Skip to main content
ARS Home » Plains Area » Houston, Texas » Children's Nutrition Research Center » Research » Research Project #431426

Research Project: The Development of an Objective Measure of Infant Temperament

Location: Children's Nutrition Research Center

2018 Annual Report


1a. Objectives (from AD-416):
Objective 1: Collect descriptive data on behaviors used to assess infant temperament at 4-months of age. This will be used to assess the variance in these behaviors at this age. Objective 2: Assess correlations between infant temperament based on direct observation using two protocols measured at 4-months and 6-months of age. Objective 3: Assess correlations between infant temperament based on direct observation using two protocols in each of two settings (laboratory and daycare).


1b. Approach (from AD-416):
Approximately three-quarters of pediatric overweight and obesity cases exhibit overweight and obesity between the ages of 2 to 5 years, indicating that obesity prevention efforts may need to begin in infancy. Infancy offers the earliest point for risk stratification and prediction, meaning that we are more likely to be able to implement interventions that prevent obesity in children, rather than treat an existing condition. To understand the role infant temperament plays in adiposity, good measures of infant temperament are needed. We will develop an efficient and objective measure of infant temperament.


3. Progress Report:
The overarching goal of the current project is to develop an objective measure of infant temperament. To achieve this, the project proposed to collect objective observations of temperament, via video recording, on 30 infants at two time points (ages four- and six- months). The project proposed to (1) analyze the psychometric properties of the temperament observation data; and (2) examine whether the temperament observation data correlates with parent reports of their infant's temperament. This year we addressed Objective 1: Develop an objective measure of infant temperament in a standardized laboratory setting (based on direct observation) and determine what aspects of observed infant temperament are associated with infant temperament as assessed by the main caregiver. To make progress towards these objectives, this year we completed data collection on 30 infants, at four- and six- months as proposed in the plan. Some infants did not complete the full proposed protocol and so we are extending this data collection to additional infants within the plan period. We have also completed follow-up on 12 infants at 12- months, which is additional to the proposed plan but will provide useful information on the stability of the temperament observation measure across time. We held weekly coding meetings among staff, during which questions on the coding of the temperament observation were addressed. For example, how individual coders defined "smile" vs. "laugh" was discussed. From these meetings we provided operational definitions (clear, distinct definitions of each behavior) of all behaviors to coded. The operational definitions have been coalesced into a user operations manual which will be made available to other researchers who wish to use the temperament observation in their own research. All of the videos which form the temperament observation for the 4- and 6-month olds have been coded. All of this information, as well as the accompanying questionnaire information has been entered into databases, using a double coding method for accuracy. We have cleaned the data, created a final database for analysis, and also completed initial analyses of the data. We proposed a factor analysis on the temperament observation data. A factor analysis would reveal which behaviors could be combined to represent an underlying construct of temperament (called a 'domain' in the literature). However, the type of data collected was not suitable for a factor analysis. Therefore, we averaged behaviors across a given task, and used alternative measures to ensure each behavior contributed information towards a common temperament construct / domain. These measures included item total-item correlations, and Cronbach's alpha, both of which suggested that the combined behaviors did each measure, at least in part, the same underlying construct / domain. Summary scores were created and these summary scores correlated with parent reports of temperament. The correlations were in the range of .25-.59 which mirror those seen between different assessments of the same behavior in other areas of child psychology; for example between parent and teacher ratings of attention, or between mechanical assessments of activity level and that of parents. Taken together, this supports the notion that the new temperament observation does measure infant temperament, to an extent. In the next phase of the study, we will disseminate these findings, both at scientific conferences and through submitting publications to peer-reviewed scientific journals. This project will ultimately allow us to measure infant temperament in an accurate way for the first time in four-month olds, and so enable an assessment of the extent to which infant temperament associated with eating behaviors and/or adiposity at this young age.


4. Accomplishments
1. Establishing assessment of infant temperament. An objective measure of infant temperament does not exist and this is a problem because temperament may be an important connection to eating behaviors and obesity at this young age of under 1 year of age. Current available measures of infant temperament are based on parent-reports of temperament and these are often biased and reflect the parent's response to the child's weight; therefore, it is difficult to get an accurate assessment of the extent to which infant temperament and infant obesity are correlated, and not skewed by parental bias. Researchers in Houston, Texas have developed an objective assessment of infant temperament, and demonstrated that the behaviors contributing to the measure can be coded reliably, and correlate with parent reports of infant temperament that is free of parental biases. This will have several impacts: our researchers will be better able to assess the extent to which infant temperament associates with eating behaviors and/or obesity at this young age; and other related researchers will be able to utilize this temperament assessment.