ParentText User Engagement and Effectiveness Study

Research has shown that the modification of parenting programmes from being held in-person to being delivered digitally has been successful. Meta-analytic and systematic reviews carried out in HMICs have illustrated that digital parenting interventions have similar effect sizes to in-person interventions. Although these results are promising, there are few gaps within the literature that have not been addressed. First, the majority of research is being conducted in High-Income Countries (HICs). Therefore, results cannot be generalised to LMICs, where differences in context and culture could have an impact on the effectiveness of digital parenting interventions. Second, parenting interventions are deployed on a variety of different technological platforms, however, a limited amount of research has focused on examining emerging platforms, with the focus primarily being on app or internet-based interventions. Therefore little is known in regards to the effectiveness of more novel intervention platforms, such as chatbots.

Whilst the digitalisation of parenting programmes allows for the widespread dissemination and utilisation of the intervention, engagement rates for digital interventions remain low. Engagement with digital behaviour change interventions can be described as the extent (e.g. amount, frequency, duration, depth) of usage, and as a subjective experience characterised by attention, interest and affect. Low engagement compromises the effectiveness of parenting programmes, as the user is required to enrol and engage with the intervention for the programme to have an influence on behaviour. Intervention developers have begun employing technological platforms, such as chatbots, that are designed with the intention to overcome low user engagement rates.

Parenting research investigating digital intervention components which facilitate engagement remains scarce. Broadening the scope beyond the field of parenting, reviews of quantitative and qualitative research examining digital behaviour change interventions, identified an assortment of behaviour change techniques and persuasive system design principles which have been found to facilitate and impede engagement and effectiveness. However, few consistent patterns emerged regarding what techniques and principles consistently facilitate engagement and effectiveness, and for whom. There is a requirement for the isolated techniques and principles utilised within digital parenting interventions to be tested, to identify which intervention components influence engagement and intervention effectiveness and apply this research to optimise parenting interventions.

The first aim of this study is to optimise user engagement in ParentText and to identify the optimal delivery of ParentText for parents/caregivers from low- and middle-income countries. The investigators will be addressing three research questions:

1. How do different component levels impact ParentText user engagement and retention in terms of a) message personalisation (high/low), b) gendered targeted messages (on/off), c) gamification of module completions (on/off) and d) frequency of messages (high/low)?
2. Are there any interaction effects between different component levels of ParentText on user engagement and retention?
3. How does the effectiveness of each component or component level on user engagement vary by different population characteristics (i.e., parent/child age and gender, parent/caregiver marital status, parent/caregiver relationship to the child, and experience of COVID-19)?

The factorial experiment will be conducted across two countries, Malaysia and South Africa. Within each country, the investigators have local partners who will be involved in deploying the intervention and recruiting participants who take care of a child aged between 0 and 17 years old. Parents/caregivers will be primarily recruited through country-specific UNICEF U-Report platforms and government partners.

The following components have been selected, based on research that concluded that these components have been efficient in increasing engagement with digital interventions.

Component A: Personalisation (High/Low): Half of the participants will receive high levels of personalisation, receiving messages that utilise their (nick)name and the child's (nick)name. Half of the participants will not receive this type of personalisation.

Component B: Gendered Targeted Messages (On/Off): Half of the participants will receive messages that are tailored towards either being a female or male caregiver. The other half will not.

Component C: Gamification of Module Completion (On/Off): Half of the participants will receive a gamified progress update on what modules that they have completed. This will be represented as the House of Support. As the parent completes parenting skills, the layers of the house will fill in, representing their progress. The other half of the participants will not be receiving gamified progress updates.

Component D: Frequency of Messages (High/Low). Half of the participants will receive 3 interactions a day (high intensity), the other half will receive 1 interaction a day (low intensity).

The current factorial experimental trial will randomise participants into the 16 experimental conditions. Although this experiment has 16 experimental conditions, it is not a 16-arm RCT. The purpose of this factorial experiment is to estimate the main and intervention effects between the components and not to compare them to one another.

The study will examine the following hypotheses:

Hypotheses

• Participants who receive high levels of personalisation will show higher levels of engagement in ParentText, than compared to users receiving low levels of personalisation.
• Participants who receive gender-targeted messages will show higher levels of engagement in ParentText, than compared to users receiving no gender-targeted messages.
• Participants who receive gamified progress updates will show higher levels of engagement in ParentText, than compared to users receiving non-gamified progress updates.

Exploratory Hypotheses

• The investigators will explore how the frequency of messages will impact engagement.
• The investigators will explore how the effectiveness of components on engagement will vary dependent on demographic characteristics.
• The investigators will explore interaction effects between experimental components.

The second aim of the study is to examine the preliminary effectiveness of the chatbot adaption of the Parenting for Lifelong Health programme, ParentText on the primary outcome of child maltreatment and the secondary outcomes of positive parenting, parental self-efficacy, parental communication about sexual abuse, financial stress, parental stress, child behaviour problems, IPV, attitudes toward gender roles and IPV, and gender-equitable behaviours. The investigators will be investigating five research questions:

1. How does overall child maltreatment (primary outcome) change over time for parents/caregivers engaged in the ParentText intervention?
2. How do the following secondary outcomes change over time for parents/caregivers engaged in the ParentText intervention: physical abuse, emotional abuse, positive parenting, parental self-efficacy, parental communication about sexual abuse, financial stress, parental stress, child behaviour problems, intimate partner violence (IPV) victimisation and perpetration, attitudes toward intimate partner violence (IPV) and gender roles, and gender-equitable behaviours?
3. How do changes in secondary outcomes mediate change in overall child maltreatment?
4. How do changes in primary and secondary outcomes vary based on different population characteristics (i.e., parent/child age and gender, parent/caregiver marital status, parent/caregiver relationship to the child, and experience of COVID-19)?
5. What is the association between user engagement and change over time in primary outcomes and secondary outcomes?

Exploratory Hypotheses

• ParentText will have a significant impact on our primary outcome, decreasing child maltreatment.
• ParentText will have a significant impact on our secondary outcomes, increasing positive parenting, parental self-efficacy, parental communication about sexual abuse, gender-equitable behaviours and improving attitudes toward gender roles and IPV, as well as reducing financial stress, parental stress, child behaviour problems, intimate partner violence (IPV) victimisation and perpetration.
• The investigators will explore how the impact of ParentText will vary depending on user engagement and demographic characteristics.
• The investigators will also explore interaction effects between experimental components.

The experiment will be conducted in five countries, Malaysia, South Africa, Jamaica, the Philippines and Sri Lanka. Country specific recruitment is outlined below:

South Africa: Investigators will collaborate with Clowns Without Borders South Africa and the Department of Social Development to deploy and recruit participants in the Northern Cape. Participants will be recruited by radio announcements. The investigators are also collaborating with UNICEF South Africa who will assist in recruiting participants over the age of 20 via their U-Report Platform (N = 79,800). UNICEF South Africa will send U-reporters an unsolicited recruitment message that is accompanied by a link that takes users to the landing site of ParentText.

Malaysia: 20-35 participants will be recruited by our implementing partner, Generasi Gemilang from their family programmes. Participants will also be recruited either by UNICEF via their U-Report Platform (N = 25,200) or through the National Population and Family Development Board (LPPKN), which will blast messages to individuals who have previously participated in the LPPNK parenting workshops (N = 20,000).

Jamaica: Participants will be recruited in collaboration with UNICEF Jamaica and Parenting Partners Caribbean as part of the Spotlight Initiative, a community-wide intervention aimed at reducing violence against women and girls. Participants will be recruited in four parishes (N = 1,000). Parents and caregivers from the Spotlight initiative will be recruited by community youth activists, leaflets, radio announcements, posters, and other community-wide activities.

The Philippines: Participants will be recruited in collaboration with the Philippine Department of Social Welfare and Development. 30 families will be targeted who are enrolled in a conditional cash transfer programme (4Ps) who live in the Valenzuela municipality of Metro Manila.

Sri Lanka: Participants will be recruited in collaboration with the Health Promotion Bureau. Over 492,640 people follow this channel. The population includes media personal, health care service providers, community leaders, employers at all levels of industry and their organisations: hotels, supermarkets, banks, Ceylon Employer's Federation and the general public who uses Viber social media in Sri Lanka. The Health Promotion Bureau will select randomly individuals who are over 18 to receive a recruitment message to ParentText.