Substitution Activities to Reduce Adolescent Drinking in Lviv, Ukraine

How to cite

Kettle, Stewart. 2014. "Substitution Activities to Reduce Adolescent Drinking in Lviv, Ukraine." AEA RCT Registry. AEARCTR-0000578. https://doi.org/10.1257/rct.578-3.0

Download: BibTeX · RIS — for reference managers without a browser translator. Zotero and most others pick up this page's citation metadata directly.

Stewart Kettle*

* Primary investigator

Registry ID
AEARCTR-0000578
DOI
https://doi.org/10.1257/rct.578-3.0
First registered
Last updated
Status
In development
Trial dates
2014-12-14 – 2015-04-30
Countries
Private
Secondary IDs
BIT Trial Number: 2014052
Registry record
http://www.socialscienceregistry.org/trials/578

Abstract

This trial aims to reduce the alcohol consumption of school students in Lviv, Ukraine. The trial design is by BIT and the project is in collaboration with the UNDP, and Samopomich. The trial will test an intervention focused around increasing the number of alternative extra-curricular activities for children as well as increasing information on them.The intervention, aimed at reducing the alcohol consumption of school students, has been designed with two specific behavioural insights in mind. The first is that the challenge of creating a successful intervention is made greater by the ‘Romeo and Juliet’ effect – the tendency, especially strong in young people, to actively react against being told what to do. The second is that talking about how bad the problem of teenage drinking may inadvertently increase the salience of drinking and lead to students thinking more of their peers drink than actually do. The intervention will therefore have two key components: 1) New activities - increasing the number of desirable alternative after-school activities for children, 2) Information on activities - increasing information on all after-school activities. The design is a cluster randomised trial with 10 schools out of 20 randomly assigned to receive the 'substitution activities' intervention.

Intervention

The intervention will have two key components:
1) New activities - increasing the number of desirable alternative after-school activities for children
2) Information on activities - increasing information on all after-school activities

Experimental design

Cluster randomised design. 20 schools with 10 randomly allocated to receive the intervention. Outcome variables recorded at the student level (self reported). We estimate the impact of our treatment on five outcomes variables: First, whether a student has drank over the past 30 days. Second, whether a student has drank over the past 7 days. Third, whether a student has drank over the past 3 months. Fourth, perceived level of drinking of peers. Fifth, consumption at hazardous levels.
We will estimate a model of the form
:
D_i=z+ β_1 T_ji + γφ_i+ U_i

Where D_i represents each of our outcome variables. z is a binary constant. This contains omitted categories, and so in this specification can be interpreted as the average propensity to drink of students in the control group. T is our treatment variable and set to 1 for all students in schools that received the substitution activities intervention. φ is a vector of control variables comprising student characteristics. Ui is an i.i.d. error term. Standard errors will be clustered at the school level.

Design details

Randomization method
20 schools chosen by UNDP form our sample. Schools were stratified based on two criteria. First schools were ranked in order of baseline levels of previous drinking. These schools were then grouped into 5 strata of 4 schools based on these rankings (i.e. taking the top four schools, then the next four, and so on). Within these strata the schools were then ranked in order of school size and then split into pairs in a similar manner. This process served to create pairs of similar schools based on these criteria, one of which would be assigned to receive the intervention and one to be a part of the control group. One school from each pair was then randomly allocated to receive treatment (based on random number generation in excel)
Randomization unit
School
Clusters
20 Schools
Observations
1700
Arms
10 schools treatment, 10 schools control
Primary outcomes
The primary outcome measure for the trial will be a binary outcome of whether a student has drank any alcohol over the previous month (self-reported). We will also use a number of secondary outcome measures including: • Any alcohol consumption during past 7 days (self-reported) • Any alcohol consumption during past 3 months (self-reported) • Perceived level of drinking of peers (self-reported) • Consumption at hazardous levels (more than 5 drinks on one occasion in the last 30 days) All outcome measures are self-reported due to two main reasons; 1) the numbers involved in alcohol related crime and alcohol related hospital admissions are too low, in 2013 there were only 126 incidents of alcohol related crime and 68 hospital admissions among young people in Lviv; 2) we were informed that testing for cerebellar functions or other measures of drinking would not be possible. Also of note is that the original design was to use the WHO AUDIT survey for detection of hazardous drinking levels (used in baseline survey – see Annex B and Annex C). However baseline results showed low levels of drinking, only 39% of students had ever drank alcohol before (see Annex D for further baseline results). We therefore chose to use any drinking of any alcohol over the past 30 days as our primary measure as this was more suitable to pick up changes in lower levels of drinking.

Cited by

No citation data available for this registration yet.

Keywords