1. Main points

  • In the UK, in 2018, almost 1 in 10 adults living in a private household (9%) had experienced some form of housing difficulty at some point in their life; the most common form was staying with friends or relatives temporarily, accounting for 62% of the experiences.

  • The main reason provided for experiencing housing difficulties was a breakdown of relationships, family problems or both, with this being reported in almost half (48%) of all cases of housing difficulties; after taking into account age, sex, region, country of birth and level of educational attainment, those who had experienced housing difficulties in the past were also significantly more likely to be currently separated or divorced than those who had not.

  • The most common method of exit from housing difficulties was moving into social or subsidised private housing (39%); a further 14% reported that they exited housing difficulties as a result of new or renewed relationships or reconciliation with family or a partner, and 12% reported gaining employment as the main reason for exiting housing difficulties.

  • Most of the instances of housing difficulties lasted for a period of six months or less (59%), however almost one-quarter (24%) lasted longer than a year.

  • After taking into account age, sex, region, country of birth and level of educational attainment, those who have had past experience of housing difficulties were more likely than those who have not to report low life satisfaction, to be happy only a little or none of the time, to feel lonely, and to have low levels of trust in others.

  • People who have experienced housing difficulties in the past were more likely than those who have not to currently be renting, living in a property with leaks or damp, and living in an area with crime, violence or vandalism, accounting for age, sex, region, country of birth and level of educational attainment; they were also more likely to currently be deprived and unable to meet an unexpected expense of £800.

Back to table of contents

2. Overview of past experiences of housing difficulties

This release uses data from the European Union Statistics on Income and Living Conditions (EU-SILC) survey in the UK in 2018, which included an ad-hoc module on past experiences of housing difficulties. Housing difficulties are defined as any periods of time when the respondent had no home of their own (either rented or owned), and encompasses the following types of homelessness:

  • staying with friends or relatives temporarily
  • staying in emergency or other temporary accommodation
  • staying in a place not intended as a permanent home
  • “rough sleeping” or sleeping in a public place

The module included questions asking about:

  • the nature of any past experiences of housing difficulties
  • the duration
  • main causes of housing difficulties
  • method of exit from the housing difficulty

These questions refer to housing difficulties at any time in the respondent’s life. If the individual had experienced more than one period of housing difficulty during their life, the answers relate to the most recent period.

The statistics in this article are based on approximately 18,500 responses to the housing difficulties question, with 1,829 respondents indicating that they had experienced some form of housing difficulty in the past.

This is the first time we have used retrospective data to examine housing difficulties. It does not replace existing measures of homelessness, but is intended to be complementary to them, providing an estimate of those who have ever experienced any of these four types of homelessness, rather than estimating those who are homeless at a single point in time. (See Related links for other relevant statistics and publications.)

While this type of data enables an assessment of housing difficulties over time, including information about the duration and reasons for entering and exiting these difficulties, there are limitations with this approach (see the Data sources and quality section for further details).

Regression analysis has been used to identify the factors that are most strongly associated with past instances of housing difficulties, examining separately factors that are fairly consistent over time, and current circumstances.

When comparing results, only statistically significant differences are commented on, that is, where confidence intervals between comparison groups do not overlap.

Back to table of contents

3. Estimates of housing difficulties

In the UK in 2018, almost 1 in 10 (9%) adults living in a private household had experienced some form of housing difficulty at some point in their life, with the most common form being staying with friends or relatives temporarily, also commonly referred to as “sofa-surfing”, which accounted for 62% of the experiences (Figure 1).

The second largest form was staying in emergency or temporary accommodation accounting for 21% of experiences; this category refers to overnight shelters for people with no usual place of residence, accommodation for persons who experienced domestic violence, or accommodation for immigrants because of their status. Staying in a place not intended as a permanent home accounted for 11%, which refers to staying in a makeshift shelter, shack or shanty, semi-permanent structure, hut or cabin or building defined as unfit for habitation. Rough sleeping or sleeping in a public space accounted for 5% of all housing difficulties experienced.

Most reported instances of housing difficulties lasted for a period of six months or less (59%), however almost one-quarter (24%) lasted longer than a year, indicating the variability of experiences of housing difficulty.

The primary cause of housing difficulties was a breakdown in relationships or family problems or both, with it being the main reason provided for almost half (49%) of all cases of housing difficulties. Other causes included financial problems or insufficient income (16%), end of rental contracts (8%) and unemployment (5%), though a large number (15%) reported “other” as the cause. These findings are largely consistent with other sources, though childhood poverty and housing market conditions have also been found to be important (see for example Bramley and Fitzpatrick, 2017).

Although the module included a second question on any other reasons for the period of housing difficulties, most respondents (84%) reported that there were no other reasons.

The most common method of exit from housing difficulties was moving into social or subsidised private housing (39%), followed by new or renewed relationships or reconciliation with family or partner (14%) and gaining employment (12%). Almost one-third (31%) reported “other” as the method of exiting housing difficulties.

Back to table of contents

4. Characteristics of those with past experiences of housing difficulties

Sex

Similar proportions of men and women have experienced housing difficulties at some point in their lives (9%) and for both, the most common form was temporarily staying with family or friends (Figure 2). However, one-quarter (25%) of women who have experienced housing difficulties in the past stayed in emergency or temporary accommodation, compared with 17% of men. Men were also significantly more likely to have “slept rough” at some time in their lives (8%) compared with women (4%).

Women were also more likely to state that their method of exit from the housing difficulty was moving into social or subsidised housing. Almost half (45%) of women who have experienced some form of housing difficulty in the past exited into social or subsidised housing, compared with one-third (32%) of men. This likely reflects the statutory duty across most of the UK to provide permanent accommodation to those in priority need; these are mainly families with children, which in practice tend to be single-parent families, commonly single mothers.

Age

The proportion of people who have experienced housing difficulties appears to increase with age as people go through their life, with 16 - to 24-year-olds having a significantly lower proportion that have ever experienced housing difficulties (6%) than either 25- to 44-year-olds (10%) or 45- to 64-year-olds (11%) (Figure 3). This is to be expected as the question refers to having ever experienced housing difficulties and the older age groups have had longer to have experienced housing difficulties than the younger age groups.

The proportion who have had past housing difficulties declines for those aged 65 years and over, with those aged 75 years and over having a significantly lower proportion that have experienced housing difficulties (5%) than those aged between 25 and 64 years old. This may reflect imperfect recall among those in the oldest age group or could reflect an increase in the prevalence of experiences of housing difficulties in recent years.

Back to table of contents

5. Modelling the likelihood of experiencing housing difficulties

Various factors may affect the likelihood of a person experiencing housing difficulties. Regression analysis allows us to assess the existence and strength of associations between an outcome – in this case having experienced housing difficulties – and an individual’s characteristics and circumstances.

Given the retrospective nature of the data, and in the absence of information about the circumstances of the individual at the time they experienced the housing difficulties, we chose to model only those factors that are likely to have been fairly consistent over time. In this case we chose region (which for many people stays broadly the same), educational attainment and country of birth, as well as age and sex, which were included as control variables.

The results from this analysis (Figure 4) show the differences within sub-groups of the population while controlling for the other factors.

Figure 4: Those with lower levels of education are more likely to have experienced housing difficulties in the past

Odds ratios of factors associated with past experiences of housing difficulties, UK, 2018

Embed code
Download the data

Notes:

  1. The error bars show the degree of confidence of the estimates; where these cross at 1 (“equally likely”) the estimate is not statistically significantly different from the reference category.
  2. The 95% confidence intervals highlight the degree of uncertainty around an estimate. Non-overlapping confidence intervals suggest a statistically significant difference between groups.
  3. Confidence intervals for many of the regions are large because of small sample sizes, for example, Northern Ireland had 34 respondents that have experienced housing difficulties in the past. Therefore, caution should be used when interpreting these results.
  4. The values reported in the chart are odds ratios and interpreted as highlighted in the Glossary.
  5. Education level is based on the International Standard Classification of Education (ISCED); some groupings have been combined because of small counts and ease of interpretation.

Highest level of education attained

Considering educational attainment, respondents whose highest level of education attained was primary or lower secondary education, were more likely to have experienced housing difficulties than those with a bachelor’s degree or above (odds ratio of 2.0). This was also the case for those whose highest level of attainment was either upper secondary or short cycle tertiary, who were more likely to have experienced housing difficulties than those with a bachelor’s degree or above (odds ratio of 1.7).

Country of birth

A higher proportion of those reporting housing difficulties at some point in their lives were born in the UK (9%) than in the EU (7%). Those born in the UK or a non-EU country were both more likely to have experienced housing difficulties in the past, than those born in an EU country (not including the UK) (odds ratios of 1.6 each).

Region

Comparing the countries of the UK, there were no significant differences in the proportions of respondents reporting past experience of housing difficulties. However, there were regional differences in the proportions of respondents reporting past experiences of housing difficulties within England, with those in London and the South West of England, and those in the North East of England significantly more likely to have experienced past housing difficulties than those in Yorkshire and The Humber (odds ratios of 1.6, 1.6 and 1.7 respectively).

It is important to note that region relates to where people were living in 2018 when they answered the survey and not necessarily where they were living when they experienced housing difficulties, though for many people this will be the same.

In order to further understand the regional variation in greater detail, data on regional median house prices and housing affordability for England, Scotland and Wales have been linked with the EU-SILC data. Housing affordability refers to the lower quartile housing affordability ratio, which is calculated by dividing house prices by gross annual earnings, based on the lower quartile of both house prices and gross annual earnings.

Controlling for age, sex, highest educational attainment and country of birth, findings show that those living in a region with a median house price above £350,000, are more likely to have experienced housing difficulties in the past than those living in a region with median prices below this (odds ratio of 1.2). Almost all the instances in which median house prices are above £350,000 are in London, with the rest in the South East of England.

Additionally, those regions for which the ratio of the lower quartile house price to lower quartile gross annual earnings is 10 or above are more likely to have experienced housing difficulties in the past than those with a ratio below 10 (odds ratio of 1.2), again controlling for age, sex, highest educational attainment and country of birth.

Regions with ratios above 10 include those in London, the South East and the South West of England. The additional analysis shows that regional variation in past experiences of housing difficulties may in part be explained by differences in house prices and the house price to earnings ratio.

Back to table of contents

6. Past experiences of housing difficulties and current circumstances

In this section, we look at how the current circumstances of individuals vary between those who have experienced housing difficulties at some time in their lives and those who have not. Alongside the estimates of experiences of housing difficulties, we used a series of models to explore the association between past experiences of housing difficulties and current circumstances.

We focus on outcomes such as housing quality and marital status, to assess whether past experiences of housing difficulties are associated with these current circumstances, controlling for the factors in the original model (that is, age, sex, region, highest educational attainment and country of birth). The outcomes were based on initial analysis, which revealed that they were strongly associated with past experiences of housing difficulties.

Marital status

Divorced or separated individuals accounted for 23% of those who have experienced past housing difficulties, compared with only 9% of those who have never experienced housing difficulties. (Figure 5). Also, just over one-third (35%) of those who have had past housing difficulties were married, compared with over half (52%) of those who have not had housing difficulties.

After accounting for age, sex, region, highest educational attainment and country of birth, those who have experienced past housing difficulties were still more likely to be separated or divorced than those who have not experienced past housing difficulties (odds ratio of 2.8). Given the most reported reason for housing difficulties is family or relationship difficulties, this finding is perhaps not surprising.

Those who have experienced housing difficulties in the past were also less likely to be married now (odds ratio of 0.4) and more likely to never have been married and to be living in a single-person household (odds ratio of 1.5 and 2.2. respectively), after controlling for age, sex, region, highest educational attainment and country of birth.

Health and disability

Of those with experience of past housing difficulties, 56% suffered from a chronic or long-standing illness or condition now; these are conditions that last for at least six months. Around one-third (33%) of those who had past experiences of housing difficulties, are currently defined as disabled (see the Glossary for a definition of disability) compared with 20% of those who had not experienced any housing difficulties (Figure 6).

A further 24% of those who had experienced housing difficulties in the past were not defined as disabled but were currently suffering from a chronic or long-standing illness or condition; these are conditions that last for at least six months.

Those who have experienced housing difficulties in the past were also more likely to report being disabled now than the rest of the population (odds ratio of 2.0), after controlling for age, sex, region, highest educational attainment and country of birth (see the Glossary for a definition of disability). It is unclear whether the health outcome occurred before or after the episode of housing difficulty, as some conditions may have been brought on or exacerbated by the housing difficulty, but they may also have been a cause of the housing difficulty.

Personal well-being and loneliness

People with experience of housing difficulties in the past have lower average (mean) ratings of life satisfaction (6.8 out of 10) currently than those who have not experienced housing difficulties (7.7 out of 10). Almost one-third (32%) of those who have experienced past housing difficulties currently also have low trust in others (scoring 0 to 4 out of 10) compared with 17% of those who haven’t had this experience. They were also more likely to be happy only a little or none of the time with 15% reporting so compared with only 6% of those that had never experienced housing difficulties (Figure 7).

This indicates that those who have experienced housing difficulties in the past have poorer levels of personal well-being currently than those who have never experienced difficulties.

A greater proportion of those who have experienced past housing difficulties currently report feeling lonely most or all of the time (13%), than those who haven’t experienced housing difficulties (5%).

Figure 7: Those who have experienced housing difficulties in the past are more likely to have lower levels of personal well-being, be less trusting of others and be lonely

Percentage of adults by subjective personal well-being and loneliness measures by whether they have experienced housing difficulties, UK, 2018

Embed code
Download the data

Notes:

  1. The 95% confidence intervals highlight the degree of uncertainty around an estimate. Non-overlapping confidence intervals suggest a statistically significant difference between groups.

  2. Low life satisfaction and low trust in others refers to those scoring 0 to 4 out of 10.

After adjusting for age, sex, region, highest educational attainment and country of birth, these patterns persisted among those who have experienced past housing difficulties. They were more likely to report low life satisfaction (odds ratio of 3.4), more likely to be happy either a little of the time or none of the time (odds ratio of 2.7), more likely to have low trust in others (odds ratio of 2.5) and more likely to report feeling lonely all the time or most of the time (odds ratio of 3.3) compared with those who have never experienced housing difficulties.

Current housing quality and tenure

Only 14% of those who had experienced housing difficulties in the past were now living in a detached house compared with one-quarter (25%) of those who had not experienced housing difficulties. Over half (56%) of those that had experienced housing difficulties were currently renting their property compared with 29% of those that had not (Figure 8).

Current living conditions also appeared to be worse for people with experience of housing difficulties in the past. Over one-quarter (27%) of those who reported housing difficulties in the past now live in a property with leaks, damp or rot, compared with only 15% of those who have not experienced housing difficulties.

Looking at the wider area in which they live, people with past experience of housing difficulties were also more likely to report they now live in an area that experiences problems with crime, violence or vandalism (Figure 8).

Figure 8: Those who have experienced housing difficulties in the past are more likely to be renting, to live in an apartment and to live in poor housing conditions now

Percentage of adults by tenure type and living conditions by whether they have experienced housing difficulties, UK, 2018

Embed code
Download the data

Notes:

  1. The 95% confidence intervals highlight the degree of uncertainty around an estimate. Non-overlapping confidence intervals suggest a statistically significant difference between groups.

After adjusting for age, sex, region, highest educational attainment and country of birth, people that have experienced housing difficulties in the past were currently less likely to own a property (odds ratio of 0.3) and more likely to be renting a property (odds ratio of 3.4) than those who have never had housing difficulties. They were also less likely to be living in a detached property (odds ratio of 0.5) and more likely to be living in an apartment (odds ratio of 2.6).

Similarly, they were more likely to currently live in a property with leaks or damp (odds ratio of 1.9) and more likely be living in an area with crime violence or vandalism (odds ratio of 1.6), after controlling for age, sex, region, highest educational attainment and country of birth. These factors are linked to areas of higher deprivation, suggesting that those that have experienced housing difficulties in the past are more likely to suffer from contemporary housing deprivation.

Deprivation and financial circumstances

There is some evidence that those who have experienced housing difficulties in the past have worse outcomes financially compared with those that have never experienced any housing difficulties. Current equivalised household income for people who have experienced housing difficulties in the past is lower, with 44% having an equivalised household income of less than £15,000 compared with 32% of those who have not experienced housing difficulties (Figure 9).

Findings from other research show that the relative income position of households is persistent over time, suggesting that housing difficulties can be both a result of low income as well as exacerbating income poverty.

Those that have experienced housing difficulties in the past were also more likely to be materially deprived currently than those that have not; material deprivation is defined as being unable to afford at least three of a list of items considered by most people in Europe to be desirable or even necessary to lead an adequate life.

Just over one-quarter (26%) of people who have experienced housing difficulties in the past were currently materially deprived, significantly greater than the proportion of those who have not experienced housing difficulties (9%). This pattern persisted after controlling for age, sex, region, highest educational attainment and country of birth, with a strong association between material deprivation and past experiences of housing difficulties (odds ratio of 3.4).

Alongside the findings for loneliness and low trust in others, this suggests that people who have had past experiences of housing difficulties are currently more likely to live in circumstances associated with social exclusion.

Figure 10: Financial outcomes are worse for those who have experienced housing difficulties in the past

Percentage of adults by financial outcomes and whether they have experienced housing difficulties in the past, UK, 2018

Embed code
Download the data

Notes:

  1. The 95% confidence intervals highlight the degree of uncertainty around an estimate. Non-overlapping confidence intervals suggest a statistically significant difference between groups.

Over half (55%) of those who have experienced past housing difficulties currently reported difficulty in making ends meet, compared with less than one-third (32%) of the rest of the population (Figure 10). Similarly, over half (54%) were currently unable to meet an unexpected expense of £800 compared with 29% of those that have not experienced past housing difficulties.

Households containing someone who has experienced housing difficulties were also significantly more likely to find total housing costs and repayment of debts from hire purchase or loans a burden, with over 60% currently reporting this.

After controlling for age, sex, region, highest educational attainment and country of birth, those who have experienced past housing difficulties are more likely to find difficulty in making ends meet currently (odds ratio of 2.5) and to be unable to afford an unexpected cost of £800 (odds ratio of 2.8). They were also more likely to currently live in a household with low work intensity (odds ratio of 2.9); low work intensity is defined as living in a household where those of working age worked less than 20% of their total potential during the previous 12 months.

Back to table of contents

7. Improving the data

The analysis presented here has demonstrated the potential for using retrospective questioning in a household survey to provide estimates of past experiences of housing difficulties, adding to the existing evidence base on homelessness. The questions included in this survey allowed for an assessment of the extent of past experiences of housing difficulties, including how long these experiences typically lasted, the reasons people experienced them and their routes out of housing difficulties. The collection of these data alongside data on well-being and living standards, including household finances, housing quality and deprivation, allowed for analysis of some of the potential longer-term impacts of housing difficulties. Nevertheless, there is scope for improvement.

Within the existing questions, our analysis revealed a large number of respondents chose the “other” category as a response option to the reason for the housing difficulty and method of exit. A wider range of response options or the ability to provide a free text response would allow a better understanding of some of the drivers of housing difficulties.

Additional questions identifying when the housing difficulties were experienced and the circumstances of the individual at that point in time would enable greater understanding of some of the drivers of housing difficulties, including whether people of specific ages are more susceptible to it, so that policy can be better directed to address these. Similarly, this type of analysis could be enhanced by also capturing factors during childhood to explore how opportunities and circumstances as a child impact on the likelihood of experiencing housing difficulties in the future.

Furthermore, capturing multiple episodes of housing difficulties rather than just the most recent would enable estimation of the scale and nature of this issue. Similarly, collecting information on whether respondents engaged with local authority or other services, would enable estimation of the level of “hidden homelessness”, where people are not counted in the homelessness statistics because they do not engage with support or other services. In both cases, this would also provide the potential to explore the characteristics of those affected and the factors driving them.

Back to table of contents

8. Housing difficulties data

Exploring past experiences of housing difficulties in the UK estimates
Dataset | Released 22 October 2020
Estimates of past experiences of housing difficulties, factors associated with past experiences of housing difficulties and accompanying regression tables.

Back to table of contents

9. Glossary

Collective household

A non-institutional collective dwelling such as a boarding house, dormitory in an educational establishment or other living quarters shared by more than five persons without sharing household expenses. This also includes persons living as lodgers in households with more than five lodgers.

Disability

To define disability in this publication, we refer to the Government Statistical Service (GSS) harmonised definition: this identifies as “disabled” a person who has a physical or mental health condition or illness that has lasted or is expected to last 12 months or more that reduces their ability to carry-out day-to-day activities.

The GSS definition is designed to reflect the definitions that appear in legal terms in the Disability Discrimination Act 1995 and the subsequent Equality Act 2010.

The GSS harmonised questions are asked of the respondent in the survey, meaning that disability status is self-reported.

Education level

The highest level of education attained is based on the International Standard Classification for Education (ISCED) for organising education programmes and related qualifications by levels and fields.

Equivalisation

Equivalisation is the process of accounting for the fact that households with many members are likely to need a higher income to achieve the same standard of living as households with fewer members. Equivalisation takes into account the number of people living in the household and their ages, acknowledging that while a household with two people in it will need more money to sustain the same living standards as one with a single person, the two-person household is unlikely to need double the income.

Loneliness

In this release, our definition of loneliness is aligned to the definition used by the Jo Cox Commission and in the Loneliness Strategy for England. This is based on a definition first suggested by Perlman and Peplau in 1981: “A subjective, unwelcome feeling of lack or loss of companionship. It happens when we have a mismatch between the quantity and quality of social relationships that we have, and those that we want”.

Material deprivation

An individual is materially deprived if they do not have the ability to afford three or more of the following items:

  • to not have arrears on your mortgage or rent payments, hire purchase instalments or other loan payments
  • to afford a one-week annual holiday away from home
  • to afford a meal with meat, chicken, fish or a vegetarian equivalent every other day
  • to be able to face unexpected financial expenses
  • to afford a telephone or mobile phone
  • to afford a colour television
  • to afford a washing machine
  • to afford a car
  • to afford to keep the home adequately warm

Odds ratio

In its simplest form, a binary logistic regression can be interpreted using an “odds ratio”. This is a comparison of the odds of reporting high anxiety for someone with a specified characteristic or circumstance and someone without that specified characteristic or circumstance, while controlling for other possible influences of past experiences of housing difficulty.

Reference group

A reference category is identified as a category within a variable that can be compared with all other categories. The most logical category was chosen for each variable; in some instances, this was based on the highest or lowest category within the scale, for example, the youngest age category. In other instances, the most normative population was identified as the reference category.

Where there was not a structured order or a normative population, reference categories were chosen based on the direction of the odds to aid clear interpretation, for example, males were chosen as the reference category as females were more likely to have experienced past housing difficulties.

Statistical significance

In reference to the statistical significance of variables in a regression model, the level of significance (also known as a p-value) tests the null hypothesis that the coefficient has no association with the outcome. A low p-value (for example, p-value less than 5%) indicates that you can reject the null hypothesis and be confident in concluding there is an association between the variable and past experiences of housing difficulties.

In addition, many figures also show confidence intervals, statistical significance of differences can also be determined based on non-overlapping confidence intervals.

Thresholds

For the life satisfaction and trust questions, ratings are grouped for ease of interpretation. The ratings are grouped as follows:

  • 0 to 4 (Low)
  • 5 to 6 (Medium)
  • 7 to 8 (High)
  • 9 to 10 (Very high)
Back to table of contents

10. Data quality and sources

Sample

The statistics in this article are based on approximately 18,500 responses to the housing difficulties question, with 1,829 respondents indicating that they had experienced some form of housing difficulty in the past.

Weighting

Information on well-being and housing difficulties was asked of each current private household member, or, if applicable for all selected respondents aged 16 years and over. Proxy interviews were not permitted for variables concerning well-being and housing difficulties, given the nature of the information being collected.

A separate weighting was given for the ad-hoc module but proxies were included in the weighting, which could cause further under-representation.

Retrospective questioning

EU Statistics on Income and Living Conditions in the UK are derived from a survey of private households, so they do not capture people living in temporary accommodation or those sleeping rough (see Exclusions). The use of retrospective questioning provides a larger sample and a broader examination of housing difficulties over time, with the respondents now having exited the episode of difficulty, such that causes, duration and methods of exit can be explored and understood in greater detail than if the event was ongoing.

The housing difficulties questions rely on retrospective recall of past experiences. This reliance on the respondent’s recall of important variables of interest, has the disadvantage that it may be inaccurate.

The cross-sectional nature of the survey also means that we have no information about the circumstances of the individual at the time they experienced the housing difficulty. As a result, in this analysis we only modelled past experiences of housing difficulties as an outcome based on factors that have been assumed to be present before the episode, with the remaining models using past experiences of housing difficulties as a factor to model current circumstances. In this way, we have identified factors that are significantly related to one another though we cannot determine the sequence of events or that specific factors caused housing difficulties.

The sample of people included in the survey are now living in private households, which suggests that these people may have been more successful than others in extricating themselves from their housing difficulties. However, we are also not clear whether people have had repeated episodes of housing difficulties or whether living in private accommodation now is an enduring circumstance for them. This means we cannot be certain about the relative and ongoing success of different routes out of housing difficulties. We have simply identified factors associated with exiting housing difficulties that people now living in private accommodation have used at some point in the past.

Exclusions

Non-original sample members from the UK moving into and out of sample households (or non-responding in some but not other waves) are given a zero base weight. This could bias downwards the representation of some people at high risk of experiencing forms of housing difficulty, putting them in the same category as people temporarily present at a household at the time of interview but not usually resident (staying with friends or relatives temporarily), who are typically excluded from most or all household surveys.

Additionally, persons living in collective households and in institutions are excluded from the target population. Again, this could disproportionately exclude people recently experiencing or at high risk of experiencing housing difficulties.

People moving from sample households to a collective household, institution or abroad are not traced. Again, this would provide downward bias of some forms of housing difficulty and groups at high risk are likely to be underrepresented.

People aged under 14 years who subsequently move out of their original sample household will not be traced – this could omit some of a fairly high-risk group of young homeless in subsequent waves.

Taken together, these factors could affect the number of people captured by the survey who have recently experienced housing difficulties as well as those for whom difficulties have been recurring over a longer period of time. Therefore, the estimates derived are likely to be an underestimate of the total level of housing difficulties by the population as a whole.

Uncertainty and quality

The statistics presented are estimates and as with all estimates, there is a level of uncertainty associated with them. Where available, 95% confidence intervals have been shown. These show the range within which we would expect the true value to lie for approximately 95 out of every 100 samples drawn at random from the population. Wide confidence intervals, often associated with small sample sizes or large sample variance, indicate a wider range of values within which we would expect the true value to lie.

Throughout this release we have assessed statistical significance using non-overlapping confidence intervals. This method has the limitation that some estimates with overlapping confidence intervals may be significantly different but will not be identified as such even though they would have been, had a formal null hypothesis statistical test been carried out (that is, the false-negative rate will be inflated). Likewise, because of the uncertainty inherent in working with a sample instead of the full population, it is inevitable that sample-based estimates will occasionally be flagged as significantly different from one another, when in reality there is no difference between the corresponding values in the population; such findings are known as “false-positives”.

In some cases, sample sizes for specific groups are small and confidence intervals are large and overlap with one another. This makes it difficult to make robust comparisons between groups. Only statistically significant differences, as defined in this section, are commented on in this release.

Missing data

Missing data can produce biased estimates and invalid conclusions, particularly if the data are not “missing at random” or, in other words, if there is some pattern to the missingness. Where data have been found to be missing at random, they have been omitted from the analysis. Other cases where the data are not missing at random data have been recoded where appropriate.

Regression technique used

A binary logistic regression was carried out to assess which personal characteristics and circumstances have the strongest association with past experiences of housing difficulties. The response variable is binary where a respondent either has experienced any form of housing difficulty or has not experienced housing difficulty. In this model we used controls such as age and sex, alongside other variables, region, highest level of education attained (ISCED) and country of birth, which were considered to be stable or relatively stable circumstances throughout someone’s adult life, so could be used to explore whether or not someone would have experienced a housing difficulty in the past.

The next stage of the modelling used variables that are contemporary and will have likely changed over the course of an adult’s life, such as marital status. In these instances, we have no information on when the events occurred in the past so can only say whether there is a significant association between having experienced past housing difficulties and the present circumstance. Individual models were run to predict each of the contemporary circumstances or characteristics in turn, in order to identify the odds of experiencing these outcomes given an experience of housing difficulty.

Limitations of the regression model

The analysis is limited to those variables that were collected in the survey. There are likely to be factors that are associated with housing difficulty that were not observed in this study but would improve the model fit if they could be observed, collected and analysed.

Additionally, because of sample size limitations and the relatively small number of respondents that have experienced past housing difficulties, some groups might be underrepresented in the data and therefore no significant difference between them can be ascertained.

Back to table of contents

Contact details for this Article

Mark Hamilton and Ben Hayes
equalities@ons.gov.uk
Telephone: +44 (0)1633 455044