00:02 OK,
00:02 I think we
00:03 started.
00:04 Um,
00:05 so,
00:05 welcome everyone.
00:06 Um,
00:06 thanks for joining us for this seminar at what I know is a very busy time,
00:10 at least at the,
00:11 at the World Bank.
00:12 Um,
00:12 we're delighted today to have
00:14 Sean Powers and Leli Dinata Diaz joining us to present
00:19 their trial from Jamaica
00:21 that was focused on violent behavior,
00:23 a violent discipline and parent parental behavior in Jamaica.
00:27 Um,
00:28 so this trial was part of our COVID recovery window,
00:31 um,
00:31 that we launched in like after,
00:33 after the COVID pandemic,
00:35 um,
00:35 because around the world we saw that,
00:37 um,
00:38 children
00:39 declined
00:40 both in terms of their cognitive development,
00:42 but also in their social-emotional skills
00:45 and their mental well-being,
00:46 um,
00:46 and so,
00:47 So we had a window solely devoted to looking at
00:50 what could quickly recover
00:52 um some of these skills that children lost during this period.
00:55 And then so this trial has been focused on kind
00:58 of recovering the social emote the loss and social-emotional skills
01:01 um through the,
01:03 the,
01:03 the pathway of of violent discipline at home.
01:07 Um,
01:07 and what's really nice about this,
01:09 as you'll see,
01:10 um,
01:11 what's really nice about this trial
01:13 is that it really follows the causal pathway,
01:15 um,
01:16 so it looks at,
01:17 it's a,
01:18 it's a parenting intervention
01:20 that looks at what happens to parents' attitudes,
01:22 what happens to their behavior,
01:24 what happens to their mental well-being,
01:26 and finally what happens to,
01:28 um,
01:28 the skills that children are showing.
01:30 Um,
01:31 and then as we,
01:32 as usual for our seminars whenever possible,
01:35 we also have,
01:36 um,
01:37 an introduction,
01:39 um,
01:39 by the
01:40 task team leader who was leading our dialogue,
01:42 um,
01:42 on education
01:43 and early childhood development
01:45 with the government of,
01:47 of
01:47 Jamaica at the time.
01:49 Um,
01:49 and so we'll have a presentation both by Sean,
01:52 who was our task team leader at the time,
01:55 and Leli,
01:55 who is,
01:56 um,
01:56 one of the principal investigators of the trial.
01:59 Um,
01:59 and then I think we should have time for questions,
02:02 um,
02:02 and then we can wrap up.
02:03 Um,
02:03 and so I will pass it on to Sean who is joining us
02:08 from Sydney
02:09 at one a.m. in the morning.
02:11 Thank you for being here,
02:11 Sean.
02:14 Thank you so much,
02:14 Alika,
02:15 um,
02:15 and it's a pleasure to be here.
02:17 Um,
02:18 Alisa is gonna be doing the,
02:19 the interesting part of the presenting,
02:21 so I will be,
02:21 be very brief,
02:22 but as Alika said,
02:23 I'll give a little bit of a background
02:25 on our dialogue in Jamaica,
02:27 and also,
02:28 um,
02:28 maybe a little bit of,
02:29 uh,
02:30 forward-looking,
02:31 um,
02:32 about how the clients are continuing to use this work and also
02:35 how it's influencing some of our other operations with the World Bank.
02:39 So,
02:40 um,
02:40 Jamaica has been a pioneer in the early childhood space for a very long time.
02:45 Um,
02:46 going back as far as the 1930s,
02:48 there was a grassroots movement there to
02:51 establish,
02:52 uh,
02:52 ECE classrooms and play spaces through,
02:54 through churches,
02:56 and of course,
02:56 they're very well known for,
02:58 um,
02:58 the long-running Jamaica home visit program randomized control trial.
03:02 Um,
03:03 where,
03:03 uh,
03:04 nurses,
03:04 um,
03:05 in the 1980s
03:06 provided,
03:07 uh,
03:07 guidance to parents on how to,
03:09 uh,
03:10 promote,
03:10 um,
03:11 stimulating learning activities for young children,
03:13 and
03:13 those children,
03:14 of course,
03:14 have been followed for,
03:15 for many decades into adulthood with quite remarkable effects on their,
03:20 their outcomes.
03:21 Um,
03:21 we have been involved,
03:23 uh,
03:23 with the Early Childhood Commission.
03:26 Which is um a coordinating agency under
03:29 Jamaica's Ministry of Education,
03:31 um,
03:31 at least since the 2000s.
03:33 So,
03:34 uh,
03:34 the,
03:34 the late 0s,
03:36 so,
03:36 uh,
03:36 quite a long time,
03:37 possibly even farther back than that,
03:39 um,
03:40 including a long-running,
03:42 actually 10-year multi-sectoral early childhood development project.
03:46 Um,
03:46 that closed in 2018,
03:48 but we maintained the relationship with the client.
03:51 Um,
03:51 and then of course,
03:53 uh,
03:53 in 2020,
03:54 um,
03:55 we had the COVID pandemic,
03:57 um,
03:57 the shutdowns,
03:58 and,
03:59 um,
04:00 in Jamaica as well as other Caribbean countries,
04:03 they're already very high baseline rates of violence against children,
04:07 and
04:07 of course there as elsewhere in the world,
04:09 we saw,
04:10 um,
04:11 elevated levels of risk for,
04:13 um,
04:13 violence,
04:14 as well as,
04:15 um,
04:15 a very acute
04:17 need to support parents who were experiencing a lot of stress during that time.
04:21 So,
04:21 we had some funding um
04:23 from the UK Foreign Commonwealth and Development Office,
04:27 FCDO
04:28 through an ASA called ELP,
04:30 the Early Learning Partnership Systems Research.
04:32 Um,
04:33 so that was the funding that enabled us to start this work in addition to the
04:37 Support that we got from CF
04:39 and um the main uh
04:41 substantive piece of work was to adapt um a
04:45 toolkit
04:46 uh to support parents through nonviolent parenting
04:49 techniques called the Erie Holmes toolkit.
04:53 Uh,
04:53 so,
04:53 Erie is a little bit of Jamaican slang,
04:55 um,
04:55 it means,
04:56 uh,
04:57 sort of everything is cool,
04:58 peaceful,
04:58 harmonious.
05:00 So,
05:00 Erie Homes was itself a long-running collaboration,
05:03 um,
05:04 between,
05:04 uh,
05:05 some researchers from
05:07 Uh,
05:07 Bangor University in Wales and the University of,
05:10 of the West Indies.
05:12 But there was already this,
05:14 um,
05:14 parent support package,
05:16 and there was a related,
05:17 uh,
05:18 toolkit called Iri Classrooms for,
05:20 for teachers.
05:21 And,
05:22 um,
05:22 really what we were able to do was,
05:25 uh,
05:25 very quickly put together
05:27 a digitized version of that.
05:29 Um,
05:29 where parents were able to access content through their phones and also,
05:33 um,
05:33 access support groups,
05:35 uh,
05:35 that were facilitated by,
05:37 uh,
05:38 officers from the Early Childhood commission.
05:40 Um,
05:41 so,
05:41 since this trial,
05:42 uh,
05:43 this program,
05:44 um,
05:44 has,
05:45 uh,
05:46 continued to be integrated with the
05:48 core business of the Early Childhood Commission,
05:50 um,
05:50 with the goal of getting it,
05:52 uh,
05:52 fully a part of the preschool services through schools rather
05:56 than just something that Early Childhood Commission officers are,
05:59 are able to deliver cause,
06:00 of course,
06:01 their numbers are limited.
06:03 And,
06:03 um,
06:04 now they are working on having 3 different
06:06 versions that teachers can use based on parents'
06:09 uh preferences and availability.
06:11 So,
06:11 obviously,
06:12 continuing the face to face parenting sessions
06:15 with some supplementation of virtual components,
06:18 or a fully virtual,
06:19 uh,
06:20 version of it,
06:21 um,
06:22 or a kind of a lighter touch,
06:24 face to face,
06:25 um,
06:25 but mostly with virtual sessions.
06:28 And in addition to that,
06:29 um,
06:30 our president,
06:30 Ajay Banga has told us to adapt and steal shamelessly,
06:34 uh,
06:35 across regions with good practices.
06:37 So,
06:37 I don't need to
06:38 tell AI seminar that the
06:40 devil's often in the details with that adaptation from one context to another,
06:44 but,
06:44 um,
06:45 we're in the process,
06:46 um,
06:47 now in the Pacific Islands region where I'm working.
06:49 Um,
06:50 of preparing a regional project that will include parent support,
06:53 and of course,
06:54 we plan to draw on the lessons from this trial.
06:57 And,
06:57 um,
06:58 Lilis is also leading some work which um she may talk about.
07:01 Um,
07:02 so,
07:02 I think much like that original,
07:05 uh,
07:06 parents support RCT in Jamaica,
07:08 I'm hopeful that this one could help inform
07:09 programming in a wide variety of countries.
07:12 So,
07:12 thank you very much again for joining,
07:14 and I'll hand it back to Alika.
07:18 Thanks so much,
07:19 Sean,
07:20 for that context,
07:21 um,
07:21 and it's,
07:22 it's good to know that you're stealing shamelessly,
07:24 uh,
07:25 uh,
07:25 when,
07:25 when,
07:26 when things work out.
07:27 Um,
07:28 so,
07:28 uh,
07:29 why don't I hand it over to Lilith who can tell us,
07:32 um,
07:32 you know,
07:32 more about the trial and,
07:34 and present results.
07:36 Alice,
07:36 over to you.
07:38 OK,
07:38 thank you so much,
07:39 Alaca for organizing the seminar also to,
07:41 to Fay
07:42 and thank you,
07:43 Sean for giving these uh introductory remarks and
07:47 good uh morning,
07:49 afternoon,
07:49 evening to,
07:51 to everyone.
07:51 Thank you so much for joining uh for the seminar.
07:54 Well,
07:55 as I like I was saying,
07:55 today I'll be presenting this paper or this project titled Violent Discipline,
07:59 Parental Behavior,
08:00 Chart and Medium Term Effects
08:02 of Digital Parenting Support to Caregivers,
08:05 and uh this is a joint work with uh some co-authors from the academia,
08:09 Saravana Ravindranang,
08:10 Manisha Shah,
08:11 and,
08:11 uh,
08:12 Helen Baker,
08:13 who is actually also the,
08:14 the developer of the,
08:16 um,
08:16 the intervention
08:17 and also a very,
08:18 a,
08:18 like an honor to be working on with Sean on this,
08:21 on this break as well.
08:23 Um,
08:23 like,
08:24 uh,
08:24 as,
08:24 uh,
08:25 I was mentioning very briefly,
08:26 this,
08:26 uh,
08:27 like the study what aims is to test,
08:28 uh,
08:29 whether
08:29 a digitally delivered parenting program can reduce violence against the children
08:34 in the context of,
08:35 uh,
08:35 Jamaica,
08:37 and,
08:37 uh,
08:37 as he mentioned,
08:38 what we did was to build on these in-person ID homes toolbox
08:42 and,
08:43 uh,
08:43 which was already evaluated and with very important effects,
08:47 and we are testing whether
08:48 if or if delivering it virtually can have similar
08:52 effects.
08:53 and uh by doing it uh digitally or virtual,
08:57 it can also help us to provide some
08:58 evidence in terms of the potential for scalability
09:02 of this intervention.
09:03 OK?
09:04 Let me now jump to
09:06 the motivation of this work
09:08 and the the main idea is like the main question is uh that you
09:13 probably you're not asking,
09:14 but yeah,
09:15 uh,
09:15 that motivated this study is why we want to study violence against children.
09:20 So,
09:20 like,
09:21 some statistics show that from UNICEF showed that
09:24 violence against children can actually remain pervasive around the world,
09:29 just to give you a sense of uh um the prevalence of this
09:33 issue is that around 75% of children with ages between 2 to 4,
09:37 very young kids,
09:38 are regularly subjected to physical punishment
09:41 or psychological aggression at home.
09:44 And these forms of violence can actually are often
09:47 seen as culturally acceptable or even necessary for discipline,
09:51 which was a particular case of,
09:53 of Jamaica,
09:54 and this is what we are also observing in the case
09:57 of Afghanistan where we are also working with a bank team
10:00 um implementing a quarantine intervention.
10:04 Uh,
10:04 the point is that the consequences of this exposure to violence are severe.
10:08 Evidence,
10:08 as you can see in the slide,
10:09 show that
10:10 early exposure to violence can harm
10:12 kids' cognitive and socio-emotional development.
10:15 It also can have some consequences in the long term,
10:18 such as an increase in the likelihood of risky behaviors
10:21 and also uh some um negative effects on academic performance.
10:25 And even in the very long term when they are adults,
10:28 it can have some negative effects on labor market outcomes and criminal behavior.
10:32 Um,
10:33 in short,
10:33 like the negative effects of child maltreatment of violence against children
10:39 is,
10:39 um,
10:40 an issue of human capital because at the end,
10:43 the argument that we want to make here is that if we reduce violence
10:46 during early childhood,
10:47 then it can have,
10:48 uh,
10:49 important effects not only in the very short term but also even
10:52 in the,
10:52 in the long term.
10:54 So,
10:54 what's the main research question?
10:56 It's actually something very straightforward that we,
10:58 um,
10:59 that we want to answer through this paper.
11:01 We want to know whether digital violence prevention,
11:03 parenting programs can change behaviors of,
11:08 uh sorry,
11:08 knowledge,
11:09 attitudes and behaviors around violent disciplines of
11:12 the caregivers of these very young kids.
11:15 Um,
11:16 in addition to that,
11:16 very basic question,
11:17 whether it works or not,
11:19 um,
11:19 we also examine whether it's,
11:21 um,
11:21 it can have some indirect effects on other outcomes,
11:23 as Alaca was saying,
11:24 we also explored effects on caregivers's mental health
11:27 and also on other kids' uh,
11:29 um,
11:30 behaviors.
11:31 Um,
11:32 and we,
11:33 we observed that or we explored that not only in
11:35 the very short term but also 9 months after.
11:39 And um,
11:40 the novelty,
11:41 we argue that the main novelty of this uh of this intervention
11:45 is that it's a fully digital model,
11:48 uh,
11:48 in the sense that it included three different elements,
11:51 such as
11:52 um the content was delivered through SMS messages,
11:55 a data-free application,
11:57 and also some online,
11:59 uh,
11:59 group sessions that were delivered by um
12:04 officers of the Early Childhood Commission in Jamaica.
12:07 And um like,
12:09 overall,
12:10 what we want to show in this paper is not just whether it works,
12:13 but we also like did an extra effort to try to explore how it works,
12:17 that is to also explore mechanisms,
12:19 and I'll give you some,
12:20 uh,
12:20 some information about that and,
12:22 and we also did the cost-benefit analysis
12:24 just to see whether we can support,
12:26 um,
12:27 uh,
12:28 just to see whether we can support the argument of scalability.
12:31 OK.
12:32 Now,
12:32 this is the,
12:33 let me give you a preview of the methodology.
12:35 Um,
12:36 right,
12:37 so,
12:37 we,
12:38 as uh Sean was mentioning,
12:39 we did this adaptation of the in-person ID home's toolbox
12:44 for a virtual delivery,
12:45 and then this adaptation included the three
12:49 main delivery channels I mentioned before,
12:52 SMS messages,
12:54 access to a data-free app.
12:56 With some complimentary materials and also some of this er er informa
13:02 information sessions with the officers from the ECC in Jamaica.
13:06 Um,
13:07 something that's very important,
13:09 uh,
13:09 in this setting,
13:10 and also because we wanted to make sure that it can be scalable,
13:13 is that we work in a very
13:16 close collaboration with the Jamaica's Early Childhood Commission
13:19 and,
13:20 uh,
13:20 in two ways.
13:21 First,
13:22 they supported the team in the recruitment of the participants,
13:26 and second,
13:27 we,
13:28 uh,
13:28 worked,
13:29 the virtual sessions were,
13:30 were delivered by the officers,
13:32 by the specialists from the ECC
13:35 OK?
13:37 Right.
13:38 So,
13:38 just,
13:38 let me give you a preview of this,
13:40 uh,
13:41 the IDT effects of the intervention.
13:44 So,
13:44 we assess the effect of the parenting program,
13:47 the,
13:47 the bundled intervention,
13:48 the three components
13:50 in three sets of outcomes.
13:52 First of all,
13:53 we find very strong effect
13:55 on a caregiver's knowledge about violence prevention techniques.
14:01 Uh,
14:02 in like,
14:03 uh,
14:03 standard deviations,
14:04 we find,
14:04 uh,
14:05 in the very short term right after the intubation ended,
14:08 that the,
14:09 the learn the knowledge index of the
14:10 treated caregivers related to the controlled caregivers was
14:14 0.52 standard deviations higher and then when we measured that knowledge,
14:19 um,
14:20 nine months later,
14:21 approximately,
14:22 the effects were 0.38 standard deviation.
14:26 Uh,
14:26 we note that in,
14:28 in terms of take up,
14:29 there were some,
14:30 there was a variation across the three channels.
14:33 The messages,
14:34 the,
14:35 like,
14:36 the share of uh reception of messages was actually pretty high,
14:39 close to 92%.
14:41 But then when we compare to the app,
14:43 the takeup was actually very low.
14:45 It was around 10%.
14:47 And then the virtual information session,
14:48 the,
14:49 the takeout was actually moderate.
14:50 It was,
14:51 uh,
14:51 to us it was actually pretty good because we were in the middle of the pandemic.
14:54 Parents were taking care of the kids while they were also working,
14:57 etc.
14:58 and it's close to 56%.
15:01 So it's actually a pretty good,
15:02 uh,
15:02 uh,
15:03 take-up rate considering the,
15:04 the conditions.
15:05 And also it's what we are observing in other virtual,
15:08 uh,
15:08 or online interventions such as a tutoring intervention that we just,
15:12 uh,
15:12 evaluated in Ukraine.
15:14 During,
15:15 uh,
15:15 during the Russia's invasion.
15:18 So,
15:18 that's the first
15:19 set of outcomes.
15:20 We
15:21 do find that uh they learn something and it seems that the take-up is
15:25 uh mainly driven first by the messages and then by the information sessions.
15:30 Second,
15:30 the effects on caregivers' attitudes and behaviors is,
15:35 again,
15:35 like there are some improvements.
15:37 In the very short term,
15:37 we observed a 0.2 standard deviations improvements in attitudes towards
15:42 physical or psychological violence against children
15:45 and 0.2 standard deviations,
15:47 reduction in physical and psychological violence itself,
15:50 like the behavior.
15:52 In the medium term,
15:53 the results are um
15:55 more,
15:56 we still observe some improvements in
15:57 attitudes towards our physical violence index
16:00 um
16:01 and also er in,
16:03 uh,
16:03 but not that many differences in terms of the,
16:06 the attitudes.
16:08 Um,
16:08 and that's something more or less expected in the sense that
16:11 it's easier to modify behaviors that are
16:14 very,
16:14 um,
16:15 it's more difficult,
16:16 sorry,
16:16 to modify attitudes which are usually very rooted.
16:19 It's culturally acceptable to,
16:22 to
16:23 like beat your child,
16:24 but then like the most important part is that we are able to modify the behaviors,
16:29 OK,
16:30 which is what we observe and that the results are persistent over time.
16:34 And then in terms of the secondary outcomes,
16:36 we observed some short-term effects on
16:38 emotional problems of the kids and uh
16:41 the also in the medium term,
16:43 we observed some reductions on or improvements on caregivers' well-being,
16:48 that is uh
16:49 improvement on depression,
16:50 anxiety,
16:51 and,
16:51 and stress.
16:52 OK?
16:53 So,
16:53 that's like a summary of uh um
16:56 the,
16:57 the paper in this the,
16:58 the next sections I want to present today.
17:01 Um,
17:02 let me start with the intervention and the experimental design.
17:08 OK,
17:08 so the intervention I was,
17:09 we have been
17:10 talking a bit about it and uh it's uh
17:14 like the base is the ID Holmes toolbox,
17:16 and evidence-based violence prevention
17:18 parenting program that was developed in Jamaica by Helink,
17:23 which is one of our goals.
17:25 So,
17:25 the original,
17:26 the one that was delivered in person had
17:28 or was implemented in approximately 8 weeks.
17:31 It was an in-person program and it was delivered through community preschools.
17:38 Um,
17:39 the target population were parents of kids between 2 to 6 years old,
17:44 and the,
17:45 the goal is to develop positive parenting practices and reduce,
17:49 uh,
17:49 physical,
17:50 like harsher punishment.
17:52 Note that um
17:54 this type of intervention where you are trying to manage kids' misbehavior,
17:58 they can start no,
18:00 not earlier than 2 years of age,
18:02 because before that age,
18:03 it's just uh like
18:05 kids' brain cannot um
18:08 control their behaviors.
18:09 That starts when kids are 2 years or older.
18:12 That's why we,
18:13 we focus the intervention in that,
18:15 um,
18:16 in that age range.
18:18 Um,
18:19 so the intervention,
18:20 the,
18:20 the Idaho toolbox,
18:22 the original is built around 4 key concepts,
18:26 um,
18:26 building positive relationship with two kids,
18:28 like between parents and the,
18:29 and the children
18:31 through
18:31 praise,
18:32 play,
18:32 and affection.
18:34 And then,
18:35 preventing misbehavior
18:37 by,
18:38 for instance,
18:38 setting clear expectations and modeling,
18:41 modeling appropriate behavior among others.
18:44 And then if preventing misbehavior is not feasible,
18:47 then the,
18:47 the intervention also include a key concept which
18:50 is uh how to manage kids' misbehavior.
18:53 For instance,
18:53 using nonviolent strategies such as redirecting the attention of the kids
18:59 or setting some limits,
19:00 etc.
19:02 And uh the fourth component or key concept is um
19:05 um the intervention delivered some techniques to support it,
19:09 um,
19:10 to help parents to support their kids uh to regulate their emotions,
19:13 which is again,
19:14 something that they can start only when they are 2 years,
19:17 not
19:18 earlier than 2 years of age.
19:20 And uh yeah,
19:22 so that's,
19:22 those are the key concept of the,
19:24 the in-person intervention.
19:26 And actually,
19:27 as I told you before,
19:28 uh,
19:28 this intervention has been,
19:29 the in-person one has been evaluated and the effects are pretty,
19:33 pretty big.
19:34 It's approximately 0.29% deviations improvement
19:37 on uh parental use of balance and then increases in
19:41 positive parent-child interactions and also reduction on behavioral problems.
19:47 Um,
19:47 the point is that,
19:49 uh,
19:49 the in-person model
19:51 is very resource intensive.
19:53 You'll need like trained facilitators,
19:56 the group sessions like in space,
19:58 infrastructure,
19:58 etc.
20:00 So,
20:00 what we wanted to do here is,
20:01 OK,
20:02 we are,
20:02 we know that we are facing a lot of um disruptions in,
20:06 uh,
20:07 like in the possibility of being face to face,
20:10 which includes pandemics,
20:11 like several emergencies,
20:12 pandemics,
20:13 um,
20:14 um,
20:15 events related to climate change
20:17 or er wars.
20:19 So,
20:20 what if we try to understand whether we can modify
20:23 the intervention.
20:24 Uh,
20:24 the,
20:25 the delivery mode,
20:26 sorry,
20:26 of the integration.
20:27 So we can,
20:28 we can see what are the effects
20:30 can be similar or at least still um statistically significant,
20:36 let's say,
20:36 uh,
20:37 when we modify the delivery mode,
20:39 OK?
20:39 So,
20:40 That's what we,
20:42 that's what we did and we adapted,
20:44 we worked with um with Helen,
20:46 a specialist from the ACC
20:49 and uh to adapt
20:51 the integration and we delivered the VIHD,
20:54 the virtual
20:56 ID Hos toolbox.
20:58 So the idea here is that we wanted to keep,
21:00 to keep the content,
21:02 we keep the four key concepts,
21:05 and we also keep um like other specific
21:08 activities that were delivered over 10 weeks.
21:11 And the idea was to extend 2 weeks because the um the level of um
21:17 um to go deeper into the different concepts.
21:20 And uh we also kept the 10 key topics that I'll show you,
21:24 I'll show you in the next slide.
21:26 Uh,
21:26 but the idea was to just to adapt all the content
21:29 using the three delivery channels I,
21:31 I mentioned before.
21:32 The first one was that caregiver received
21:35 these 3 SMS SMS messages per week
21:38 for a total of 30 messages.
21:41 And each of the messages gave very short actionable punching chip,
21:45 like,
21:46 try to give your child clear instructions and
21:48 praise them whenever they follow your instructions.
21:50 Praise encourages positive behavior.
21:53 And the
21:54 Part of short and actionable parenting tip is very important
21:58 because what we also did in another intervention in El Salvador
22:01 was to write very like longer
22:04 uh parenting tips
22:06 in the form of a blog and um
22:09 unfortunately,
22:10 like almost no one read the,
22:11 the,
22:12 the blog.
22:12 So that's what we learned from that experience and
22:15 then here like what within in Jamaica is,
22:16 OK,
22:17 let's try to modify this in a way that's gonna be attractive for,
22:20 for caregivers.
22:22 Um,
22:24 second,
22:25 we,
22:25 each of the messages were linked
22:28 to
22:29 additional content
22:31 that was stored in a data-free app.
22:34 And that content included short videos,
22:36 visuals,
22:37 and some weekly ID challenges,
22:40 which were like some ways in which the caregivers
22:42 could practice the content they were learning at home,
22:45 OK?
22:46 The third part was these virtual meetings with an ECCO specialist.
22:51 Uh,
22:52 so we grouped the participants in,
22:54 uh,
22:55 in groups of 8 to 9,
22:57 uh,
22:58 caregivers,
22:59 and,
22:59 uh,
23:00 so the,
23:01 the groups were led by a trained officers from the ACC.
23:05 Um,
23:05 the length of the session was approximately one hour,
23:08 and the,
23:09 the idea was to reinforce the key lessons,
23:12 uh,
23:12 offer some practice opportunities,
23:14 and also to provide some sort of peer support space.
23:17 That is,
23:17 for instance,
23:18 if I,
23:19 um,
23:20 if like I had the opportunity to also meet other parents in the group and
23:24 try to create some sort of network
23:25 because that could be actually a potential channel
23:27 through which this,
23:28 uh,
23:29 this type of integration can have some effects,
23:31 OK?
23:33 So that's like um
23:35 that's the description of the integration.
23:36 As I told you before,
23:37 we had these 10
23:40 key topics and each of them were delivered in each of the,
23:44 the weeks of the iteration,
23:45 OK?
23:47 Uh,
23:48 it's important to note that the,
23:49 the content of this intervention has a very key concept,
23:53 but at the end,
23:54 um,
23:55 the topics
23:56 were modified and were,
23:58 er,
23:59 adapted to,
24:00 no,
24:00 were not modified,
24:01 but they are actually very Very specific to the Jamaican context,
24:05 they may need some adaptation in case they want to be implemented
24:08 in other settings,
24:09 such as,
24:09 for instance,
24:09 chillax is not a a concept
24:11 that's gonna be understood probably in other settings.
24:14 So they,
24:15 it needs some adaptation,
24:17 but we know that's uh
24:19 that's something
24:20 um that needs to be done anyways,
24:23 OK.
24:24 Yeah,
24:24 here it is.
24:25 OK.
24:26 What about the recruitment of the,
24:27 that's the integration.
24:28 So now let's move to how we recruited the participants.
24:33 Yeah,
24:33 so we collaborated very closely with the ECC but we also have another partnership
24:39 with the Trend Media,
24:41 which is a subsidiary of Digital,
24:44 the largest mobile provider provider in Jamaica.
24:48 So,
24:49 Trend Media,
24:50 the,
24:51 uh,
24:51 the partnership with them was very important in the sense that,
24:53 that allowed us to have access to a larger um
24:58 set of participants that were not necessarily
25:00 already engaged in quarantine integration.
25:03 And um,
25:04 so they,
25:05 and they also help us to,
25:06 for the,
25:07 for the implement
25:08 yeah,
25:09 in the implementation that is the delivery of the messages and,
25:12 um,
25:13 and also the um
25:14 like to set up the,
25:16 the app where we uploaded the materials.
25:19 So,
25:19 in terms of the recruitment,
25:21 Trend Media
25:22 was um
25:24 uh
25:25 what they did was um during 25 days
25:29 uh in mid 2021,
25:32 they sent SMS messages that invited
25:37 um some of their adult subscribers
25:40 across all 14 parishes in Jamaica.
25:43 Uh,
25:44 so the,
25:44 we,
25:45 approximately 6000 people clicked the link
25:49 and approximately 2000 of them
25:51 met the eligibility criteria for the study,
25:54 that is that they were living with the child between 2 to 6 years of age,
25:58 they own a smartphone or tablet,
26:01 and that they also provide a consent to
26:03 participate in the intervention and uh the evaluation,
26:06 OK?
26:07 And out of those,
26:08 then uh once they confirmed the participation,
26:11 then we got their information
26:14 and we reached out to 2 to 2000
26:18 and uh to,
26:19 to start with the,
26:20 with the study and uh we were able to collect baseline data
26:25 from approximately 1,113 caregivers.
26:29 OK?
26:29 And that's gonna be our analytical sample.
26:33 Uh,
26:33 you can see in the papers,
26:35 several checks of how representative is the sample,
26:37 uh,
26:38 from the overarch caregiver in Jamaica,
26:41 uh,
26:41 how representative is the distribution across the different parishes,
26:44 etc.
26:45 and I,
26:47 as you were already expecting,
26:48 this sample is positively selected,
26:50 particularly in terms of,
26:52 um,
26:53 education.
26:54 Um,
26:55 so yeah,
26:55 that's a caveat that you should keep in mind,
26:57 uh,
26:58 when you are thinking about the results and potential challenges,
27:01 etc.
27:02 OK?
27:03 So,
27:03 then,
27:04 um,
27:05 we,
27:05 like,
27:06 the study included uh four main um
27:10 points in time.
27:11 We collected this baseline data in August 2021 before the integration started,
27:16 then the integration um took place between September and November 2021,
27:20 uh for the 10 weeks,
27:22 and then,
27:23 uh,
27:24 we conducted the first short-term follow-up survey in December 2021.
27:28 Right after the integration ended,
27:30 and then we reach out again to the caregivers and conducted a medium,
27:34 medium term follow-up
27:37 in September 2022.
27:39 All surveys were conducted over the phone.
27:42 And then,
27:43 once we completed the baseline survey,
27:45 we randomly assigned the 1,113 caregivers,
27:49 uh,
27:49 half to treatment and half to control,
27:52 and we're certified by caregiver gender.
27:55 And also a recruitment channel,
27:56 uh,
27:57 just because,
27:58 like,
27:58 uh,
27:58 as I told you before,
27:59 the exposure to some activities already developed by,
28:03 let's say,
28:03 the ACC
28:05 where uh can actually explain some of the,
28:07 some of the results,
28:08 so that's why we wanted to clarify by recruitment channel.
28:11 At the end,
28:12 approximately 90% of the participants of the 1000 participants were recruited
28:17 through umre media.
28:20 It's uh approximately 10% who are recruited through the ECC channels
28:25 and also the um the preschool principals also help us to,
28:28 to send some invitations.
28:31 Uh,
28:31 right,
28:32 so what,
28:33 um,
28:33 it's clear what the caregivers in the treatment group received,
28:37 which was like the SMS messages,
28:39 uh,
28:40 the access to the app,
28:41 and then the option to join these online sessions.
28:45 Since we wanted to keep the control group engaged,
28:48 we also,
28:49 um,
28:50 uh,
28:51 sent them the SMS messages
28:54 and the content was related to COVID-19 safety practices.
28:59 Um,
29:00 so yeah,
29:00 which is,
29:01 uh,
29:01 orthogonal to what we were trying to,
29:04 or what,
29:04 what's included in the as the content in the intervention,
29:07 OK?
29:08 It was just to keep them engaged with,
29:10 with us.
29:11 Um,
29:13 right,
29:13 so overall the ronomization worked pretty well
29:16 and we have balance across demographic,
29:19 demographics and the baseline outcomes,
29:22 um,
29:22 with no significant differences across the 27
29:26 variables that were included.
29:29 OK,
29:29 before,
29:30 um,
29:31 giving,
29:31 giving you more details about
29:34 the specific data we collected in the
29:37 three rounds of data collection,
29:39 let me try to give you an idea of what we are,
29:41 um,
29:42 what's our conceptual framework here.
29:45 As you can see in the video,
29:45 like,
29:46 we know that there are these high levels of violence against children
29:49 in households in low and middle-income countries,
29:51 OK?
29:52 Um,
29:53 which are 75% more or less.
29:55 So,
29:55 what
29:56 our integration is this information-based
30:00 positive parenting program
30:01 that was delivered
30:03 er through messages,
30:05 apps and online meetings.
30:08 Then our or the theory of change or the arguments behind the the potential
30:14 change pathways of this intervention are mainly three.
30:18 The first one is the caregivers
30:19 are gonna learn
30:21 some specific content
30:23 and practices to reduce violence,
30:24 violent behaviors.
30:26 Um,
30:27 we also think that by giving them some tools
30:30 on how they can manage their kids' behaviors,
30:32 for instance,
30:33 that's gonna increase their self-efficacy,
30:35 uh,
30:36 related to parenting.
30:38 And then,
30:39 by having the opportunity of meeting other caregivers in the,
30:42 in the virtual sessions,
30:44 then um there can be some effects driven
30:47 by the creation of these networks of parents,
30:50 OK?
30:50 So,
30:50 those are the main uh change pathways.
30:54 And then we argued that through these channels,
30:57 we can observe some effects on what we
31:00 call the direct outcomes and indirect outcomes.
31:03 The direct could be
31:04 Uh,
31:05 an increased knowledge and then changes on attitudes,
31:09 and finally changes on behaviors,
31:11 OK?
31:13 Uh,
31:13 it can also have some indirect effects
31:16 such as improvement on kids' behaviors,
31:19 uh,
31:20 but also like some effects on
31:23 caregivers' well-being.
31:25 Uh,
31:25 if,
31:26 uh,
31:26 for instance,
31:26 now I,
31:27 I know how I can manage my,
31:30 uh,
31:30 kids' tantrums,
31:31 then
31:32 I'll probably be less stressed
31:35 or I can be more,
31:37 um,
31:38 yeah,
31:38 less anxious,
31:38 etc.
31:40 And uh that's important in the sense that we,
31:43 and we are,
31:44 sorry,
31:44 uh we are
31:45 uh calling this parental well-being as secondary outcomes
31:48 in the sense that they are not,
31:49 um,
31:50 because that's not the main goal of this intervention.
31:52 It did not,
31:53 uh,
31:54 aim to modify,
31:56 uh,
31:56 parental well-being,
31:57 which is something different to what we did in,
32:00 uh,
32:00 El Salvador and what we're trying to do in Afghanistan.
32:03 It's just trying to,
32:05 uh,
32:05 to provide tools to the parents and how they can reduce their violent behaviors,
32:08 OK?
32:11 OK,
32:11 so given this conceptual framework,
32:13 we collected
32:15 er this,
32:16 I'm sorry.
32:18 We collected these three sets of outcomes.
32:21 The first one is uh caregivers' outcomes which include,
32:24 as I told you before,
32:25 um,
32:26 attitudes to violence against children,
32:28 violent behaviors,
32:30 and also outcomes related to,
32:31 to their mental health,
32:33 including depression,
32:34 anxiety,
32:34 and stress.
32:36 All these,
32:37 no,
32:38 sorry,
32:38 the instruments that we use to measure attitudes
32:42 and behavior were based on the UNICEF,
32:44 the mixed questionnaire,
32:45 but it was actually adopted by a bank team
32:48 and
32:49 uh
32:50 to be collected over the phone.
32:52 Then we also collected uh outcomes from the kids,
32:56 but they were reported by the,
32:57 by the parents.
32:59 Um,
32:59 and we collected information on child emotional conduct problems
33:03 using the strengths and difficulties questionnaire.
33:06 Again,
33:06 it was also adopted by a bank team
33:08 for um
33:10 uh for surveys or for data collection conducted over the phone.
33:13 And then for the mechanisms,
33:15 we collect data on parental self-efficacy and these uh caregivers uh networks.
33:20 Um,
33:21 to measure learning,
33:23 we developed a module just to ask about
33:26 the specific con uh concepts of the integration,
33:29 and we also collected from the,
33:31 from trained media administrative information on the message delivery,
33:36 app usage,
33:37 and also the attendance uh to the organizations,
33:40 uh,
33:40 which actually the attendance was provided by the,
33:43 um,
33:43 the ECC officers.
33:46 Right.
33:46 So,
33:47 um,
33:48 let's just start with some of the,
33:50 the main,
33:51 the,
33:51 the results.
33:52 Let's first to see whether these interventions is gonna,
33:56 uh,
33:56 have some the compliance that,
33:58 that's expected and also whether the take-up is gonna be good enough so we can
34:02 argue that,
34:02 uh,
34:03 uh,
34:03 participants attended the sessions that received the messages,
34:06 etc.
34:06 and therefore we can see effects of knowledge.
34:09 And as I mentioned at the,
34:12 in the introduction of the,
34:13 the presentation.
34:15 Um,
34:16 so we have the three delivery methods.
34:18 And here what we have,
34:19 uh,
34:20 is,
34:20 uh,
34:20 uh,
34:21 like a data from chain Media
34:23 showed that they were able to deliver the messages to approximately 92% of the,
34:28 of the caregivers,
34:29 OK?
34:30 And then we also asked the caregivers in the survey,
34:33 uh,
34:33 did you receive these messages,
34:34 uh,
34:35 uh,
34:35 about parenting interventions and approximately 91.38%.
34:40 91% of them reported that they did receive the messages.
34:45 These are only the,
34:46 the treatment group.
34:47 Then,
34:48 um,
34:49 out of those uh 92%,
34:52 97% also reported that they read the messages and that they found the message,
34:58 the messages useful.
35:00 Um,
35:01 so that's in terms of the messages.
35:03 Then in terms of the app,
35:04 data from,
35:05 uh,
35:06 red media shows that the takeup was actually very low.
35:09 Um,
35:09 out of the 10 sessions,
35:11 the caregivers,
35:12 um,
35:13 only joined a person or only,
35:15 not joined the session,
35:16 but,
35:16 um,
35:16 only accessed.
35:18 Uh,
35:19 one of the,
35:20 one of the sessions,
35:21 the information in the app for one of the sessions
35:24 on average,
35:25 and they spent approximately 7 minutes
35:29 total over 10 weeks in the app.
35:31 So,
35:32 at the end we can confidently say
35:35 that
35:36 any of the effects are not driven by this app.
35:40 And that's something actually that we are still trying to understand,
35:43 not in this project,
35:43 but in other projects that we are trying
35:45 to understand why the take-up of online integration are
35:48 too low.
35:49 And I think that's an important agenda
35:51 because if we want to implement these integrations at scale,
35:54 then we may need to understand what's the specific design of the
35:58 um the app or the online delivery that we'll need in order to,
36:03 to keep uh
36:04 at least a very,
36:06 uh,
36:06 a moderate engagement of the participants.
36:09 Um,
36:10 yeah,
36:11 so,
36:12 note that there is something particular here that
36:15 we conducted focus group discussions with the,
36:18 er,
36:19 some of the participants.
36:20 And what we found in those discussions that some of them
36:24 found that the app was less intuitive
36:27 or that they prefer to receive the information directly through SMS,
36:31 um,
36:33 you know,
36:33 there are these,
36:34 uh,
36:34 like the receiving information through SMS is more like a supply push,
36:38 whereas accessing the app is more likely demanded,
36:41 so at the end,
36:42 we,
36:42 you need to do an extra effort to assigning,
36:45 uh,
36:46 signing in
36:47 to,
36:47 to the app and then looking at the content,
36:49 etc.
36:50 So,
36:50 it's uh like just a supply push seems to be at least in this
36:54 context in this integration so that uh that uh works and or it's easier,
36:58 what's easier for them.
37:00 And then
37:02 finally,
37:02 in terms of the weekly online group sessions,
37:04 the attendance was moderate.
37:06 It was approximately 4.55
37:10 out of the 10 sessions and 79%
37:14 of the participants attended at least one of the sessions.
37:18 And uh at the end,
37:19 this level of participation,
37:20 we think that's um
37:21 quite reasonable,
37:22 as I told you before,
37:23 at the end,
37:24 it was in the middle of the pandemic,
37:25 so,
37:25 uh,
37:26 the caregivers were,
37:27 um,
37:28 had some competing
37:29 responsibilities,
37:30 taking care of the kids,
37:31 uh,
37:31 the household,
37:32 plus work in some cases,
37:34 etc.
37:35 OK.
37:36 OK,
37:36 so the takeout was good for the messages and er acceptable for the
37:42 for the sessions,
37:43 then uh
37:45 let's
37:46 measure whether they learned something because they
37:47 are reporting that they received the messages
37:49 and that they read them,
37:50 that they liked them,
37:51 so let's see whether they learned anything.
37:54 Uh,
37:54 and as I told you before,
37:55 to make sure that we designed this instrument in which we collected,
38:00 we
38:01 collect information for 8,
38:04 through 8 questions,
38:05 sorry,
38:05 and create an index of,
38:07 uh,
38:08 uh,
38:08 I'm sorry,
38:09 a knowledge,
38:10 knowledge index.
38:12 And what we are finding is that in the very short term,
38:14 there are some improvements in knowledge and
38:15 this data for both treatment and control.
38:17 The previous day was only a data for the treatment.
38:20 What we are observing in terms of differences between treatment and control is that
38:24 uh the treated caregivers had a knowledge index
38:29 which is on average 0.53 standard deviation higher
38:32 than the control group,
38:34 and nine months later,
38:36 which is the second panel in this table,
38:38 we also are finding that the results are still pretty,
38:41 pretty strong
38:42 and it's approximately 0.40 standard deviations.
38:45 Uh,
38:46 indicated that at the end that any gain in terms
38:48 of knowledge was actually not a short to leave them.
38:52 Um.
38:53 Then,
38:54 what about,
38:54 what,
38:55 OK,
38:55 so they,
38:56 it seems that they learned something.
38:57 Let's see whether it also changed
38:59 their attitudes and behaviors toward violence.
39:03 And you know that it did,
39:04 but
39:04 let me just show you more details about it.
39:06 What I'm showing you in this slide is the,
39:08 this figure in which here in the vertical axis you have the,
39:11 the outcomes.
39:12 So,
39:13 for both attitudes and behavior,
39:14 we measured like an aggregated uh indicator of attitudes.
39:20 toward violence and then we separated
39:22 the er by attitudes towards physical violence
39:26 and attitudes towards er
39:28 psychological violence.
39:30 So here what I'm showing you is uh the estimated effect,
39:34 uh,
39:34 the symmetrical efficient,
39:35 sorry,
39:36 which is the P
39:37 and then uh the p values for the significance,
39:40 and then uh so the,
39:42 the circle
39:43 is um
39:44 um
39:45 is the short term effect,
39:46 estimated effect,
39:47 and then the triangle is the medium term.
39:50 So,
39:50 overall what we are finding is that uh
39:54 like immediately right after the program ended,
39:57 the carriers in the treatment group scored 0.20% deviations,
40:01 higher on the overall
40:03 altitudes index,
40:05 meaning that they were less supportive of violent discipline.
40:09 Nine months later,
40:11 we're observing uh a still statistically significant effect
40:15 and
40:16 the magnitude is 0.14 standard deviations.
40:20 Um,
40:21 these results are actually driven by
40:23 reductions on attitudes to physical violence,
40:27 which is between 0.19 in the short term and then 0.
40:31 15 in the medium term.
40:33 In terms of psychological violence,
40:35 uh,
40:35 we observed some
40:37 improvements in the short term,
40:38 but those effects dissipate over,
40:40 over time,
40:41 which is as,
40:42 as I was telling earlier.
40:44 Um,
40:45 at the end,
40:46 modifying some more extreme forms of violence seems to be,
40:50 uh,
40:50 working through this intervention but more
40:53 subile or related to psychological violence which are very rooted in terms of,
40:58 in terms of norms,
41:00 then that may be a bit more difficult to modify through this intervention.
41:04 What about behaviors?
41:06 So the structure of the figure is the same,
41:07 and again,
41:08 we collect information on the aggregated
41:11 index of behaviors
41:12 and then on um behaviors
41:15 of physical violence
41:17 and then psychological violence.
41:20 So what we are observing is that in terms of like aggregated index,
41:24 we do observe some improvements in terms of um violent behaviors both in the
41:29 short and medium term and the magnitudes actually do not change after 9 months.
41:33 Which is uh what we more or less are observing
41:36 in both physical and psychological violence.
41:38 Note that
41:39 we,
41:40 it's,
41:41 we argue here that it's a bit different.
41:42 Here it's like probably dissipation of defect,
41:45 uh,
41:45 here,
41:46 not necessarily because at the end the
41:47 magnitude of the coefficients are very similar.
41:50 It's just that the sample in the medium-term follow-up was smaller.
41:54 Therefore,
41:54 that's why we are expecting some increases in the confidence intervals,
41:58 um.
42:00 But in fact,
42:00 we test for differences between the two coefficients and we
42:03 find we cannot reject the hypothesis that the two coefficients are
42:08 similar in terms of magnitude.
42:11 OK.
42:12 Now,
42:12 what about,
42:13 so,
42:13 at,
42:14 uh,
42:14 in the integration worked,
42:16 uh,
42:16 modifying the main outcome which were attitudes
42:20 and then behaviors.
42:22 What about other secondary outcomes,
42:24 caregiver mental health.
42:26 We collected information in the short term of depression and anxiety,
42:30 and then depression,
42:31 anxiety and stress in the middle term.
42:32 What we are observing is that there were some positive effects,
42:35 not in the short term,
42:36 but in the medium term.
42:38 Uh,
42:38 showing that probably it's like that's a potential challenge.
42:41 Like now I have the tools set,
42:42 I am more confident that I can manage or prevent this behavior of my kids,
42:47 and therefore that's gonna help with my mental health.
42:49 OK?
42:50 So that's probably how it's,
42:51 uh,
42:51 how it's working.
42:53 What about kids' outcomes.
42:55 We do observe some improvements,
42:57 particularly on the emotional problems in the short term,
43:00 and,
43:01 uh,
43:01 but those effects dissipate.
43:03 It's like about 1/3 in the medium term,
43:06 uh,
43:06 but we don't observe anything on conduct problems.
43:08 So,
43:09 at the end,
43:09 like,
43:10 um,
43:11 we probably need to,
43:12 to keep these results on kids more sustained.
43:14 We probably need some
43:16 more intensive interventions,
43:17 OK?
43:20 And then for the mechanisms,
43:22 we measured 22 main ones.
43:24 One,
43:24 as I told you,
43:25 parental self-efficacy and the other one were the net networks.
43:28 And we don't find any evidence on that the net net networks were
43:33 um
43:34 uh are potential are driving the,
43:36 the main results.
43:37 We do observe,
43:38 however,
43:38 that parental self-efficacy can,
43:40 can be driving the effects that we observe it like some positive effects,
43:45 particularly in the one related to uh acceptance,
43:49 OK?
43:51 Right.
43:51 Um,
43:52 so,
43:53 we,
43:54 er,
43:54 explore additional,
43:56 like,
43:56 uh,
43:57 alternative explanations and,
43:59 uh,
43:59 which are also some proposal checks
44:01 and
44:03 specifically we,
44:04 um,
44:06 we conducted,
44:07 because all the outcomes were self-reported,
44:09 the potential
44:10 concern that you may have is that uh we are actually um
44:14 er facing some experimental demands.
44:17 Uh,
44:17 effect.
44:17 So,
44:18 what we did was to,
44:19 since we can we included a
44:21 module on social that measures,
44:23 social disability bias,
44:25 what we did was to include it as a control variable and then
44:27 we tested whether the effects were different for parents with higher versus lower.
44:33 Uh,
44:34 social desirability bias and what we find is that the results are the same.
44:38 That is,
44:38 uh,
44:38 it's not that,
44:39 er,
44:40 more socially desirable
44:43 parents or parents with those type of attitudes
44:45 are the ones er driving the positive effects.
44:48 We don't find differences between the,
44:50 the treated groups with high or low social desirability bias,
44:54 OK?
44:55 Uh,
44:56 then another thing,
44:57 another concern of potential discrimination is that
44:59 there was some sort of displacement,
45:01 displacement of violence toward older kids in the household.
45:05 Like,
45:05 for instance,
45:06 if I'm now not,
45:07 uh,
45:07 hitting my child,
45:09 the youngest one,
45:10 maybe I'm actually hitting the,
45:12 the oldest one,
45:13 OK?
45:14 We collected reports from the caregivers of whether they,
45:19 they conducted any violent behavior
45:22 for uh with their kids who were between 7 to 12 years of age.
45:26 And for that sample,
45:27 what we are observing is that at the
45:29 end there is actually a reduction in violent behaviors
45:32 and the,
45:33 the impacts are very similar to er to those in the,
45:38 in the target age.
45:40 And more importantly,
45:42 it's a,
45:43 mm sorry.
45:44 We,
45:44 when we compare,
45:45 when we actually restrict the sample to only those who were,
45:49 uh,
45:50 who had to,
45:51 to,
45:52 um,
45:53 the kids in the two age ranges,
45:55 we,
45:55 we actually observed very,
45:56 very similar effects.
45:59 The third concern
46:01 is that in our setting,
46:03 we have some
46:05 um differential attrition
46:07 in
46:09 in the first follow up.
46:10 So then what we did was to try to adjust use
46:13 the rebounds and also we do a balanced panel estimation,
46:16 the results are robust.
46:18 And lastly,
46:19 uh we use uh we address the sensitivity from
46:22 the selection of control variables using a double lossome.
46:26 Um,
46:27 right,
46:27 so just to wrap up two more slides and then we move to the,
46:31 to the Q&A
46:33 section and uh let me start with the policy implications.
46:38 Uh,
46:39 let me like try to discuss uh what these results may mean for policy and,
46:43 and operations.
46:45 First,
46:46 what we are finding is that
46:48 the digital parenting support can be a very powerful preventive,
46:52 preventive tool against violence,
46:54 particularly for young kids.
46:56 And,
46:57 uh,
46:58 and this is actually relevant because the
47:00 traditional parenting programs faces some constraints,
47:03 that is they have limited staff,
47:04 uh,
47:05 logistics and also the costs.
47:08 With this intervention,
47:09 we show that
47:11 the drug can be delivered remotely.
47:13 It's actually very,
47:14 very low cost,
47:16 very cheap,
47:17 and it can still produce
47:19 a large behavioral changes.
47:20 I added here in the slide to
47:23 um like the comparison in terms of the cost between the in-person and the virtual um
47:30 I homes toolbox.
47:31 The virtual cost per person approximately 62.4.
47:36 And uh then the in-person cost approximately 123.
47:41 Now,
47:41 you're gonna tell me,
47:42 yeah,
47:42 but the in-person actually produced the larger
47:45 effects that which was 0.29% deviations.
47:48 When we try to use like to translate
47:51 those costs into like cost per standard deviation,
47:55 at the end,
47:56 they are very,
47:57 they are very similar.
47:58 That is,
47:58 uh sorry,
47:59 the virtual
48:01 intervention is still very,
48:02 um
48:03 very cost effective
48:05 and it's actually addressing something um
48:08 Something that's important in our setting,
48:10 which is the,
48:11 the availability of a specialist
48:14 who can deliver these interventions in person,
48:16 OK?
48:16 Which is not considered or taken into account in those estimations of the costs.
48:21 Um,
48:23 second,
48:24 we something that we learned is that
48:25 like this close collaboration with the government,
48:28 which was wonderful with the,
48:29 at least with the ECC,
48:31 but also like uh try to um to collaborate with
48:34 private partners in our case stream media or digit software.
48:40 Very important partners in the sense that they help us with the,
48:43 the delivery of the integration,
48:44 the recruitment of the participants.
48:46 Things that in the context of,
48:48 let's say,
48:48 emergencies,
48:49 uh,
48:50 um,
48:50 for governments it's very difficult to,
48:53 uh,
48:53 to manage,
48:54 uh,
48:55 which is also what we did in the case of Ukraine,
48:57 another intervention integration
49:00 where we collaborated with an,
49:01 with an NGO for the implementation.
49:04 And third,
49:05 um,
49:06 the,
49:07 uh,
49:07 the integration also,
49:08 um,
49:09 this digital tools can help us to,
49:11 to reach these parents who are,
49:13 cannot attend in person for different reasons and also like,
49:17 uh,
49:17 it's very important,
49:18 that's what we are trying to test right now in Afghanistan,
49:21 and to see whether these integrations can work in
49:24 settings where mobility is um is very restricted.
49:28 And,
49:28 uh,
49:29 and also literacy is uh
49:31 um is very low,
49:32 OK?
49:32 So we're gonna see whether
49:34 uh the delivery mode can be adapted in order to reach to the most vulnerable parents.
49:39 And lastly,
49:41 something that I think it's still very important to understand,
49:43 particularly in the AI era and how we need to,
49:47 to use digital tools to,
49:49 to deliver integration to,
49:50 to different beneficiaries.
49:52 And it's why the takeup of this,
49:54 uh,
49:55 the apps,
49:55 uh,
49:56 uh,
49:56 is actually not as high as we would uh want to,
50:00 and,
50:01 uh,
50:01 so maybe we need to,
50:03 to explore how the different components of these integrations
50:06 um can be improved to,
50:08 to increase engagement.
50:09 OK?
50:10 And very quickly,
50:11 just to conclude,
50:12 er we're showing this experimental evidence
50:14 of a virtually delivered information uh parenting program and uh
50:19 these are very important,
50:20 has important implications for the scalability of
50:22 this integrations and how we can bring
50:25 this um um these programs er to,
50:29 to very er isolated er participants.
50:33 And yeah,
50:34 so that's a summary of the,
50:35 of the results.
50:36 I'm,
50:36 I'm gonna actually
50:37 just move forward to,
50:38 to see whether you have any specific questions.
50:41 And yeah,
50:41 so,
50:42 thank you so much.
50:44 Thank you very much,
50:45 Lilith.
50:46 Um,
50:46 so I think we have
50:48 time for one round of questions,
50:50 so maybe we'll collect,
50:52 um,
50:52 up to 3 questions,
50:54 um,
50:55 and,
50:56 and then give it back to Lilith.
50:59 So I,
50:59 I'm just seeing,
51:00 I'm gonna call as I see people raise their hands.
51:03 So I see Brianna's raised her hand,
51:05 so go ahead,
51:05 Brianna.
51:08 Hi,
51:08 everyone.
51:09 Thanks so much for this,
51:10 this great presentation.
51:11 I am
51:12 recently departed,
51:13 uh,
51:13 COTAS team leader for Jamaica,
51:16 uh,
51:16 on social protection,
51:17 on our social protection interventions.
51:20 Um,
51:20 and so I'd be curious to know what,
51:22 if any,
51:23 um,
51:23 coordination or collaboration there's been with the PATH,
51:26 the CCT program.
51:28 Um,
51:28 we have been doing,
51:30 um,
51:30 several years of technical assistance on improving their case management,
51:34 uh,
51:34 system.
51:35 And I think there's a lot of complementarities and at least from,
51:39 from my understanding in country,
51:41 and this is,
51:42 you know,
51:42 um.
51:44 Uh,
51:44 also an internal government,
51:45 uh,
51:46 issue,
51:46 but there has not been much coordination with this service,
51:50 uh,
51:50 with the Ministry of,
51:51 um,
51:53 Social Security and Labor.
51:54 So thank you.
52:02 OK,
52:02 so we have uh questions,
52:04 some questions
52:06 in the chat.
52:07 One I just easily answered,
52:08 Lois,
52:09 you presented ITTFX.
52:11 Um,
52:12 there's a question about what sort of take-up rate,
52:15 um,
52:16 you know,
52:16 for your,
52:17 and I guess enrollment.
52:20 I think he's,
52:20 I think he's talking about the eligible,
52:22 um,
52:23 you know,
52:23 represent of the intervention overall.
52:25 Um,
52:26 and then
52:27 did you notice much heterogeneity across impacts
52:30 with respect to pre-existing discipline practices,
52:34 um,
52:35 either in the in-person,
52:36 which I think is Helen's previous research,
52:38 or the digital versions of the interventions.
52:44 And then I think we have,
52:45 we can pick up one more question,
52:47 um.
52:50 Does anyone else have their hand raised?
52:52 I can't,
52:53 I can't see everyone all at once.
53:00 No,
53:00 OK,
53:01 so then I'll add a question from myself,
53:04 so it's what I or just some kind of note something interesting,
53:07 um,
53:08 so,
53:09 um,
53:10 I,
53:10 I get the sense,
53:11 I don't know if anyone has like,
53:12 you know,
53:12 there's a,
53:13 a paper on this,
53:14 but from
53:15 our portfolios experience that
53:17 when
53:18 A digital intervention is paired with some
53:21 kind of human follow up either because facilitators
53:25 of the digital program
53:26 already visit these households or already know these households,
53:30 there's much,
53:31 that's when you get,
53:32 um,
53:32 you know,
53:33 take
53:33 impact
53:34 rather than just,
53:35 you know,
53:36 some online thing dropped out of nowhere on people.
53:39 What's interesting is you,
53:41 you do have
53:42 Kind of,
53:43 uh,
53:43 you know,
53:45 you don't have total take,
53:46 you know,
53:46 complete take up of the in-person thing,
53:48 but you do have some partial takeup of that.
53:51 So just your sense of the literature on that is that your sense as well
53:55 that,
53:55 you know,
53:56 these digital
53:57 interventions are successful when at least there's some light touch
54:01 of a human
54:02 kind of facilitating,
54:03 um,
54:04 kind of use.
54:07 So maybe we'll go with those three questions,
54:09 um,
54:10 and then,
54:10 um,
54:11 then we can close.
54:13 OK,
54:13 sounds good.
54:14 Thank you.
54:15 Uh yeah,
54:16 Brianna,
54:16 very nice to meet you and uh
54:18 yeah,
54:19 so,
54:19 to be honest,
54:20 I'm not sure if there is any coordination between the uh the Ministry of Labor and the
54:26 um and whether it has been embedded into any other like larger project,
54:32 and I don't know if Sean has any insights on that,
54:35 but we'll be very happy to,
54:37 to explore and,
54:38 and see whether
54:39 Um,
54:40 to explore that with the ACC,
54:42 um,
54:43 and see whether there is any,
54:44 any coordination.
54:46 Um,
54:47 my,
54:49 I think Corin was gonna be here,
54:50 but,
54:50 uh,
54:50 Charlene was gonna be here,
54:51 but yeah,
54:51 we can,
54:52 we can,
54:53 from,
54:53 from the ACC but we can explore that,
54:55 uh,
54:56 that later and I'll get back to,
54:58 to you on,
54:58 on that.
55:00 Um.
55:01 Then in terms of the heterogeneity relative to pre-existing practices,
55:05 so there are two things here.
55:06 The first one is that,
55:08 uh,
55:09 yeah,
55:09 so Sean said that we put on the ACD pre but not disintegration.
55:13 So yeah,
55:13 thank you,
55:14 Sharon.
55:15 Uh,
55:15 then,
55:16 uh,
55:16 yeah,
55:16 so heterogeneity,
55:18 we
55:19 do,
55:19 there are two issues here.
55:20 One is that we didn't have enough power to
55:22 explore or the heterogeneity result that we wanted to.
55:26 Uh,
55:26 but then for some of the variables for which we are powered enough to,
55:31 to explore some differences,
55:32 but we
55:33 do observe is that indeed,
55:35 um,
55:36 like parents who had worse practices at baseline are the
55:40 ones who are experiencing the more gains from the,
55:43 from the intervention.
55:45 OK?
55:45 So,
55:46 um,
55:47 so yeah,
55:47 that's,
55:47 uh,
55:48 that's something that we are finding,
55:49 which is,
55:50 again,
55:51 good news uh in terms of,
55:52 uh,
55:53 because that can help us too.
55:55 Um,
55:55 to argue that we are potentially benefiting those who,
55:59 who have the worst practices at baseline.
56:02 And then Ajaca that's actually very good,
56:05 a very good point,
56:06 and you know,
56:07 we,
56:08 this is not causal evidence,
56:09 but
56:10 we conducted this focus group discussions
56:12 and what the parents were reporting is that they
56:15 see that these components in some cases are like um
56:19 like substitutes,
56:20 not necessarily complementary elements from the,
56:23 in terms of the delivery.
56:25 Uh,
56:26 for instance,
56:26 some of them were not attending the sessions because they thought
56:29 that they were receiving enough information through the SMS messages.
56:33 So it's like,
56:33 at least from,
56:34 this is not causal evidence,
56:36 but at least from this,
56:37 this quality work we learned that uh probably people are,
56:40 um,
56:41 like,
56:42 uh,
56:42 are seeing these elements again as substitutes,
56:45 uh,
56:46 but we are still seeing,
56:47 as you said,
56:48 uh,
56:48 from the in other papers that when they're combining elements,
56:51 uh,
56:52 human elements plus virtual.
56:55 Uh,
56:55 they seem to have very important effects,
56:57 and that may be true because in the
56:58 case of El Salvador with this quarantine integration,
57:00 it was fully virtual,
57:02 there was no,
57:03 um,
57:04 face to face intervention,
57:06 and at the end we found negative effects,
57:09 particularly for male caregivers,
57:12 um,
57:12 and they reported that the,
57:15 in terms of stress and also interaction with their kids
57:17 and what they were reporting is that they
57:19 They just,
57:20 it was in the middle of the pandemic,
57:22 they were like managing different responsibilities
57:25 and uh they,
57:26 when they received just information,
57:29 they saw like,
57:30 uh,
57:30 OK,
57:31 I'm not doing anything
57:32 like good with my kids,
57:34 so I'm gonna withdraw and I'm not gonna engage anymore with my kids.
57:39 Um,
57:40 so at the end it had uh,
57:43 my sense,
57:43 but it's not evidence,
57:45 evidence-based,
57:46 is that if we had included in the case of El Salvador,
57:50 a,
57:51 a human interaction,
57:53 probably male caregivers will have the
57:56 opportunity to discuss the information and.
57:59 that could have increased their,
58:01 their self-efficacy and we could have
58:04 prevented the negative effect that we're finding in integration.
58:07 So yeah,
58:07 it's um there is no causal evidence on that and so that's also,
58:11 that,
58:11 that's another very interesting research question that we can,
58:15 we can see whether we can explore in other projects,
58:17 but thank you.
58:19 So,
58:19 yeah,
58:19 that's uh
58:21 that's all from my side,
58:23 and thank you everyone for,
58:24 for the great questions and uh and joining us in this,
58:27 in this seminar.
58:29 Yeah,
58:29 and thank you to Lilith and Sean for joining us and thanks to all of you.
58:33 Let's give them like a
58:34 virtual round of applause,
58:36 um,
58:36 and kind of we look forward to engaging you
58:39 more through this seminar series.
58:42 Thank you.
- add-style
- lp-body-content
Violent Discipline and Parental Behavior: Effects of Digital Parenting Support in Jamaica
Wednesday, October 15, 2025, 10:00 AM Eastern Time
Abstract
Approximately 5% of children aged 2 to 4 worldwide are regularly subjected to violent discipline. We study the impact of a digitally-delivered intervention on positive parenting practices in Jamaica using a randomized control trial. Results show that the intervention improves caregiver knowledge (0.52 SD), and attitudes around violence (0.2 SD). Caregiver violence against children decreases (0.12 SD). As a result of the program, children experience fewer emotional problems (0.17 SD). We also find reductions in caregiver depression (0.12 SD), anxiety (0.16 SD), and parental stress (0.16 SD) nine months later. The digital delivery has important scalable policy implications which could help decrease violence against children worldwide.
Presenters