Digital Hoarding: What Has Actually Been Measured

Forty thousand unread emails. Nine years of screenshots. Four backups of the same folder, none of which you trust enough to delete the others.

There is research on this, and more of it than there was five years ago. There is a validated questionnaire, a scoping review, a workplace literature, and a study of 801 people linking it to everyday cognitive slips.

There is also one experiment, no longitudinal study, no intervention trial of any kind, and an unresolved argument about whether digital hoarding is harmful, harmless, or in some settings mildly useful.

This article is about what has been measured and what has not, because the second list is longer than anyone writing about this subject admits.

What has been measured

The foundational work is qualitative. Sweeten, Silence and Neave (2018) interviewed 45 people aged 20 to 52. Three themes emerged: over-accumulation, difficulty deleting, and anxiety attached to both, differing between personal and workplace contexts.

Vitale, Janzen and McGrenere (CHI 2018) interviewed 23 people in their own homes, watching them navigate their own devices, and found something more useful than a category: a spectrum, with hoarding at one end and deliberate minimalism at the other. People sit at different points on it for different kinds of data, ruthless with email, sentimental about photographs, and both poles turned out to be doing identity work rather than file management.

Then the instruments arrived. Neave, Briggs, McKellar and Silence (2019) built the Digital Hoarding Questionnaire across samples of 424 and 203 adults. Ten items, two subscales, and good reliability: α = .905 and .824 in the first study, .945 and .873 in the second.

A separate group has since produced a larger instrument. Sedera, Lokuge and Grover (2022), in Information & Management, surveyed 846 people and modelled digital hoarding across three dimensions, difficulty discarding, digital clutter, and excessive acquisition, linking it to anxiety through attachment theory. We could not retrieve the full statistics and are not printing them.

The two things the questionnaire separates

The DHQ's structure makes a distinction most writing on this subject collapses.

Accumulating, four items, is how much you take in and keep. Difficulty deleting, six items, is what happens when you try to get rid of something.

Someone who downloads everything and never revisits it is doing something different from someone who opens the folder monthly, looks at a file they have not needed in four years, and closes the window again. The first is a storage habit. The second is a decision that keeps failing to get made.

The workplace is where the evidence is strongest

This is the part of the literature with real numbers in it, and it is the part nobody quotes.

In the second Neave sample, employees who had formal data-protection responsibilities held significantly more material than those who did not:

Item	Data-protection duty	No such duty	Statistic
Unread emails	128.5 (SD 199.0)	29.8 (SD 33.9)	F(1,195) = 13.095, p < .001
Read emails	359.3	254.0	F(1,195) = 4.030, p = .046
Presentation files	163.8	74.5	F(1,195) = 6.190, p = .014
Photographs	215.7	116.7	F(1,195) = 5.211, p = .024

Roughly four times as many unread emails among the people whose job is to be careful with records. And a null that matters just as much: there was no significant difference in deletion behaviour between the two groups. They keep more. They do not delete differently.

They also feel more exposed by it. Asked how personally consequential it would be if their files were released, the data-protection group rated every category at around 3.5 to 3.9 on a seven-point scale against 2.2 to 2.6 for everyone else.

The qualitative work explains the mechanism. McKellar and colleagues (2020) interviewed 20 high-scoring employees across an academic and a commercial organisation and identified four underlying dimensions: anxiety, disengagement, compliance and collection. A follow-up study of 11 employees in a large commercial organisation found the strongest driver was anxiety arising from a blame culture, the need to be able to account for a decision later. Those participants understood their retention policies, GDPR included, and used workarounds to keep data longer than the policy allowed.

That is the most concrete finding in the field. It is also, note, eleven people.

Does it actually do any harm?

Here the literature stops agreeing with itself.

The strongest harm finding. Liu and Liu (2025) surveyed 801 university students and found digital hoarding correlated with everyday cognitive failures at r = 0.36 and with fatigue at r = 0.37. Fatigue mediated: the indirect effect was 0.15, 95% CI [0.11, 0.19], and the direct path dropped from β = 0.36 to β = 0.21 once fatigue was in the model. Mindfulness moderated it, β = 0.37 among low scorers against β = 0.12 among high scorers. The authors state plainly that causal inference is not available from their design.

The finding that runs the other way. Gao and Yu (2025) studied Chinese employees across three samples and found digital hoarding positively associated with self-rated work performance, β = 0.397, p < .01, replicating at β = 0.461. Two mediators pulled in opposite directions: job burnout carried a negative indirect effect (-0.054) and thriving at work a positive one (+0.081), with the positive dominating. The authors concede they cannot establish temporal precedence and flag reverse causality.

The null. Bravo-Adasme and colleagues (2025) tested 327 students and found digital hoarding had no significant direct effect on academic performance and no relationship with academic engagement. Its only significant path was to academic burnout at β = -0.192, negative, meaning more hoarding went with less burnout. The model explained 14.2% and 4.6% of variance respectively.

Three studies, three directions. None of them can establish causation, and all three are cross-sectional.

What the workplace evidence does support is narrower and more interesting than "clutter is bad." A 2025 study of 344 employees found the dominant path ran through felt loss of control (β = 0.627) rather than through volume itself, which had only a weak direct effect (β = 0.218). The problem, on that reading, is not how much you have. It is whether you feel on top of it.

The argument the field is having with itself

Digital hoarding correlates .55 with physical hoarding and .58 with obsessive-compulsive symptoms, and in Thorpe, Bolster and Neave's regression of 282 people, those two were the only survivors, β = .377 and β = .362, together explaining 41.4% of the variance.

So a reasonable person can ask whether "digital hoarding" is a separate thing at all, or whether it is what people who hoard and people with obsessive-compulsive traits do when you give them storage.

A 2026 study of 344 people argues for distinctness on the grounds of ordering rather than difference: stress, enjoyment and memory predict both physical and digital hoarding, but memory ranks second for the digital form while enjoyment ranks second for the physical one. The same study returned three nulls worth reporting: anxiety, attention and decision-making showed no significant effect on either behaviour, which contradicts both the physical-hoarding literature and Sedera's anxiety-based model.

The 2026 scoping review of 36 articles is blunter. The field is small, about 31% qualitative, disproportionately focused on photographs, and theoretically over-anchored in physical-hoarding models that fail to capture what is distinctive about digital environments.

To be precise about diagnostic status: hoarding disorder is in DSM-5, added in 2013. Digital hoarding is not a diagnosis, not in DSM-5, not in ICD-11, nowhere. The clinical framing traces to a single case report from 2015 which proposed it as a subtype. One patient, one proposal, cited ever since as though it settled something.

What the field still does not have

One experiment, and it is not the one you would want. Luxon and colleagues (2019) randomly assigned 101 Pinterest users, using deception, to believe that either one item or a group of items would be deleted from their account, measuring emotional state before and after. Electronic object attachment mediated the relationship between hoarding severity and distress at the prospect of discarding. It is a genuine randomised manipulation, but it is a deletion-threat analogue, not a test of whether accumulation does anything to anyone.

No longitudinal study. Not one. Every quantitative result above is a single-timepoint survey, and multiple author teams say so themselves.

No intervention study of any kind. We searched deliberately for trials, protocols and preregistrations covering deletion prompts, storage quotas, digital-decluttering programmes and inbox interventions. There are none. The scoping review offers its model as a guide for designing interventions, that is, for building something that does not yet exist.

No measurement of time or money. This is worth being blunt about, because the internet is full of figures. Nobody has measured hours lost searching for files, nobody has costed the storage, and no study reports a security incident caused by retention. The paper most often cited for organisational cost, a 2012 information-systems piece on data hoarding and clutter, contains no original data at all. Its numbers are borrowed from third parties and its best-known figure is a worked hypothetical about an email sent to too many people.

A geographically narrow and institutionally concentrated base. The quantitative work comes overwhelmingly from three groups, one British, one Chinese, one Chilean, sampling students and knowledge workers. The Chilean group authored the scoping review as well as several of the primary studies it summarises, so the review is not independent of the literature it assesses.

A boundary worth stating

This site has a separate article on having too many browser tabs open, and that article records something relevant here: the digital hoarding literature does not cover browser tabs. No study in this field measures them, and the validated instrument contains no tab item.

That is not a technicality. Tabs are volatile, they cost working memory rather than storage, and they usually represent unfinished intentions rather than kept possessions. The research on this page concerns accumulated material you have decided to keep. It does not transfer, and we are not going to pretend it does.

What you can reasonably take from it

Three things survive contact with the evidence.

Know which half you are. Accumulating and difficulty deleting are separable, measured separately, and reliable as separate subscales. Someone who keeps everything and someone who cannot bring themselves to delete anything need different things.

The workplace pattern is not pathology, it is incentives. The people with formal data-protection duties hold four times the unread email of everyone else and delete no differently, and the qualitative work traces it to a blame culture rather than a disposition. Deleting carries an asymmetric risk: keeping a file you never open costs almost nothing, while deleting one you later need can cost a great deal, and storage has been effectively free for fifteen years. Under those conditions, keeping everything is a correct answer to the incentives.

The mechanism, if there is one, appears to be control rather than volume. The strongest workplace path ran through felt loss of control, not through how much was stored. That points at organisation and findability rather than deletion, which is fortunate, because deletion is the thing nobody can bring themselves to do.

Whether any of this produces real harm remains genuinely open. One study of 801 people links it to cognitive slips. Another finds it associated with better self-rated work performance. A third finds no effect on academic performance at all. Until somebody runs a longitudinal study or an intervention, that is the honest state of knowledge, and anyone telling you otherwise is filling the gap with confidence rather than data.

Digital accumulation is one of several ways a system's defaults shape behaviour, see the screens, attention and focus guide for the wider picture, too many tabs open for the volatile version of the same instinct, and digital amnesia for what happens to memory when storage is external.