Flow is a concept with a long pedigree in psychology, first described by Mihaly Csikszentmihalyi as the state of deep absorption and intrinsic enjoyment people experience when fully engaged in an activity. Short video flow adapts the idea to the swipeable feed: the sensation of losing track of time while algorithmically curated clips unfold one after another, each personalized to the viewer’s tastes. Previous research has linked this state to adverse mental health outcomes, but most studies were cross-sectional snapshots or conducted in Asian cultural contexts where platforms like Douyin dominate. Whether flow actually precedes and worsens distress — rather than simply co-occurring with it — remained an open question, particularly outside Asia.

To address that gap, Akbulut employed a two-wave longitudinal design with a five-month interval, tracking 495 young Turkish adults aged between 18 and 40 years. The sample was predominantly female, at 84.2 percent, with a mean age of 21.23 years and a standard deviation of 2.18, meaning most participants were clustered between 19 and 23 years old. At both time points, participants completed self-report measures assessing short video flow, loneliness, and psychological distress. The five-month gap matters: it is long enough to separate genuine developmental change from momentary mood fluctuations, yet short enough to minimize attrition and keep the same life context roughly intact.

The statistical engine of the study was cross-lagged panel analysis, a technique that allows researchers to test directional relationships over time while controlling for what each variable predicts about itself. Every measure at Time 1 naturally correlates with the same measure at Time 2 — a phenomenon known as an autoregressive effect. By statistically holding these stability paths constant, the model isolates the cross-lagged paths: does short video flow at Time 1 predict loneliness at Time 2 above and beyond earlier loneliness levels? Does loneliness at Time 1 predict distress at Time 2 beyond prior distress? Only if these directional paths emerge can researchers argue that flow leads to loneliness, rather than lonely people simply flowing more.

The results supported exactly that directional chain. Short video flow at Time 1 significantly predicted higher levels of loneliness at Time 2, with a standardized coefficient of β = 0.14, p < .001. In turn, loneliness at Time 1 significantly predicted increased psychological distress at Time 2, with β = 0.21, p < .001 — the strongest path in the model. Short video flow at Time 1 also directly predicted psychological distress at Time 2, with β = 0.10, p < .05, indicating that flow harms mental wellbeing through channels beyond loneliness as well. Crucially, the indirect effect flowing through loneliness was statistically significant, with an unstandardized estimate of B = 0.017, a 95 percent confidence interval of [0.005, 0.030], and a standardized β = 0.030. Because the direct path remained significant alongside the indirect one, the study found partial rather than full longitudinal mediation.

In plain terms, the model suggests that greater experiences of short video flow may indirectly exacerbate psychological distress by increasing feelings of loneliness over time. The mechanism is plausible on theoretical grounds: flow states are inherently solitary, demanding deep absorption that crowds out social interaction, and the algorithmic feed replaces genuine companionship with parasocial engagement. Over months, hours once spent with friends may migrate toward the screen, and the personalized feed — however entertaining — cannot reciprocate emotional support. Loneliness, in turn, is one of the most robust predictors of depression and anxiety symptoms in young adults, so any experience that deepens it is likely to deepen distress as well.

The model’s fit statistics tell a more nuanced story. Incremental fit indices met conventional thresholds, with CFI = 0.94, IFI = 0.94, NFI = 0.93, TLI = 0.92, and GFI = 0.91, all at or above the 0.90 cutoff commonly used to judge adequate fit. However, RMSEA = 0.094 with a 90 percent confidence interval of [0.085, 0.104] exceeded the recommended 0.08 threshold, and χ²/df = 5.38 surpassed the recommended cutoff of 3, indicating some degree of model misspecification. Akbulut reports these limitations transparently, noting that the model captures the broad pattern of longitudinal relationships but may omit additional paths or variables that would improve fit. Readers should therefore interpret the effect sizes as evidence of real but modest longitudinal associations rather than precise causal quantities.

Several further caveats temper the conclusions. The sample was heavily female and drawn from a single national context, so generalization to other populations requires caution. Self-report measures can be inflated by shared method variance, and five months is only one slice of the developmental timeline — flow and loneliness may also reinforce each other in shorter feedback loops the design cannot detect. The study was approved by the Afyonkarahisar Health Sciences University Non-Intervention Scientific Research Ethics Committee and conducted in accordance with the Declaration of Helsinki, and the author received no external funding for the research. The author also notes that artificial intelligence tools were used solely for language editing and proofreading, with full responsibility for the content remaining with the author.

Even with these limitations, the study is notable for extending flow research into a non-Asian cultural context and for using a design that can speak to directionality — a rarity in a literature dominated by cross-sectional surveys. The findings align with a growing body of work suggesting that highly immersive, algorithmically personalized media may carry psychological costs that accumulate gradually rather than appearing immediately. They also echo classic media psychology concerns about displacement: time absorbed by flow is time not spent on the social interactions that protect mental health, and the displacement effect may be strongest precisely when the media experience feels most rewarding.

For the millions of young adults who lose track of time in the feed each night, the practical message is not necessarily to abandon short video platforms but to notice when absorption starts replacing connection. Akbulut suggests that interventions aimed at reducing loneliness and promoting healthier patterns of digital media engagement may be particularly beneficial in mitigating the negative psychological effects of short video platforms. That could mean designing apps that encourage social co-viewing, building digital literacy programs that teach users to recognize flow states, or simply encouraging young people to balance screen absorption with in-person interaction. As short video consumption continues to expand globally, understanding how its most pleasurable moments may quietly deepen isolation could prove essential for protecting the mental health of a generation raised on the swipe.

Subject of Research: Short video flow, loneliness, and psychological distress among young Turkish adults: a cross-lagged panel study

Article Title: Short video flow, loneliness, and psychological distress among young Turkish adults: a cross-lagged panel study

Article References: Akbulut, Ö. F. (2026). Short video flow, loneliness, and psychological distress among young Turkish adults: a cross-lagged panel study. BMC Psychology. https://doi.org/10.1186/s40359-026-05761-2

Image Credits: AI Generated

DOI: 10.1186/s40359-026-05761-2

Keywords: short video flow, loneliness, psychological distress, young adults, cross-lagged panel analysis, BMC Psychology, digital media, mental health, Short, video, flow, psychological