Evaluating Learning Interactions (ELI)

The ELI intervention will send 3–4 text messages per week to parents in two treatment groups. One treatment will send parents messages promoting unconstrained learning interactions, such as guides to conversation and creative activities that parents can do with their children. The second treatment will send parents messages promoting constrained learning interactions, such as literacy lessons on letter recognition, phonics, and vocabulary. We will not have a control group because the intent of ELI is to test the relative efficacy of the two different approaches to encourage parents to interact with their children to promote school readiness in the domain of literacy skills. We focus only on literacy skills so that we can concentrate messaging on only one skill to maximize impact and minimize the length of the assessments. Each treatment will be delivered over six months by text messages in Spanish and English.

Treatment 1: Messages promoting unconstrained learning interactions. Perhaps the most important unconstrained learning activity for parents to engage in with their child is simple conversation. Research shows that differences in language input result in disparities among children in lexical and grammatical development, both within and between socioeconomic groups (Hart & Risley, 1995; Hoff, 2003; Huttenlocher et al., 2010; Pan et al., 2005; Rowe & Goldin-Meadow, 2009). Parents in this treatment group will receive 3–4 messages per week that are guides to conversation and creative activities that parents can do with their children. Examples of messages taken from our PEAKS pilot as follows: “Ask [CHILD NAME] ‘If you could have one superpower, what would it be?’ Text back what superpower [CHILD NAME] chose!”; and “Let [CHILD NAME] choose a story and read it together! Text back the title of the book [CHILD NAME] chose.” Over time the messages also include encouragement to expand the conversation, such as “Ask [CHILD NAME] ‘If you could have one superpower, how would you use it to help others?’”; and “Let [CHILD NAME] choose a story and read it together! And then ask [CHILD NAME] to tell you what happened in the story in his/her own words!”

Treatment 2: Messages promoting constrained learning interactions. This treatment will send parents text messages each week that provide specific literacy-related lessons for parents to engage in with their child. Consistent with the Head Start Approaches to Learning, these lessons will focus on letter recognition, letter sounds, and vocabulary. For example, a text to parents in this treatment might be as follows: “Today focus on the letter B. Help your child think of 4 words that start with the letter B.”

Throughout both treatments we will occasionally send parents “engagement prompts.” An engagement prompt is a request for parents to engage their child in an activity or a conversation and then text back a response indicating that the task was completed. An example of an engagement prompt for the unconstrained learning interaction treatment groups is as follows: “Ask [CHILD NAME], ‘What would be your perfect afternoon?’ Text back the response.” For the constrained learning interaction treatment groups an engagement prompt might be the following: “Ask [CHILD NAME], ‘What rhymes with big?’ Text back the response.”

We will also employ additional strategies to boost and maintain parents’ engagement with the program. In both treatment groups we will use parent identity salience framing for many of the messages because, as noted above, this framing improved parent involvement in a prior text-based parent engagement program that we evaluated. An example of identity salience framing is this: “Be the parent you want to be for [CHILD NAME]!” This is then followed by the message particular to the treatment group of the parent.

In both treatment groups, we will also provide inducements throughout the program to boost attention to the messages. These inducements can include texting back to parents when they respond to an engagement prompt, providing recognition to parents who respond, and so on. These will be the same for both treatment groups.

2022-11-28

2023-05-31

We will assess the change in children’s literacy skills and their change in curiosity about future learning. Reliable administrative data for preschool-aged children’s skill development does not exist, and this study will include some children who are not enrolled in formal preschool; therefore, we will have to assess children’s skills as part of ELI. Our research team has extensive experience conducting child skill assessments.

At baseline and follow-up we will briefly survey parents on demographic characteristics, attitudes toward learning interactions at home with their child, and investments in their child’s school readiness. In the survey at the end, we will ask for user feedback on their experience with the ELI messaging program. We will also look at differences in the rate of parent feedback to periodic engagement prompts.

We will recruit low-income families with preschool-aged children in collaboration with our implementation partners. All recruited parents will be offered half of their incentive payment upon enrollment in ELI and the other half upon completion of the postintervention parent survey and child assessments. Consent will be actively obtained from participants and contact information and demographic data will be collected upon enrollment. Enrolled parents will be randomized into one of the two treatment groups. Baseline parent surveys and child assessments of literacy and curiosity will be administered just prior to the start of the intervention. As part of the 6-month intervention, parents will receive 3-4 text messages per week in their native language of English or Spanish promoting either constrained or unconstrained learning interactions with their children. Periodically parents will be asked to text back a response indicating they are engaged with the ELI program. At the end of the intervention, endline parent surveys and child assessments will be administered.

Consented and enrolled parents will be randomized using a computer program - likely Stata.

Randomization will be at the family level.

No

500 families

500 families

250 families in treatment promoting unconstrained learning interactions; 250 families in treatment promoting constrained learning interactions

We set power to 80% and the significance level to p=.05. MDES= 0.1 if Treatment 1 (T1) scores twice as many points as Treatment 2 (T2) as long as T2 answers 10% of questions correctly. When the difference between T1 and T2 drops to 50%, we would be able to detect the treatment effect as long as T2 answers 25% of questions correctly.