People trust thermometers because generations of researchers compared instruments and gradually established shared standards. That process took more than a century. Galileo’s early device, built around 1600, had no scale, while later thermometers used different materials and reference points, allowing devices in the same room to produce different readings.
Some errors persisted even after measurement became more standardized. In the 1860s, German physician Carl Wunderlich used more than a million armpit readings from around 25,000 patients to establish 37°C as a normal body temperature. Physician Philip Mackowiak later found that one of Wunderlich’s thermometers read almost two degrees higher than modern instruments, yet the figure remained widely used for more than a century.
Measurement problems can have unequal consequences. Pulse oximeters were calibrated largely using people with lighter skin. Research published in 2020 found that the devices failed to detect low oxygen about three times as often in Black patients as in White patients. During the COVID-19 pandemic, oxygen readings influenced access to some treatments, and later research found that the bias delayed recognition of low oxygen in Black and Hispanic patients.
Loneliness measurement is less developed. Although loneliness has become a public health priority and governments increasingly include it in national surveys, many rely on one question, such as how often a person feels lonely. This approach is inexpensive and has helped produce population data, but its reliability varies.
A study involving more than 25,000 respondents across all 27 European Union member states compared three loneliness measures. A three-item scale produced results that could be compared across every country. The single-question measure met the researchers’ minimum validity standard in only 19 countries. The study also noted that single- and multi-item measures can correlate weakly, with correlations as low as .27, while two surveys conducted in the same year produced country estimates that differed by about four percentage points on average.
These discrepancies can affect policy and funding. A country may appear more or less lonely than it actually is, while a program may seem ineffective because its measurement tool failed to capture the change it produced. In the EU data, the single question was more closely associated with emotional distress than with people’s social lives, meaning that a program designed to rebuild social networks might have little effect on the score.
The article also describes a 2021 study of more than one million adolescents in 37 countries that reported school loneliness had almost doubled between 2012 and 2018 and linked the trend to smartphones. When the authors later examined the study’s measurement, they found that no country met the standard for any level of measurement invariance. As a result, the study could not establish that its loneliness score had the same meaning across countries or across the two years.
Unlike temperature, loneliness is not a single, uniform quantity. Interviews with 354 people in eight countries found that people assess relationships through factors including trust, reciprocity and the ability to feel safe being themselves, but they give those factors different importance. One participant in rural Zimbabwe described severe loneliness despite being surrounded by family and neighbors, while a university student in the United States associated loneliness with moments when nobody answered the phone.
A single question requires people to compress these different experiences into one answer. Some people describe loneliness through anger or self-doubt, while others see time alone positively. Shame can also affect responses in places where being socially connected is treated as a moral obligation. Consequently, people with the same score may need very different kinds of support, while someone lacking important relationships may report a low level of loneliness.
The researchers argue that surveys should therefore examine which aspects of a person’s relationships are falling short, rather than measuring only how lonely someone feels. They recommend using multi-item scales tested for comparability whenever possible and treating country comparisons based on a single question cautiously. For individual support, they suggest asking what is specifically missing before deciding what help to provide.
Sources: Psychology Today.