Examining Shelter’s statistical framing

Statistics can change how an entire sector is perceived.

A single number repeated often enough can shape public policy, influence legislation and define how millions of landlords are viewed.

Shelter understands this better than most organisations operating in the housing debate.

Headlines citing Shelter research frequently appear in national newspapers:

“45% of renters face illegal acts.”

“Evictions surge.”

“Millions of tenants at risk.”

But what happens when you look behind those headlines and examine the underlying data?

This was precisely the kind of question that interested David Knox. He believed statistics should always be read with the same discipline as financial accounts.

When you apply that discipline, the picture becomes more complicated than the headlines suggest.

Case study: the “45% illegal actions” claim

Shelter has previously cited polling suggesting that 45% of private renters have experienced illegal actions by landlords.

On its face, that figure is alarming.

However, survey-based polling differs fundamentally from enforcement statistics or representative national datasets such as the English Housing Survey.

Polling samples often:

• Capture self-reported experiences.
• Include a broad definition of “illegal action”.
• Reflect respondent interpretation rather than legal adjudication.

The term “illegal action” may include:

• Failure to protect a deposit correctly.
• Administrative breaches.
• Harassment allegations.
• Misunderstood tenancy procedures.

The public hearing “45%” may assume widespread criminal conduct.

The underlying dataset may reflect a wider spectrum of compliance issues, some technical, some disputed, some historic.

Without contextual comparison to:

• Total number of tenancies.
• Local authority enforcement statistics.
• Court findings.

The headline percentage alone creates a distorted picture.

David Knox’s consistent approach was to ask: what is the base number? What is the definition? What is the comparator?

Those questions apply here.

Section 21 and the arithmetic of growth

Another recurring example is eviction reporting.

Shelter press releases frequently highlight percentage increases in Section 21 notices or possession claims.

An increase of 20% sounds dramatic, however, percentage increases do not reveal:

• Whether volumes remain below pre-pandemic levels.
• Whether the rise reflects court backlog normalisation.
• Whether economic conditions temporarily distort the figures.

For example, if possession claims rise from 10,000 to 12,000, that is a 20% increase. It is also 2,000 additional cases in a rental market comprising millions of tenancies.

Both statements are true.

Only one dominates headlines.

The omission of denominator context

When percentages are used without reference to total market size, the scale of risk can appear disproportionate.

The private rented sector in England contains millions of households.

If a statistic references thousands of cases, the percentage relative to the total sector often appears small.

If the same statistic is expressed as year-on-year percentage growth, it appears urgent.

Neither framing is inherently false.

The choice of framing shapes narrative.

That narrative influences legislation.

Historical pattern

This pattern is not new.

Earlier Property118 articles challenged Shelter’s presentation of “rogue landlord complaint” increases where headline growth rates were emphasised without immediate disclosure of base figures.

Similarly, NRLA open letters have responded to Shelter press releases by adding omitted Ministry of Justice context.

This is not a dispute about data existence.

It is a dispute about emphasis.

David Knox’s concern was always that selective emphasis can influence public debate disproportionately to the underlying scale.

Media amplification

Once a percentage appears in a press release, it is frequently reproduced without methodological caveat.

Headlines compress nuance.

“Half of renters face illegal treatment”

“Evictions surge by 25%”

Such language becomes part of the political atmosphere in which housing reform is debated.

By the time policymakers cite the statistic, the original survey design or enforcement context is rarely discussed.

The figure has taken on symbolic weight.

Why this matters

Shelter plays a significant role in shaping housing reform.

Its data and polling inform public opinion and parliamentary debate.

If headline statistics are framed in a way that magnifies relative change while minimising denominator context, that framing carries legislative consequence.

Landlords are regulated not only by statute, but by narrative.

A narrative built on incomplete context can produce disproportionate response.

That does not require deliberate distortion. It requires only selective emphasis.

The discipline David applied

When David Knox FCA analysed Shelter’s accounts, he did not accuse. He reconstructed.

The same discipline should apply to statistical claims.

For any headline percentage, three questions should follow:

1. What is the base number?
2. What is the denominator?
3. What is the historical comparator?

If those are not immediately visible, interpretation becomes vulnerable to exaggeration.

Moving from statistics to structure

In the final article in this series, we will step back from individual numbers and consider a broader question: what role does Shelter occupy within the housing ecosystem?

Is it primarily:

• An advice charity?
• A campaigning organisation?
• A policy influencer?
• A publicly funded contractor?

That structural identity shapes how both financial figures and statistical claims should be interpreted.

About David Knox FCA

David Knox FCA, who wrote for Property118 under the pseudonym “Appalled Landlord”, passed away on 21 January 2020. His investigative work, including his scrutiny of Shelter’s published accounts, remains available in the Property118 archive. This series revisits the same type of publicly available source material in the analytical spirit of his work. A tribute to David can be read here.