This report presents results based on the longitudinal findings from Wave One and Wave Three of the Life Opportunities Survey (LOS). Analysis focuses on Wave One and Wave Three to make full use of the longitudinal design of LOS. Considering impairment status at Wave One and Wave Three only also simplifies the classification of adults for analysis.Back to table of contents
The Life Opportunities Survey (LOS) is a large-scale longitudinal survey1 of disability in Great Britain. It is the first major national social survey which explores disability in terms of the barriers to participation that people experience. Prior to the development of the LOS, the last major survey on disability conducted in Great Britain had been the Family Resources Survey disability follow-up in 1996 to 1997. Before this, it was the Office of Population Censuses and Surveys (OPCS) national surveys of disability, carried out during 1985 to 1988. In 2005, the Department for Work and Pensions (DWP) carried out a feasibility study2 which recommended a new survey dedicated to meet the evidence needs of policy for disabled people in Great Britain. Subsequently, the Office for Disability Issues (ODI) (part of the DWP) commissioned the Office for National Statistics (ONS) to develop the LOS.
The survey is wholly funded by the DWP and is carried out by ONS. The LOS was conducted over 3 waves with each wave of the survey spanning 2 years. Respondents in Wave Two were followed up approximately 1 year after their initial interview. Interviews with Wave Three respondents were carried out approximately 2 and a half years after Wave Two, and 3 and a half years after Wave One. Wave One of the LOS took place between June 2009 and March 2011; Wave Two began in June 2010 and finished in March 2012; Wave Three commenced in October 2012 and was completed in September 2014. This is the final wave of the survey.
Notes for 1.1 Background of the Life Opportunities Survey
A survey where respondents are regularly re-interviewed after a set interval (for example, 1 year).
Purdon et al. (2005) Meeting DWP’s long-term information needs on disability: A feasibility study.
The following section of the report contains a brief summary of how respondents were followed up across the 3 waves and how they are classified into groups for the purpose of longitudinal analysis.
1.3.1 Follow-up of respondents from Wave One to Wave Two
A total of 37,500 households were sampled for the LOS Wave One survey, using a single stage (unclustered) design. In this first wave, all adults (with and without impairment) responding to the survey were interviewed. Classification of impairment status then took place after the interview. Each respondent’s impairment status at Wave One was then used to assign them to 1 of 3 different groups which determined how they were followed up in Wave Two.
Adults with at least 1 impairment
This group included all adults who reported at least 1 impairment at Wave One1 (29%). These adults, as well as all adult members of their households, were interviewed in person at Wave Two.
Of the 71% of adults who did not report having an impairment at Wave One, a subset was selected to form a "control" group. The adults in this group were chosen to provide a comparison group that is similar to the adults with at least 1 impairment on several main characteristics associated with impairment: sex, age, region of residence, and the urban or rural classification of residence. The size of the control group was designed to be half that of the group of adults with at least 1 impairment, that is, for the Wave Two sample there were 2 adults with impairment for every 1 person in the control group. All of the adults in the control group, as well as all adult members of their households, were interviewed in person at Wave Two.
All of the remaining adults who did not have an impairment at Wave One and who were not selected for the control group, were assigned to the "onset screening group". These adults were not interviewed in person at Wave Two, unless they or an adult member of their household had acquired an impairment between Wave One and Wave Two. A brief telephone interview was conducted with the onset-screening group between Wave One and Wave Two in order to establish if anyone in their household had acquired an impairment.
For more information on the composition of the 3 follow-up groups, see the Wave Two Interim Technical Report (2.26 Mb Pdf).
1.3.2 Follow-up of respondents from Wave Two to Wave Three
All adults interviewed at Wave Two LOS were followed up with a face-to-face interview at Wave Three. This included adults who reported at least one impairment at Wave Two, as well as all adult members of their household and adults without an impairment at Wave Two and all adults in their household.
Adults in the onset screening group were also followed up at Wave Three. A brief telephone interview was conducted with the onset-screening group between Wave Two and Wave Three in order to establish if anyone in their household had acquired an impairment. These adults were not interviewed in person at Wave Three, unless they or an adult member of their household had acquired an impairment between Wave Two and Wave Three.
Diagram 1.1 shows a summary of time periods of waves and the LOS longitudinal design. The data collection periods overlap slightly for Wave One and Wave Two, although adults were interviewed for Wave Two approximately a year after their first interview. The time period between interviews at Wave Two and Wave Three varies between 28 months and 30 months. This variation in time lag is due to the move from quarterly to monthly data collection implemented at Wave Two. The time lag between Wave One and Wave Three is between 3 years and 4 months and 3 and a half years.
Diagram 1.1: Timescale for LOS interviews
Classification of individuals for Wave Three longitudinal analysis
Diagram 1.2 shows how adults have been classified for the Wave One to Wave Three longitudinal analysis. The classification takes into account impairment status at Wave One and Wave Three only. Diagram 1.2 also illustrates how adults who did not report any impairment at either Wave One or Wave Three flow through the survey. The more complex flow of adults through all 3 waves is illustrated in Annex 2 (79.3 Kb Pdf) .
The following 4 groups are featured in the longitudinal analyses of this report:
adults who had at least 1 impairment at Wave One and at Wave Three are "adults with impairment at both waves"
adults who had at least 1 impairment at Wave One but no impairments at Wave Three are "offset adults"
adults who did not have any impairment at Wave One but had at least 1 impairment at Wave Three are "onset-acquired adults"
adults who did not have any impairment at Wave One or at Wave Three are "adults without impairment at both waves"
Diagram 1.2: Classification of adults for LOS Wave Three longitudinal analysis
Notes for 1.3 Brief outline of the LOS methodologyBack to table of contents
Outline of chapters
This report explores impairment transitions over time, as well as adults’ experiences at Wave Three of various life areas by impairment status.
Chapter 1 (this chapter) sets out the background for the LOS and outlines the LOS methodology. This chapter also discusses the importance of the main measures used in the analysis and highlights both the strengths and weaknesses of using LOS data.
Chapter 2 will examine the scale of onset and offset transitions that have occurred between Wave One and Wave Three.
Chapter 3 will describe differences across Wave One and Wave Three in work status transitions, with relation to impairment status transitions and other characteristics and circumstances such as age, economic status, occupation and caring responsibilities.
Chapter 4 provides an overview of participation restriction across life areas followed by a focus on participation restriction to the life area of leisure. Chapter 4 also looks at barriers to participation, such as difficulties in accessing culture, sport and leisure activities, and the effect of impairment status on how adults choose to live their lives.
Chapter 5 explores how the level of social contact experienced by adults varies according to impairment status. It also looks at barriers to social contact, such as difficulties in getting in or out of the home, by impairment status.
Chapter 6 provides a summary of the findings of the report.
As noted above, there is a focus on 3 of the 8 LOS life areas: work, leisure activities and social contact. These topics were chosen to understand more about the experience of onset and offset transitions over time and to examine some of the barriers to participation and access in these life areas at Wave Three.
The data have been weighted to provide estimates for the general population. This is done to improve the estimates, adjusting for sample non-coverage and non-response. As LOS is a longitudinal survey, this adjustment has been done based on the population at Wave One. LOS is therefore not suitable for calculating cross-sectional estimates of impairment prevalence at Wave Two and Wave Three. Section 1.6 has more information on the specific strengths and weaknesses for LOS.
Further information on weighting of LOS Wave One to Wave Three data can be found in Section 1.5.
Longitudinal analysis groups
Throughout the report comparisons are made between the 4 groups of adults as described in Section 1.2 - adults with impairment at both waves, offset adults, onset-acquired adults and adults without impairment at both waves. The 4 groups reflect the diversity of impairment status, in that impairment status may be stable, or may change over time. A person may have impairments at both waves or be without impairment at both waves. Alternatively, they may no longer have impairments or they may acquire an impairment at Wave Three.
Participation restriction and barriers
An adult has a participation restriction in a life area if they experience at least 1 barrier in that area. For example, in community, leisure and civic life, respondents were asked if there were any activities that they would like to do more of but were not able to1. If they answered "yes", they were then asked to specify the barrier(s) experienced. Barriers in this life area included:
too busy/not enough time
no-one to go with
fear of crime
fear of crowds
lack of availability
lack of help or assistance
a health condition, illness or impairment
disability related reasons
attitudes of other people
difficulty with transport
difficulty getting into buildings
difficulty using facilities
feel that I am not welcome
For another life area, the barriers reported may be different. For example, "lack of job opportunities" is a barrier applicable to work, whereas "difficulty getting to stop or station" is a barrier applicable to the use of public transport such as local buses and long distance trains. Analysis of barriers is presented in charts showing the top barriers that were reported by each group at Wave Three.
The top barriers form the focus of our analysis given the small number of respondents selecting other barrier types. The top 4 barriers are considered in Chapter 4 and the top 6 in Chapter 5. Where direct comparisons between analysis groups or age groups are made within the text, these are statistically significant at the 95% confidence level unless otherwise stated. Where the text describes the most common barriers, or the ordering of barriers within an analysis group, no significance testing has been carried out.
Why is it important to study barriers to participation to understand disability?
The results from the longitudinal analysis of the LOS will be used to further inform a new understanding of the dynamic nature of the disabled population and of the barriers faced by people with impairment. It is important to understand more about the dynamic nature of disability and the barriers people with impairments face. Such findings can be used to inform the development of policies relating to health and disability.
Notes for 1.4 Wave Three AnalysesBack to table of contents
Units, presentation and accuracy
Detailed definitions of the terms and categories used in this report are provided in the Glossary.
All estimates presented in the tables and charts have been analysed using longitudinal weights so that they have been adjusted for non-response, differential sampling probabilities of selection at Wave One, and loss to follow-up at Wave Two and Wave Three.
The weighting of the Wave One to Wave Three data was carried out in 2 steps:
Step 1: An adjustment to loss to follow-up was applied to the Wave One to Wave Two weights of all cases that were present in Wave One and responded in Wave Three. The same method was applied that was used in Wave Two, described in Annex 1 of the Wave Two report. The adjusted weights were then calibrated to the Wave One population totals.
Step 2: The dataset used in step 1 was expanded to add the cases from the screening group that received a face-to-face interview in Wave Three (because they had "onset" to impairment). The same weight adjustment procedure was applied that was used in the Wave One to Wave Two dataset, described in Annex 1 of the Wave Two report.
Percentages reported throughout the document have been calculated using population estimates rounded to integers. When calculating percentages, the values of residual categories such as "no answer" or "don’t know" have been excluded from the denominator.
Sample attrition, household and individual response rates
The response rate for LOS Wave Three can be calculated at both household and individual level. The LOS Wave Three achieved a household level response rate of 66%. In this case responding households were those in which information was collected in person or by proxy for all adults. The survey achieved an individual response rate of 64%, among adults who were eligible for a face-to-face interview.
For longitudinal surveys, the sample is subject to attrition moving from 1 wave to the next. For the LOS, a subset of the original sample at Wave One were not contactable, did not respond, or refused further participation at Wave Two and Wave Three. More information on sample attrition and the household and individual response rates can be found in Annex 3 (106 Kb Pdf)
Unless otherwise stated, changes and differences mentioned in the text have been found to be statistically significant at the 95% confidence level. Different tests were used when comparing differences between waves (non-independent) and differences between analysis groups or age groups (independent).
There are 3 notable exceptions to this:
For comparisons of offset and onset rates by impairment type (Tables 2.7 and 2.10), unless they involve comparing age groups, no significance testing has been carried out. This is because individuals can experience more than 1 impairment type, so any differences are not independent and would require a different form of significance testing.
Significance testing has not been carried out on Tables 2.1 to 2.4 and 2.11 to 2.14, comparing the severity and frequency of impairment types. This is because individuals can experience more than one impairment type, so any differences are not independent and would require a different form of significance testing.
The analysis of the types of barriers adults experienced are presented as a description comparing the top barriers. Where the text describes the most common barriers, or the ordering of barriers within an analysis group, no significance testing has been carried out. Where direct comparisons between analysis groups or age groups are made within the text, testing has been undertaken and differences are statistically significant at the 95% confidence level unless otherwise stated.Back to table of contents
Specific strengths of using the LOS:
LOS is a longitudinal survey that has been designed to follow respondents over time to provide an insight and understanding into the transient nature of impairment. An important key component of the survey design is the ability to compare the experience of barriers to participation between people with impairment and those without impairment over time to assess the equality in access in different life areas. The formation, at Wave Two, of a control group from within the respondents who did not have any impairments at Wave One, allows experiences to be compared between the 2 groups. The analysis in this report compares Wave One with Wave Three.
As the first longitudinal survey of its kind, the data collected provides a unique and rich source of data for use by analysts. It provides understanding about the transitions into and out of impairment, as well as different types and severity of impairment experienced by adults in Great Britain.
Survey design. The LOS uses an unclustered sample designed to produce efficient estimates for the sample size and Great Britain. The unclustered sample allows greater precision to be achieved in LOS outputs because it helps to make the sample more representative of all groups in society1.
The size of the Wave One sample was sufficient to accommodate for attrition over time. As a result, at Wave Three, the sample remains of a sufficient size to allow for comparisons and analysis to be conducted among subgroups of the sample.
Specific weaknesses of using the LOS:
Potential under- or over-reporting of impairment. The LOS collects self-reported impairment data which is dependent upon the respondent’s perception of the limitations and severity of the impairment at the time of the interview. A self-reporting measure of this kind is prone to both over- and under-reporting as the perception of the existence and effect of impairment is dependent upon each respondent’s own personal opinion on the day they are interviewed.
One consequence of this is that the respondent's perception of the impairment may differ over time causing transitions to be recorded that may or may not be real. For example, the LOS definition of impairment is dependent upon the respondent reporting more than "mild" difficulty. If at Wave One they report "moderate" difficulty, they will be classed as having an impairment according to the LOS definition. However, if at Wave Two we assume that there has been no change in the impairment and the respondent reports "mild" difficulty, LOS will no longer class them as having an impairment. Using this example, the respondent will be treated as having offset from impairment at Wave Two. This approach has also been applied at Wave Three.
Cross-sectional weights are not available on the LOS Wave Two and Wave Three datasets and therefore estimates of impairment prevalence cannot be calculated. Overall estimates of disability prevalence are available from the Family Resources Survey2.
The LOS is a sample survey and so is subject to weaknesses which also exist for other surveys, including:
Non-sampling error. Estimates from all sample surveys, including those derived from the LOS data, will contain a degree of uncertainty due to non-sampling and sampling error. Non-sampling error generally arises from errors in the collection, measurement and processing of the data3. To minimise non-sampling error, the LOS questionnaire was administered using computer-assisted interviewing (CAI), which ensured that the questionnaire routing was followed correctly during the interview and inconsistencies and potential collection errors were able to be queried, and where appropriate, corrected during the interview1. Again, to minimise non-sampling error, as a result of errors arising from data collection, all of our ONS interviewers received both generic interviewer training and specific LOS training prior to working on the LOS. The LOS Wave One Technical Report provides further information about the data collection and processing of the data3.
Sampling error. This relates to the fact that the sample chosen is only one of a large number of samples that could have been chosen. It follows then that any estimate derived from the sample is only one of a large number of estimates that could have been made. For users of survey data, the precision of the estimate is of interest, that is, how much the estimate would vary when different samples are drawn4. In general, the larger the sample size from which the estimate is derived, the more precise, or robust, is the estimate.
Non-response and missing information. The information from a household which co-operates in the survey may be incomplete. This can be due to one of the following reasons: a partial refusal (for example, to income); because information was collected by proxy, therefore certain questions were omitted from the interview; or a particular item was missed because of a lack of understanding or an error.
Survey coverage. The LOS Wave One sample was designed to cover private households in Great Britain. Therefore individuals in care or retirement homes, for example, will not be included. This means that figures relating to the severely disabled and the most elderly individuals may not be representative of the Great Britain population, as many will have moved into homes where they can receive more frequent help.
The survey at Wave Two followed individuals and their households and remains a private household survey. This means that any respondents who moved into institutions (care or retirement homes, prison, etc) between Waves One and Two were not followed at Wave Two and were recorded as not eligible to participate in the survey. Similarly, respondents who moved into institutions between Waves Two and Three were not followed at Wave Three and were recorded as not eligible to participate in the survey.
The LOS Wave Two set sample comprised only of households of individuals who had participated at Wave One and did not include any new households. This approach does not allow the sample to be refreshed to include any new population sub-groups that may have entered the country since the sample was drawn for Wave One. However, any adult or child who had entered a LOS household since the Wave One interview are included as new survey entrants at Wave Two. This approach has been applied at Wave Three.
Survey bias. Although work has been undertaken to try to ensure that the figures that have been collected are valid and that adjustments are made for attrition, survey bias may still remain. This may be partly due to misreporting of certain data items and sampling error, but also may reflect non-response biases not controlled for in the weighting factors. As with Waves One and Two, efforts were continually made at Wave Three to minimise these issues, for example, through greater reliance on documentary evidence at the interview and maintaining response rates.
Notes for 1.6 Strengths and weaknesses
- Wave One Technical Report
- Family Resources Survey 2013 to 2014
- For more information on non-sampling errors please see Page 200 of the Wave One Report
- The precision of the estimate can be gauged by its confidence interval – the range of values within which the population parameter is estimated to lie. The narrower the interval, the more precise is the estimate
A summary of this report is available in alternative formats such as Large Print, Braille, Audio or Easy Read. The Easy Read version (1.35 Mb Pdf) is available via our website. Large Print and Braille copies are available from the ONS LOS team. The Audio version is available via the DWP website1. Contact the LOS team by email LOS@ons.gov.uk or by telephoning +44 (0)1633 455703.
The LOS Wave Three datasets will be made publicly available in 2015, via the UK Data Service2.