults oversee child’s eating (p 203 – 204; Birch et al., 2001).” The CFSQ includes two scales, demandingness and responsiveness which are used to group adults into four feeding styles. Demandingness is the “extent to which parents show control, maturity demands, and supervision in their parenting responsiveness is the “extent to which parents show affective warmth, acceptance, and involvement (p 84; Hughes, Power, Fisher, Mueller, & Nicklas, 2005).” The feeding styles reflect scores on these two dimensions as follows: authoritative (high responsiveness, high demandingness), authoritarian (low responsiveness, high demandingness); indulgent (high responsiveness, low demandingness), and uninvolved (low responsiveness, low demandingness). This study will use these terms to refer to these definitions throughout. The CFQ and CFSQ instruments have demonstrated the impact of parent mealtime strategies on children. One study with 50 early ECETs modified the CFSQ into a checklist format (Hughes et al., 2007). Other studies used the CFQ and CFSQ among ECETs but without the goal of examining scale properties or comparing performance to the use of scales among parents (Dev, McBride, et al., 2014).

Educator Feeding Questionnaire Development.

We modified the CFQ and CFSQ instruments to be applicable to the ECET population, and we refer to this modification and combination as the Educator-Child Feeding Questionnaire/Educator-Caregiver Feeding Styles Questionnaire (E-CFQ/E-CFSQ). All items from the original measures (CFQ, CFSQ) were retained where possible to ensure the greatest possible consistency between parent and ECET measures. Item wording was slightly altered from “parent” to “educator” and “home” to “classroom” terminology. Consistent with prior adaptations, questions asked ECETs to reflect on how they feed their class as a whole. Of the seven CFQ scales, there were two item modifications made on the E-CFQ: 1) only one item was collected for the perceived child weight scale, “most children you teach are…?” and 2) one item was added to the concern about child weight scale, “how concerned are you about the children in your classroom eating too little?” No items were added or deleted for the two E-CFSQ scales (demandingness, responsiveness). Scales of measurement were consistent with the original instruments with responses on a 1 to 5 scale on the CFQ and CFSQ items.

Food Insecurity.

Food insecurity was measured using a two-item screen for food insecurity based on the USDA’s Household Food Security Survey Module (United States Department of Agriculutre, 2012) and previous validation of brief screening (Swindle, Whiteside-Mansell, & McKelvey, 2013). ECETs responded about the frequency of the two screening items on a 1 to 3 scale (1 = Never, 2 = Sometimes, or 3 = Often). Items asked about food running out without money to get more and having to cut the size of meals or skip meals because of lack of money for food. Consistent with published scoring recommendations for the brief screening, ECETs were categorized as food insecure if they answered “Sometimes” or “Often” to either item. FI)].

Data Analysis

Data were analyzed using SPSS Version 25 (IBM, 2017). Item descriptives examined means and ranges of items. All adapted items for the E-CFQ/E-CFSQ have missing percentages less than 10% and 12%, respectively, and listwise deletion was used for all analyses. Item scales were created to match those from the published parenting measures. Means, ranges, and internal consistency (i.e., Cronbach’s alpha) values for each scale were computed and compared to published values from the parenting literature for the CFQ. One-sample t-tests were used to compare E-CFQ/E-CFSQ scales in this study among ECETs to published CFQ and CFSQ values from parents. ECETs were grouped as authoritarian, authoritative, indulgent, or uninvolved based on the E-CFSQ responses consistent with Hughes and colleagues (Hughes et al., 2012). Next, the relation between the categorical distinctions suggested by the E-CFSQ and the scale means of the E-CFQ were examined consistent with Hughes and colleagues (Hughes, Power, Fisher, Mueller, & Nicklas, 2005). Then, differences by FI status were tested using independent sample t-tests for E-CFQ scales and chi-square tests of independence for E-CFSQ feeding styles. For all tests reported, a conservative p-value (α = .002) was adopted consistent with a Bonferroni correction for multiple comparisons (α = .05/38 tests = .001).

Results

Participant Demographics.

A total of 506 ECETs completed the survey. ECETs in the study indicated they were African American (55%), White (43%), or other (2%); nearly 20% indicated they were Hispanic (19.8%). Nearly all (99.2%) were female. On average, participants were 41.6 years old (SD = 12.3, min = 18, max = 76) with 12.3 years of experience teaching (SD = 9.3, min = 0, max = 41). Educational attainment was primarily an associate’s degree (31.4%) followed by a bachelor’s degree (27.9%), some college (16.6%), a high school diploma (9.0%), a master’s degree (13.5%), or other training (1.5%). Many ECETs worked for either a Head Start agency (68%) or state-funded preschool program (16%); other ECETs (16%) were employed by other types of programs such as private, home-based, or faith-based programs.

Examining properties and performance of the E-CFQ and E-CFSQ.

Tables 1 and 2 presents scale means, ranges, and internal consistencies for E-CFQ/E-CFSQ scales in the current sample, respectively. Cronbach’s alpha values ranged between 0.70 and 0.96 and were within 0.10 units for parents and ECETs for scales on the E-CFQ. Internal consistencies were all greater than .70 and within .03 units for comparing parents and ECETs on the E-CFSQ scales. All scales in the ECET sample were within acceptable range for evaluating skewness (min = - 0.18, max = 0.75) and kurtosis (min = −1.37, max = 1.55).

Table 1.

Child Feeding Questionnaire: Scale Means, Ranges, and Internal Consistencies for Parents in Published Samples and Educators in Our Preliminary Studies

Scales	Mean (SD)			Cronbach’s Alpha
	Parent† (N = 295)	Educator (N = 506)	Cohen’s d	Parent†	Educator
Perceived	4.5 (0.65)	3.1 (1.06)	1.53	0.73	0.73
Responsibility*
Perceived Adult Weight	NA	3.3 (0.53)	-	NA	0.71
Perceived Child Weight	NA	3.0 (0.61)	-	NA	NA*
Concern about Child	1.7 (0.91)	2.4 (1.20)	0.64	0.85	0.91*
Weight*
Pressure to Eat*	2.7 (1.07)	3.0 (0.96)	0.28	0.62	0.70
Restriction*	3.3 (0.92)	2.8 (0.91)	0.56	0.75	0.81
Monitoring*	4.0 (1.03)	2.8 (1.47)	0.95	0.86	0.96

^†Values for parents are based on published data from a sample that was entirely low-income with 16% African American, 68% White, 8% Hispanic, and 8% other/more than one race (Goulding et al., 2014).

NA= Not Available. Only one item collected for the Perceived Child Weight scale for educators: “Most children you teach are…”; One item added for the Concern about Child Weight scale for educators: “How concerned are you about children in your classroom eating too little?” Comparisons were made using one-sample t-tests. Significance is noted above:

^*p < .001

Table 2.

Child Feeding Styles Questionnaire: Scale Means, Ranges, and Internal Consistencies for Parents in Published Samples and Educators in Our Preliminary Studies

Scales	Authoritative			Authoritarian			Indulgent			Uninvolved
	Parent† Mean (SD)	Educator Mean (SD)	Cohen’s d	Parent† Mean (SD)	Educator Mean (SD)	Cohen’s d	Parent† Mean (SD)	Educator Mean (SD)	Cohen’s d	Parent† Mean (SD)	Educator Mean (SD)	Cohen’s d
Demandingness	3.11* (.23)	2.57 * (.18)	2.61	3.44* (.46)	2.85* (.44)	1.31	2.26* (.37)	2.00* (.23)	0.84	2.36* (.35)	1.91* (.37)	1.25
Demandingness	1.31* (.09)	1.65* (0.12)	3.21	1.05* (.10)	1.26* (.18)	1.44	1.40* (.16)	1.76* (.15)	2.37	1.08* (.09)	1.33* (.15)	2.02

^†Values for parents are based on published data from a sample of 130 Hispanic and 101 African American parents (Hughes et al., 2005) Demandingness: Parent α = 0.85; Educator α = 0.82.

Responsiveness: Parent α = 0.86; Educator α = 0.83; Child-centered α = 0.71 (parent);Child-centered α = 0.70 (educator)

Comparisons were made using one-sample t-tests. Significance is noted above:

^*p < .001

Comparing scale properties for ECETs to published parent samples.

Table 1 presents ECET means and standard deviations on E-CFQ scales alongside published values for parents on the same scales (Goulding et al., 2014). For the CFQ (Table 1), the lowest scale mean for both parents and ECETs is concern about child weight. The highest mean for parents was perceived responsibility; the highest mean for ECETs was perceived adult weight. One-sample t-test comparisons indicated a significant difference between parents and ECETs on the 5 available CFQ scales for comparison with one small effect size, two medium effect sizes, and two large effect sizes observed. ECETs had lower perceived responsibility [t (1, 498) = −29.44, p < .001], restriction [t (1, 489) = −11.55, p < .001], and monitoring [t (1, 487) = −18.86, p < .001], but higher concern about child weight [t (1, 489) = 12.75, p < .001], and pressure to eat [t (1, 486) = 6.42, p < .001].

Table 2 presents scale means for demandingness and responsiveness for each of the E-CFSQ feeding style categories mirrored against the same published information for parents with one exception. The lowest levels of demandingness are seen among ECETs in the uninvolved category (M = 1.91, SD = 0.37), different from parental patterns which show lowest demandingness among indulgent parents. One-sample t-tests showed that, overall, ECETs had lower demandingness [t (1, 468) = −17.44, p < .001] and higher responsiveness [t (1, 468) = 26.42, p < .001] than parents in published samples (Hughes et al., 2012). All mean differences exhibited large Cohen’s d effect sizes (Table 2).

Table 3 presents the E-CFQ scale means for each E-CFSQ feeding style category along with the same published information for parents. In the published parent data, authoritarian style parents have the highest pressure to eat and highest levels of restriction while authoritative parents have the highest monitoring scores. This pattern was replicated among ECETs in our sample. In parents, the lowest pressure to eat was among indulgent parents; the lowest restriction was among indulgent parents, and the lowest monitoring was among uninvolved parents. This was the same for ECETs for monitoring; however, ECETs with the lowest pressure to eat and restriction scores were the uninvolved. All E-CFQ scales were significantly different for parents CFQ scale means of the same feeding style at p < .001 with ECETs in every group exhibiting lower demandingness, pressure to eat, restriction, and monitoring scale means as well as higher responsiveness means than parents. Compared to parents, slightly more ECETs were authoritative (14.7% parents, 17.9% ECETs) and uninvolved (14.3% parents, 17.3% ECETs); a similar number were authoritarian (36.4% parents, 35.6% ECETs); and fewer were indulgent (34.6% parents, 29.2% ECETs; Table 4). Effect sizes were not calculated for these comparisons because the article presenting these findings in parents did not report standard deviations with these means. Taken together, these results provide partial support for Hypothesis 1; the adapted E-CFQ/E-CFSQ demonstrated similar psychometric properties in educators as they did for parents in prior studies (e.g., internal consistency, range) as well as a similar pattern for relative means of scales. However, there were several mean differences between ECETs in our sample and parents in prior samples.

Table 3.

Means on Child Feeding Questionnaire scales by Child Feeding Styles Questionnaire Feeding Styles

	Authoritative		Authoritarian		Indulgent		Uninvolved
	Parent† Mean	Educator Mean (SD)	Parent† Mean	Educator Mean (SD)	Parent† Mean	Educator Mean (SD)	Parent† Mean	Educator Mean (SD)
Pressure to Eat	3.81*	3.28* (0.81)	3.82*	3.38* (0.85)	3.43*	2.61* (0.95)	3.60*	2.50* (0.89)
Restriction	3.65*	2.95* (0.85)	3.79*	3.18* (0.84)	3.31*	2.62* (0.87)	3.52*	2.31* (0.87)
Monitoring	4.59*	3.37* (1.63)	4.00*	3.16* (1.27)	4.26*	2.38* (1.43)	3.87*	2.02* (1.21)

^†Values for parents are based on published data from. in a sample of 130 Hispanic and 101 African American parents (Hughes et al., 2005); Standard deviations for these scales were not reported in the Hughes sample. Therefore, Cohen’s d cannot be reported for these comparisons. Comparisons were made using one-sample t-tests to assess for differences between educators in this study and parents in published samples within each feeding style. Significance is noted above:

^*p < .001

Table 4.

Child Feeding Questionnaire Scale Means and Child Feeding Styles Questionnaire Scale Means and Categories by Food Security Status

	Current FS N = 297	Current FI N = 209	Cohen’s d
	M (SD)	M (SD)
Perceived Responsibility*	2.86 (1.07)	3.32 (1.00)	0.44
Perceived Adult Weight	3.33 (0.54)	3.31 (0.52)	-
Perceived Child Weight	2.98 (0.56)	3.03 (0.66)	-
Concern about Child Weight*	2.23 (1.13)	2.65 (1.25)	0.35
Pressure to Eat*	2.86 (0.94)	3.19 (0.96)	0.35
Restriction *	2.71 (0.97)	2.99 (0.81)	0.31
Monitoring*	2.86 (0.94)	3.04 (1.40)	0.15
Demandingness*	2.26 (0.49)	2.56 (0.54)	0.58
Responsiveness*	1.52 (0.26)	1.44 (0.27)	0.30
	%	%	Z
Authoritative	17.5	18.4	0.2
Authoritarian	27.2	46.8	2.7
Indulgent	34.7	21.9	−1.9
Uninvolved	20.5	12.9	−1.5

FI = Food Insecure; FS = Food Secure; Comparisons were made using independent-sample t-tests.

^*p≤.001

Documenting rates of food insecurity and examining relations between food insecurity and ECETs feeding practices.

Current FI was observed for 41.3% of ECETs. E-CFQ/E-CFSQ means by FI status are found in Table 4 with significant differences indicated based on independent-sample t-test comparisons. Independent-sample T-tests indicated that ECETs who were FI had significantly higher scores from ECETs who were FS on scales of perceived responsibility [t (1, 462) = −4.87, p < .001], concern about child weight [t (1, 409) = −3.79, p < .001], restriction [t (1, 478) = −3.42, p = .001], pressure to eat [t (1, 437) = −3.75, p < .001], monitoring [t (1, 450) = −3.67, p < .001], and demandingness [t (1, 407) = −5.95, p < .001] as well as significantly lower responsiveness [t (1, 412) = 3.05 , p = .002]. Cohen’s d effect sizes were small for all significant difference expect for demandingness, which demonstrated a medium effect (See Table 4). For feeding styles, FI was associated with a greater occurrence of authoritarian (Z = 2.7) and less frequent occurrence of indulgent (Z = −1.9) feeding styles (χ² = 22.63, p < .001) than ECETs with FS as examined by chi-square tests of independence. These findings support Hypothesis 2 which predicted an association between FI and ECET feeding practices.

Discussion

The purposes of this study were to examine the performance of the CFQ and CFSQ in an ECET sample, compare properties of these instruments in ECETs to published parent samples, and examine relations between FI and ECET feeding practices. Mean patterns and internal consistency values on the E-CFQ/E-CFSQ were similar to those observed in parent samples on the CFQ and CFSQ in published studies. However, despite seemingly small differences between means for parents and ECETs, significant statistical differences were observed that were accompanied by medium to large effects for most means. Internal consistency values for the scales all were high and consistent with those observed for the corresponding scales for parents in published work. ECETs exhibited similar relations between E-CFQ means and E-CFSQ feeding styles as documented in parents on the CFQ and CFSQ with only one minor difference, providing support for concurrent validity. These findings support the use of the E-CFQ/E-CFSQ instrument for measuring important elements of feeding practices for ECETs and support the modification made to the wording of the CFQ and CFSQ to develop versions suitable for ECETS. Further, this study illustrates that FI among ECETs may be related to feeding practices in important ways.

Despite evidence of similar performance of the scales in ECETS as in parents (e.g., internal consistency, mean patterns), mean differences between ECETs in this study and parents in prior studies are notable. These differences highlight considerations both for future measurement and for understanding children’s feeding environments. For measurement purposes, future studies using the E-CFQ and E-CFSQ, particularly to assess interventions, can have baseline information about how parents and ECETs differ in their feeding practices independent of intervention. This can contribute to more accurate conclusions about intervention effects. Prior studies using the CFQ and CFSQ in ECETs did not examine these differences or compare properties of the instruments in parent and ECET samples (Dev, McBride, et al., 2014; S. Hughes et al., 2007). The current study provides support for their adaptation and use as well as data to suggest differences to anticipate for parents and ECETs. The study also lays the groundwork to understand how the combination of feeding environments that children experience in their lives may interact to predict child eating behaviors and diet. Specifically, the findings of this study suggest that children are likely experiencing different feeding practices from adults at home and at school. It is unclear how these inconsistencies may impact children. For example, studies have not explored how an indulgent feeding style at home may interact with an authoritative style at school to have unique influences on children. However, consistent measurement tools will provide a foundation to explore such questions.

The pattern of ECET feeding found in this study reflect that, despite center policies, educator training, and feeding program guidelines, differences in approaches to feeding are present for ECETs, similar to parents. In parenting research, detrimental feeding practices (e.g., restricting feeding, pressure to eat) contribute to negative outcomes for children including overeating and increased weight (M. Faith, Scanlon, & Birch, 2004). Positive feeding approaches (i.e., authoritative feeding styles) relate to improved outcomes such as increased fruit and vegetable consumption (Rodgers et al., 2013). Similar exploration of the link between ECETs feeding practices and child outcomes may be facilitated by use of the E-CFQ/E-CFSQ. For example, the effect sizes for influence of ECET practices on child outcomes (e.g., dietary practices, self-regulation) could be compared with those from prior studies with parents. Further, the data provided here for the E-CFQ/E-CFSQ can support studies that compare the impacts of parents’ and ECETs’ feeding practices on children across time. Specifically, researchers could examine feeding practices of parents and ECETs for the same child and examine unique and interactive effects on child outcomes.

Similar to previous studies and reports (McKelvey et al., 2018; Song et al., 2016; Swindle et al., 2018), ECETs in the current study experienced FI at rates much higher than the US population as a whole (United States Department of Agriculture Economic Research Service, 2017). In fact, this study documented the highest food insecurity rate among ECETs to date. Nearly half of ECETs experienced FI in the last year. Thus, a vast number of the early care and education workforce are impacted by the experience of inadequate food. Concern about the low wages of this workforce and the consequences for the women majority who fill these roles has been well-articulated since the late 1990s. (as cited in Bronte-Tinkew, Zaslow, Capps, Horowitz, & McNamara, 2007; Whitebook, Phillips, Howes, 2014). The disproportionate burden of FI among women in the early care and education workforce may be contributing to gendered poverty in our communities, where access to childcare is a constant need (Semega, Kollar, Creamer, & Mohanty, 2019). Further, food insecurity is likely to be entangled with other documented poor outcomes of ECETs including high rates of depression and poor physical health (Whitaker, Becker, Herman, & Gooze, 2013). Previous studies have highlighted that ECETs feel undervalued (Shdaimah, Palley, & Miller, 2018), and feelings about the unjustness of their salary impact the care they provide in the classroom (e.g., lower emotional support; Cassidy, King, Wang, Lower, & Kintner-Duffy, 2017). Comprehensive policy and interventions to address FI, improve wages, and support the health of ECETs have the potential for robust impacts on adults and children in early care settings alike.

ECETs FI in our study was associated with distinct feeding patterns. Specifically, ECET FI demonstrated associations with both scale means and feeding style categorizations such that FI was linked with higher levels of perceived responsibility, concern about child weight, pressure to eat, restriction, monitoring, and demandingness. ECETs with FI were also significantly more likely to be Authoritarian in their feeding practices. In parents, higher concern about child weight is associated with greater restrictive feeding (Branch et al., 2017); restriction predicts increased weight status (Faith, Scanlon, Birch, Francis, & Sherry, 2004) and eating in the absence of hunger (Birch. Fisher, & Davison, 2008); pressure to eat is associated with lower fruit and vegetable intake (Fisher, Mitchell, Smiciklas-Wright, & Birch, 2002); and Authoritarian feeding style predicts lower availability of fruits and vegetables as well as lower vegetable intake for children (Patrick, Nicklas, Hughes, & Morales, 2005). Thus, our data suggest that ECETs who experience FI are more likely to feel responsible for feeding children in their care but may rely on detrimental practices to fulfill this role (i.e., pressure, restriction), In extant parenting literature, FI has demonstrated an indirect effect on feeding practices through overall parenting and through increased parental concern about child weight (Bronte-Tinkew et al., 2007). That ECETs experience FI at such a high rate and that many of the feeding practices associated with FI have been shown to be detrimental in parenting research suggests an important area of concern for the early care and education environment.

This study has limitations and strengths. First, the feeding instruments used in this study, although adapted with the childcare setting in mind, were not designed to capture the nuance of feeding in a group setting. Other instruments designed for that purpose could be paired with the ones used in this study to provide a fully nuanced understanding of ECET feeding practices. Another possible limitation of the study is the focus on two Southern states with high rates of food insecurity; findings may not generalize to other areas. Further, this study did not aim to conduct demographic comparisons. Future studies with this purpose could be designed with sampling methods to ensure adequate representation from a diversity of racial and ethnic perspectives and to examine measurement invariance between groups. However, while they may be some limits in generalizability, this study did includes groups who are typically most at risk for food insecurity. A key strength of this study is a large sample size, which far exceeds the sample sizes for the studies in which the CFQ and the CFSQ were validated. Further, this study recruited from two states and resulted in a demographic mix of participants that strengthens external validity of the study, at least for the states targeted.

In sum, our study provides data on the properties and performance of the E-CFQ/E-CFSQ for use with ECETs while demonstrating that food insecurity was related to responses on the tool. In addition to the opportunities for further research with the E-CFQ/E-CFSQ discussed prior, there is substantial room for further work on the correlates of FI for ECETs. For example, future research should consider exploring the role of FI on observed ECET mealtime behaviors and uptake of classroom-based nutrition programs. Interactions with food in the classroom may act as an emotional trigger for FI ECETs and promote use of detrimental feeding practices (i.e., pressuring children, comparing children, discouraging food manipulation (Gibson et al., 2012; Ward, Bélanger, Donovan, & Carrier, 2015). Testing interventions aimed at ECETs designed to modify negative responses and replace them with evidence-based practices could be beneficial to the field. While this would be beneficial knowledge for all ECETs, it will be especially beneficial for those who have experienced food insecurity and for whom that experience increases the use of detrimental feeding practices. Interventions to target ECETs personal factors that are driving their feeding practices can be designed with sensitivity and understanding. Another possibility is that a full de-implementation approach aimed at reducing/removing detrimental feeding practices is a necessary precursor to implementation of classroom-based nutrition interventions (Montini & Graham, 2015). This would require addressing underlying root causes of FI and FI experiences. Replication of this study’s results across diverse types of childcare settings and in other geographic locations will be key to informing policy changes and supporting early childhood ECETs related to their experiences with FI.

Ethical Statement.

This study was conducted according to the guidelines laid down in the Declaration of Helsinki and all procedures involving research study participants were approved by the University of Arkansas for Medical Sciences (IRB # 204c384). Written consent was waived by the institutional review board because it would have been the only document linking participants to the study. The survey included an informational sheet informing participants of the purpose of the study and their participation as well as the anonymous and voluntary nature of the survey. They were informed there were no consequences for electing not to participate or for withdrawing at anytime.

Abbreviations:

ECET

Early care and education teacher

USDA

United States Department of Agriculture

AFC

About Feeding Children

CFQ

Child Feeding Questionnaire

CFSQ

Child Feeding Styles Questionnaire

EFS

Educator Feeding Survey

FS

Food Secure

FI

Food Insecure