One Lone Item: Assessing loneliness with a single-item direct measure

Across the most widely-used loneliness scales, the words ‘loneliness’ and ‘lonely’ are commonly avoided due to concerns about bias. These efforts, although motivated by caution, may have undermined scales’ abilities to assess loneliness accurately. In this article, we validate the One-item Loneliness scale (1iL; “I often feel lonely”), and show using multi-rater data (Ntotal = 1,262 people, recruited from Prolific) that it induces no more bias than popular indirect scales. The 1iL had retest reliability superior to some short multi-item scales (rtt = .74) and comparable cross-rater agreement (rCRA = .52). Using multi-rater adjustment, we find near-perfect convergence with the Three-Item Loneliness Scale (rtrue = .95), the most popular short measure of loneliness, and the Emotional subscale of the De Jong Gierveld (rtrue = .97). We confirmed these findings in a highly-powered preregistered replication. Crucially, the 1iL demonstrated clear discriminant validity from the De Jong Gierveld's Social subscale; contrary to popular belief, loneliness is unidimensional, can be assessed directly, and is psychologically distinct from judgements about one’s social circle.