The Effect of In-School Sexual Education: an Evaluation using Survey and Administrative Data

We study the effect of a school-based sexual education program on adolescents' sexual behavior and fertility outcomes. The intervention consists of 90-minute interactive group sessions delivered in school by trained educators, to 9th and 10th grade students. The curriculum covers biological knowledge, contraception, self-efficacy, communication skills, and gender norms. The sessions are conducted in mixed-gender groups. We plan to complete the interventions in the 2025/2026 and 2026/2027 academic years.

2026-04-01

2027-07-01

Measured by administrative records, and in the follow-up survey:
-Number of pregnancies
-Number of births
-Number of abortions
-Number of sexually transmitted infections

Measured in the follow-up survey:
Self-reported sexual behavior and contraceptive use

In the administrative records, we plan to measure the number of pregnancies, births, and abortions from health records using ICD codes. We can link administrative records by school and cohort identifiers. The admin data consists of a 50 percent sample of the Hungarian population, thus the sample size is 50% in the admin. (The number of schools (clusters) is the same) In the admin, we potentially include students who were absent on the day of the sessions. As a consequence, for these outcomes, we can measure intent-to-treat effects. We measure these outcomes 1 year, 3 years and 5 years after the intervention for girls. For both boys and girls, we measure the number of sexually transmitted diseases in the administrative records (ICD codes A50-A64).

We measure the survey-based outcomes one year after the intervention. Number of pregnancies, births, abortions for girls, and number of pregnancies, births, and abortions of sexual partners of boys are self-reported in a self-administered online anonymous questionnaire.
Sexual behavior is measured through the question, "Have you ever had sexual intercourse (Have you slept with someone)?", contraceptive use is measured by the multiple choice question "What method(s) of protection did you use the last time you had sex?" with the following choices: We did not use any protection; Condom; Birth control pill; Other method, namely: ............

Short-term outcomes: outcomes related to sexual knowledge and attitudes (capability-, motivation-, and opportunity-related outcomes)
Long-term outcomes: outcomes related to the education level (obtaining secondary education) and labor market participation

Furthermore we test effect heterogeneity across:
- sexes (male vs. female),
- educational performance and aspirations,
- and socio-economic background

Short-term outcomes:
The secondary outcomes related to sexual knowledge and attitudes are all measured in the follow-up survey, with specific questions.

Questions about sexual knowledge test knowledge about contraception methods, pregnancy risk, and STDs:
“Mark ALL answers you think are correct for the questions below:
The advantages of a condom include that it: helps prevent sexually transmitted diseases / can be purchased at a pharmacy / has no dangerous side effects / does not require a prescription / none of the above is true / I don't know
Which method provides protection against sexually transmitted diseases? Condom / contraceptive pill / IUD (intrauterine device) / oral sex / none of the above / I don't know
A woman can get pregnant during sex: during withdrawal/coitus interruptus / even if there is no penetration / if she does not take the contraceptive pill regularly / during menstruation / none of the above is true / I don't know
The 72-hour pill (emergency contraception): is available without a prescription / has no harmful consequences if taken frequently / effectively prevents pregnancy if taken within 72 hours after intercourse / protects against sexually transmitted diseases / none of the above is true / I don't know
How likely is it that a girl will become pregnant within a year if she has sex regularly without contraception? she will definitely not get pregnant (0% chance) / she is very unlikely to get pregnant (10% chance) / she will get pregnant in about half of cases (50% chance) / she will almost certainly get pregnant (90% chance) / none of the above is true / I don't know”


Questions related to attitudes include:
- Measures of boundary-setting capabilities:
"How much do you agree with the following statements? Mark one answer from completely agree / rather agree / rather disagree / completely disagree for each question:
I don't like to say no, because it can lead to conflict.
I am able to say no without giving an explanation.
When necessary, I put my own needs first."

- Measures of communication skills:
"Which of the following do you think are true about condom use? (Mark one answer from completely agree / rather agree /neither agree nor disagree / rather disagree / completely disagree in each row)
It is very difficult to bring up the topic of condom use during sex.
It is very embarrassing to buy a condom."
“How confident are you in the following situations? (Mark one answer from “Not confident at all / Rather not confident / Moderately confident / Rather confident / Completely confident”):
Talking with your sexual partner about any topic related to sex.”


- Measures of the presence of a trusted person (with whom the respondent can discuss sexuality-related questions):
"There is someone in my life with whom I can discuss my questions and problems related to sex. Yes / No"
"Have you ever talked about sex with someone you trust? (You may mark more than one answer)
There is no such person in my life / With my father or mother / With another adult / With a friend / With my partner / With a classmate or schoolmate / With someone else, namely:….."

- Measures of self-efficacy in contraception:
“How confident are you in the following situations? (Mark one answer from “Not confident at all / Rather not confident / Moderately confident / Rather confident / Completely confident” in each row):
Talking with your sexual partner about preventing unwanted pregnancy.
Saying no to sex when you are not in the mood.
Saying no to sex if your partner refuses to use a condom or other contraception.”


Long-term outcomes: obtaining secondary education and labor market status are measured from administrative records 5 years after the trial.

We will conduct a randomized controlled trial among Hungarian secondary schools with a history of high teenage pregnancy rates. We selected the eligible schools based on past administrative records. Eligible schools are those where the number of teenage pregnancies per girl was above 0.025, in the cohorts who were in 10th grade in the years 2015-2017. In addition, the schools have to be large enough: the average number of girls in the administrative records per cohort should exceed 15 (which means that in the population their number exceeds about 30 per cohort). This way, we ended up with 259 schools in our sampling frame. We contact schools by e-mail and phone and assign them randomly to treatment and control group upon entry into the study, based on a 'rolling randomization" process, where the probability of assignment to the treatment group is 67%. Randomization is stratified by regions and previous average math competence score (low vs. high) as they are associated with teenage pregnancy rates.
A baseline and a follow-up survey, 1 year after the baseline, are conducted in both treated and control schools. The baseline and follow-up survey can be linked individually through anonymous identifiers. The group sessions are conducted in treatment schools right after the students have answered the baseline survey.

Not available

Randomization is done in an office by a computer, upon the entry of a school into the study, following a rolling randomization process. The probability of assignment to treatment group is 67%. Randomization is stratified by the region and the previous average math competence score (low vs. high).

School

Yes

90 schools. This number is the best guess as there is much uncertainty about the willingness of schools to participate because of the abovementioned institutional surroundings.

The average number of students in two cohorts in our sample is 208. Around 90*208=18720 students (if both girls and boys participate in all schools) fill in the baseline survey. Among them around 11440 are from treated and 7280 are from control schools. Calculating with a 50% retention rate, around 9360 students fill in the follow-up survey (5720 from treated, 3640 form control schools). Number of observations coming from the administrative data: Around 5720 students from treated schools in the administrative data (primary outcomes for both sexes: STDs; for girls: admin births, abortions, pregnancies ). Around 20800 students from control schools coming from the administrative data.

In the case of outcomes coming from the administrative data: 200 schools control, 55 schools treated
In the case of outcomes coming from the surveys: 35 schools control, 55 schools treated

For power calculations, we use data based on 2015 admin records. We chose schools where pregnancy rates were above 0.025/girl, and where the number of girls per cohort was at least 15. We end up with a sample where we have 55 girls per school in the admin data, the mean number of pregnancies is 0.144, and rho=0.087. We use alfa=0.05 and power of 0.8. The minimum detectable effect size with these parameters in the case of 55 treated schools and 200 control schools is a 0.46 decrease in the number of pregnancies, which corresponds to a 32% decrease in the pregnancy rate. While this effect size is large, the high teenage pregnancy rates and the lack of sexual education point to a substantial expected effect of the program.