The Nuts and Bolts of SMS for Parental Engagement
How to cite
Lichand, Guilherme; Madeira, Ricardo; Bettinger, Eric; Cunha, Nina. 2016. "The Nuts and Bolts of SMS for Parental Engagement." AEA RCT Registry. AEARCTR-0001381. https://doi.org/10.1257/rct.1381-1.0
Download: BibTeX · RIS — for reference managers without a browser translator. Zotero and most others pick up this page's citation metadata directly.
* Primary investigator
Abstract
A growing education literature suggests that supporting parents through text messages (SMS) can positively impact students’ behavior and educational attainment. While those studies highlight the potential of text messages for producing cost-effective educational results, there is limited evidence on the optimal design of SMS campaigns. What it the optimal frequency of texting, so as to most effectively capture parents’ attention without saturating it? At what time should messages be sent? Should parents get messages always at the same time? Is interactive content more effective? The answers to those questions are critical as governments and international organizations consider scaling up successful SMS interventions. This paper cross-randomizes different features of the design of a typical SMS campaign targeted at making parenting a habit among families of public schools’ 9th graders in Brazil. Those experiments assess the impacts of alternative campaign parameters: (i) frequency , (ii) time of the day, (iii) consistency of time of delivery, and (iv) interactivity, on student’s attendance, grades, and drop-out rates.
Intervention
Typical SMS campaign targeted at making parenting a habit among families of public schools’ 9th graders in Brazil.
Experimental design
Campaign parameters are randomly assigned at the student level, comprising a sample of 6000 students within of 180 classrooms at 60 Brazilian public schools.
Assignment to each treatment branch across the four experiments is cross-randomized, except in what comes to the control group, since those receiving no messages cannot be assigned to other campaign parameters.
Experiment 1 randomly assigns the frequency at which SMS messages are delivered. The control group receives no messages. The decision to assign 1/3 of the sample to this group is based on maximing power for Experiments 2 through 4. Treatment 1A (1/3 of the remaining subject pool) receives 1 message a week, a suggestion of activity for parents to do along with their children (delivered on Wednesday). Treatment 1B (also 1/3 of the remaining subject pool) receives 2 messages a week, a ‘fact’ with information about how an activity is linked to children’s development (delivered on Monday) and a suggestion of activity for parents to do along with their children (delivered on Wednesday). Treatment 1C (also 1/3 of the remaining subject pool) receives 3 messages a week, a ‘fact’ with information about how an activity is linked to children’s development (delivered on Monday), a suggestion of activity for parents to do along with their children (delivered on Wednesday), and a reinforcement of that activity, which tries to make it a habit (delivered on Friday).
Experiment 2 randomly assigns the time of the day at which messages are delivered. Treatment 2A (1/3 of the sample) receives messages at the evening (7pm), while Treatment 2B (also 1/3 of the sample) receives messages at the afternoon (noon).
Experiment 3 randomly assigns the consistency of SMS delivery. Treatment 3A (1/3 of the sample) receives messages at always the same time of the day (either noon or 7pm), while Treatment 3B (also 1/3 of the sample) receives messages at alternating times (at the scheduled time, 1 hour before and 1 hour after, following a 3-week cycle).
Last, Experiment 4 randomly assigns whether content is interactive. Treatment 3A (1/3 of the sample) receives a follow-up message (delivered on Thursday) asking whether the parent complied with the activity suggested the day before – to which parents can reply ‘yes’ or ‘no’ –, while Treatment 3B does not receive follow-up messages.
Design details
Cited by
No citation data available for this registration yet.