Information from television game shows has recently been used to measure women's risk aversion. Researchers have abstracted from this evidence to explain the underrepresentation of women at senior levels in politics, business and management. But how reliable is this type of data? Using data for 483 male and female participants in a simulation of the TV game show 'Deal or no Deal', we find that women on average exit 0.45 rounds earlier than men, confirming the higher risk aversion for women. We also find that if we were to select women with comparable earnings and education to men, being female is less of an obstacle towards risk-taking behaviour than in the absence of these controls. Specifically, women would now be seen to exit 0.12 rounds earlier, rather than 0.45 rounds earlier. Experiments need to be mindful of controlling for these background factors when assessing the nexus between risk-taking and gender.