The Challenge

As technology reshapes economies, the information technology (IT) sector offers strong career prospects for young people. Yet access to these opportunities is uneven: women represent only about 35 percent of the global tech workforce. Entry into the sector often depends on professional networks that circulate information about openings and influence beliefs about who belongs in IT. In male-dominated industries, these dynamics risk reinforcing gender stereotypes and limiting women’s economic opportunities.

These dynamics may play out differently among young people without established professional networks. In Panama, around 25 percent of youth are not in education, employment, or training (NEET), leaving them disconnected from pathways into growing sectors like IT. Addressing this gap is both an economic and social priority: without action, promising industries may struggle to find talent, while a generation of young people remains excluded. To address this, the youth development organization Glasswing International runs a six-month IT career training program for young secondary school graduates who are currently NEET. Due to their status, this group does not have experience or professional networks compared to career employees. But even then, questions remain: do gendered perceptions of ability and preferences continue to shape choices and limit women’s participation in IT?

The Evaluation

In collaboration with IPA, researchers examined gender beliefs among applicants to Glasswing International’s IT career training program. To do this, they collected data via mobile phones and used monetary incentives to elicit beliefs about men’s and women’s general cognitive abilities, IT-specific abilities, and preferences for IT-related tasks. A total of 244 applicants participated in the study.

First, researchers examined first-order beliefs–participants’ own beliefs about men’s and women’s abilities and preferences. To generate benchmarks, a small group of eligible applicants had previously completed tests of general cognitive ability, IT-specific ability, and a ranking task. Participants in the study were shown descriptions of these tasks and asked to state who they believed performed better—a randomly selected man or woman from the benchmark group. They reported three beliefs (cognitive ability, IT ability, and preferences), and one was randomly selected for payment under a scoring rule, allowing them to earn up to USD 5.

Researchers then examined second-order beliefs—what participants believed others thought about gender differences in general cognitive ability. Participants were informed that people earlier answered the same questions about first-order beliefs and then asked to guess what a random man and woman chose as their answers. One of the first-order beliefs was randomly selected for payment, and participants could earn USD 5 for guessing the others’ answers correctly.

Finally, to confirm that beliefs about ability and interest predicted beliefs about actual application to the program, researchers asked participants to bet on the application decisions of 20 randomly selected name pairs (10 mixed-gender, 5 woman-woman, 5 man-man) by identifying one person from each pair. If the chosen individual later applied to the program, participants won USD 5.

Results

The findings suggest that women’s underrepresentation in IT may stem less from doubts about their abilities and more from perceptions of their career preferences. On average, participants believed that women had stronger general cognitive and IT-specific abilities than men. Yet they also believed that men were more interested in IT tasks and that men were more likely to apply for IT training. This perception—that women are less inclined toward IT work—could play a stronger role in keeping women out of the sector than assumptions about skill.

The study also revealed important dynamics in how people think others view gender. Participants correctly anticipated that women would see women as having higher general cognitive ability. However, men significantly misjudged the views of their peers: they assumed other men favored men much more than they actually did. This “false consensus” suggests that men may be overestimating gender bias within their networks, which can influence their own attitudes and behavior.

Finally, the evaluation showed that collecting data through mobile-phone surveys with incentives can be both cost-effective and reliable. Participants understood instructions, faced minimal technical issues, and gave consistent answers. This method could therefore be applied in other interventions or contexts where low-cost, scalable data collection is needed.

Implementing Partner