Economists spend much of their lives talking about and correcting for sample selection. Recent evidence from behavioral economics documents that participants in lab experiments don't account for selection effects when they interpret conditional distributions. This "selection neglect" can distort expectations in settings where individuals learn from comparisons with other people who differ in important ways from themselves. In this paper, we test if selection neglect can be replicated outside the lab. We examine a real-life application where selection effects are particularly strong: selection into part-time jobs. Using new representative survey data from Germany, we elicit employees' expected earnings for part-time and full-time jobs. We then explore if people expect lower wages for part-time work if they observe more selective reference groups. For identification, we exploit that average wage differentials between full-time and part-time workers are large but almost entirely driven by non-random selection into part-time. Specifically, we use heterogeneity in selectivity across occupations. Contrary to experimental studies from the lab, we find no strong evidence for selection neglect in this setting. This discrepancy can be explained by information frictions in real markets.