The effects of parental involvement in financial literacy education. Evidence from a randomized experiment.

Schools are assigned to the following four experimental conditions:
1) Control group: Students do not follow the course on financial education.
2) Treatment group 1: Students follow the course on financial education.
3) Treatment group 2: Students follow the course on financial education and complete a homework assignment without their parents.
4) Treatment group 3: Students follow the course on financial education and complete a homework assignment together with one of their parents.

2018-01-12

2018-06-30

We measure financial literacy by a test that contains 13 questions. This test consists of questions that measure financial knowledge as well as financial behavior.

Schools that registered for participation were randomized to the aforementioned four experimental conditions. We assessed the level of financial literacy of all students before as well as after followed the course. Students assigned to the control group completed the same tests at the same time as students in the treatment groups, without following the course between the tests. The experiment was implemented in two waves: In the first wave, students in grade 8 (the second year of secondary education in Flanders, age 13-14) participated in the experiment. In the second wave, students in grade 9 (the third year of secondary education in Flanders age 14-15) participated, completing the same course and questionnaires.

Before the intervention, all students (control group + treatment groups) filled in an online questionnaire to test their financial literacy. Students also reported several personal characteristics like gender, school grades, and socio-economic background. Teachers in the treatment groups received the course material, including instructions for implementation, after all students filled in the questionnaire.
Approximately one month after filling in the first test, students in the treatment groups followed a 4-hour financial education course on means of payment. The course material was designed as a serious game and students had to solve questions on means of payment in groups of two students. Students could find the necessary information to solve the questions in a paper booklet that complemented the digital game. As incentive for good performance during the course, the winning team received a small present. All course material was developed by high school teachers with consideration for the age and the ability of the students.
In treatment groups 2 and 3 (see above), students had to complete a homework assignment before the start of the course. In treatment group 2, students did the homework without their parents. In treatment group 3, students did a variation of the homework assignment together with their parents.
At the end of the course, students filled in a test to measure financial knowledge and financial behavior. At the same time, students in the control group also filled in this test without having followed the course. This test was designed as variation of the first test, i.e. with questions analogous to the test students completed before the intervention.

Schools were randomly assigned to the different experimental conditions by a random number generator in Stata.

In both waves, we randomized the treatment at school level. All students and teachers in the same school were assigned to the same condition. In this way, all teachers in the same school received the same teaching material and instructions in order to minimize the possibility of spill-over effects and contamination of the different treatment conditions.

Yes

Given that the intervention has already been executed and data have been collected for both the pretest and the posttest, we report the actual number of students and schools for which we have observations for both the pretest and the posttest, i.e. the final sample for our analysis. The calculation of the minimal detectable effect size is also based on this final sample.
Wave 1: 41
Wave 2: 25

Wave 1: 2,228 Wave 2: 551

Wave 1: grade 8
Control group = 739 pupils, 13 schools
Treatment group 1 = 400 pupils, 9 schools
Treatment group 2 = 598 pupils, 12 schools
Treatment group 3 = 491 pupils, 7 schools

Wave 2: grade 9
Control group = 111 pupils, 5 schools
Treatment group 1 = 161 pupils, 7 schools
Treatment group 2 = 124 pupils, 6 schools
Treatment group 3 = 155 pupils, 7 schools

Total sample
Control group = 850 pupils, 18 schools
Treatment group 1 = 561 pupils, 16 schools
Treatment group 2 = 772 pupils, 18 schools
Treatment group 3 = 646 pupils, 14 schools

Average number of schools per condition = 16.5
Average number of pupils per condition = 707.25
Average number of pupils per school = 42.9

The computation is based on List et al. (2011) and accounts for intracluster correlation in the calculation of the minimal detectable effect size. In our experimental setting, there are on average 16.5 schools in each experimental condition. Each school on average contains 42.9 students. Computation in Stata based on the posttest shows that the intracluster correlation in the final sample equals 0.1,. In the analysis, this intracluster correlation can be reduced by controlling for characteristics of schools and students. Using the conventional power of 0.8 and a significance level of 0.05, the calculation results in a minimal detectable effect size of 0.34 standard deviations. Details of the calculation: According to List et al. (2011), in a clustered design, the minimum number of observations in each experimental group can be computed as follows: n=2(t_(α/2)+t_β)²(σ/δ)²(1+(m-1)ρ) This implies that the minimum detectable effect size is equal to: δ=σ/√(n/(2(t_(α/2)+t_β)²(1+(m-1)ρ))). Or the minimum detectable effect size expressed as a fraction of a standard deviation is equal to: δ/σ=1/√(n/(2(t_(α/2)+t_β)²(1+(m-1)ρ))), i.e. δ/σ=1/√((707.25 )/(2(1.96+0.84)²(1+(42.9-1)0.1)))=0.34. Reference List, J., Sadoff, S. and Wagner, M. (2011), So you want to run an experiment, now what? Some simple rules of thumb for optimal experimental design, Experimental Economics 14, 439-457.

This paper provides causal evidence on the effects of parental involvement on stu- dent outcomes in a financial education course based on two randomised controlled trials with a total of 2779 students from grade 8 and 9 in Flanders. Using an exper- imental design with three treatment groups, the impact of parental involvement in homework is distinguished from the standalone impact of the classroom intervention and homework itself. Intention-to-treat analysis reveals that access to the intervention effectively improves students’ financial literacy in the two dimensions of knowledge and behaviour. The classroom intervention combined with a homework assigned to be completed with the parents increases financial literacy by 0.38 standard deviations. On average, the added value of prompting parental involvement in homework is not statistically significant. Yet, stimulating parental involvement has significant positive effects on behaviour for disadvantaged students.

Maldonado, J.E., De Witte, K., and Declercq, K. (2022). The Effects of Parental Involvement in Homework - Two Randomised Controlled Trials in Financial Education. Empirical Economics 62(3), 1439-1464.

https://doi.org/10.1007/s00181-021-02058-8