Improving measurement of harms from others' drinking: Using item-response theory to scale harms from others' heavy...

The heavy drinking of others may negatively affect an individual on several dimensions of life. Until now, there is scarce research about how to judge the severity of various experiences of such harms. This study aims to empirically scale the severity of such harm items and to determine who is at most risk of these harms.

We used population-based survey data from 10 countries of the GENAHTO project (Gender and Alcohol's Harms to Others, data collection: 2011-2016). Questions about harms from others' drinking asked about verbal and physical harm, damage of belongings, traffic accidents, harassment, threatening behaviour, family and financial problems. We used item response theory methods (IRT) to scale severity of the aforementioned items. To acknowledge culturally based variations in different countries, we assessed 'differential item functioning'.

The items 'family problems', 'financial problems' and 'clothes and property damage' as well as 'physical harm' were scaled as more severe in most countries compared to other items. Substantial differential item functioning was present in more than half of the country pairings. The item 'financial problems' was most often differentially scaled. Younger people who drank more, as well as women (compared to men), reported more harm.

Using IRT, we were able to evaluate grades of severity in harms from others' drinking. IRT scaling yielded in similar rankings of items as reported from other studies. However, empirical scaling allows for more differentiated severity scaling than simple summary scores and is more sensitive to cultural differences.