Improving the effectiveness of the Future Accommodation Model to increase private renting behaviour amongst UK Service personnel

This research was completed by a Ministry of Defence (MOD) employee to contribute towards a MSc in Behavioural Change at a UK university. The research was supervised by a Government Social Researcher at the MOD and a lecturer in behavioural change at the host university. The views expressed in this report are the authors and do not necessarily reflect those of the government.

This research was completed in 2021 during the Future Accommodation Model (FAM) Pilot programme. The research findings have been used to inform the development of a new accommodation offer for the UK Armed Forces, which was announced in September 2023.

The MOD considered various options for the design of a new accommodation offer, including making financial savings for Defence. The MOD is no longer seeking to make financial savings or sell existing service accommodation due to this change in accommodation policy. The reader should note this context when reading this report.

List of abbreviations

BCTs Behaviour change techniques 

BCW Behaviour change wheel 

FAM Future Accommodation Model  

HMNB His Majesty’s Naval Base  

MoA Mechanisms of action 

MOD Ministry of Defence  

RAF Royal Air Force  

RN Royal Navy  

RQ Research question

SFA Service Family Accommodation  

SLA Single Living Accommodation  

SP Service personnel  

Abstract

The Ministry of Defence (MOD) provides Service personnel (SP) and their families in the UK Armed Forces with subsidised accommodation to aid workforce mobility, operational readiness, and capability. The MOD is now examining options to give SP more choice and reduce costs by replacing some service accommodation with contributions towards renting in the private sector. The Future Accommodation Model (FAM) pilot offers information and financial support for SP to change behaviour from occupying service accommodation to renting, but change has been limited.

This research used a behaviour change approach to understand the barriers and facilitators to renting for SP. The study aims to propose appropriate improvements to the FAM pilot that are likely to increase renting amongst SP.

The FAM pilot’s active ingredients or behaviour change techniques (BCTs) and hypothesised mechanisms of action (MoAs) were identified through documentary analysis of materials (websites, guidance) produced for the SP. Focus groups with SP were used to identify likely influences on renting. The behaviour change wheel (BCW) process was used to evaluate whether the MoAs were addressing the most important influences and to propose improvements.

The FAM pilot’s BCTs and their hypothesised MoAs had limited relevance to the likely important influences on behaviour (physical opportunity influences linked to administration, cost and availability of rental properties, and reflective motivation influences linked to the characteristics of properties and communities).

Suggested improvements include offering greater practical support, equivalence between financial incentives offered for renting and alternate behaviours, persuasive communication around the positive characteristics of renting and, in extremis, restricting alternate behaviours.

Whilst the effectiveness of the policy as piloted is likely limited, by taking a systematic, theory-based approach to evaluation and modification, this study offers policymakers an improvement route and contributes to the evidence-base of the contexts in which interventions work.

Introduction

Increasing accommodation costs in recent decades have prompted many UK public sector bodies to reduce subsidisation of housing, such as reductions in social housing under the “right to buy” in the 1980s (Stewart, 2005), the removal of police “section housing” in 1994 (Gaskarth, 2016), and more recently, the 2013 social housing “bedroom tax” (Gibb, 2015). The Ministry of Defence (MOD) provides subsidised accommodation to service personnel (MOD, 2020), yet other countries for example New Zealand (New Zealand Defence Force, 2021) have moved away from direct provision of accommodation to subsidising renting in the private sector or removing the subsidy altogether.

The MOD is examining options to give Service personnel (SP) more choice in how they live and reduce costs by releasing the MOD’s more expensive service accommodation and replacing it with contributions towards renting privately. Currently, the most prevalent behaviour of SP is to occupy service accommodation (MOD, 2021a) therefore a behaviour change in SP to occupy rental accommodation is required for the MOD to make savings. Whilst there is no research specifically into the negative or positive influences (the barriers or facilitators) on accommodation behaviour amongst SP, related studies may help identify or explain influences on behaviour in the population of interest.

Overview of existing research

The first time many personnel change accommodation behaviour is transition (the point of leaving the Armed Forces). Although most SP transition successfully, homelessness stemming from poor transition has a significant impact on those affected and costs the UK around £5.5 million per annum (Forces in Mind Trust, 2013). Barriers identified to accessing accommodation at transition include debt, the need for long-term planning, not involving family members and complexity of systems. Facilitators include simplification of processes; and financial, third sector and social network support (Heal and others, 2019; Rolfe, 2020; Fleuty and others, 2021). The MOD’s current approach to facilitating transition is to provide information on housing but some personnel still prepare too little or too late (Scottish Veterans Commissioner, 2021). Support for renting whilst still in service could help address barriers (Rolfe, 2020), whereas providing large, subsidised accommodation is linked to unrealistic expectations that social housing will be automatic, abundant, cheap, and unaffected by geographic rental variations (Rolfe, 2020; Scottish Veterans Commissioner, 2021).

The experience of those leaving local authority care as young adults offers a parallel of changing accommodation behaviour after time in a system where accommodation is provided. Here, restricted employment opportunities, lack of affordable housing, and unsupportive welfare policies can be barriers to establishing and remaining in a home. Facilitators include more choice of accommodation and greater practical support (Biehal & Wade, 1999).

Rental subsidies have been used for populations other than SP, such as housing voucher schemes in the United State of America that encourage movement from less to more affluent neighbourhoods to address deprivation. Despite the financial assistance, barriers remain to changing accommodation behaviour under such schemes. Two-thirds of families intending to relocate under a scheme in Chicago ultimately did not participate, with barriers including availability of accommodation, bureaucratic delay, health problems, lack of experience, and work or personal pressures preventing engagement with the process (Pashup and others, 2005). Amongst those who participated, large numbers returned to less affluent areas. Barriers to sustaining changed behaviour included landlord difficulties, poor-quality accommodation, social isolation, and financial challenges (Boyd and others, 2010), whereas formation of strong social networks was a key facilitator (Boyd, 2008). Across multiple schemes, the availability of relocation counselling, for example non-financial support, whether used or not, was found to increase satisfaction with the new home (Varady and others, 2001).

Amongst SP there may be institutional influences on behaviour that differ from other populations. For example, socio-economic background can predict smoking prevalence on joining the services, but frequent cigarette breaks and access to duty-free cigarettes in early service increase social and physical opportunity for smoking (Bray and others, 2013). Social opportunity has also been found to increase sexual misconduct and harassment (Taber, 2017), and weaken the success of brief alcohol interventions (Doherty and others, 2017) in service populations.

The influences on accommodation drawn from voucher schemes, care leavers and SP at transition may have relevance to the context of accommodation behaviour amongst serving personnel. However, as with smoking, drinking, and sexual misconduct, there may be influences specific to the services creating a clear research gap.

Future Accommodation Model Pilot

To change behaviour and increase renting, the MOD has designed the Future Accommodation Model (FAM) pilot. You can find out more about the pilot by visiting Future Accommodation Model: what you need to know.

The FAM pilot offers rental subsidies and information about renting to personnel assigned to three pilot sites (HMNB Clyde, Aldershot Garrison, and RAF Wittering) ahead of a planned UK-wide rollout of a new accommodation offer to 150,000 personnel. This new rental support is offered alongside existing service accommodation options (service family accommodation and single living accommodation) and a new option providing financial support for home ownership (see Figure 1).

Figure 1: Accommodation options offered under FAM

Image shows four housing options for a SP. Two options are living in service accommodation where charges are deducted from pay. Two options are living in private accommodation and subsidy is received through pay

Uptake by eligible personnel of service accommodation and home ownership is broadly in line with intentions expressed before the pilot (MOD, 2017), but only 3% are renting (unpublished data c/o MOD Accommodation Team, 2021). This is slightly above the 2% renting outside the pilot (MOD, 2021) but below the 10% who expressed the intention (MOD, 2017). Intention to change behaviour is significant in predicting behaviour, but is not the only factor (Sutton, 2001) and FAM was not designed with a behavioural theoretical basis. FAM’s theory of change is that access to information and a subsidy will change behaviour, but this appears not to fully address the intention-implementation gap and therefore components promoting implementation may be lacking or insufficiently effective (Hagger, 2010).

Behavioural change frameworks and tools

The design of complex behaviour change interventions should draw on existing evidence and be underpinned by a theory-based understanding of how the proposed intervention is likely to work (Craig and others, 2008). As well as increasing the chance of success (Michie and others, 2011), a systematic approach to design, combined with clear terminology, benefits the field of behaviour change by adding to the evidence-base of what types of intervention work in what contexts and why (Michie and others, 2009; Bauer and others, 2015). With the aid of artificial intelligence and machine learning, this can become part of future knowledge systems designed to help policymakers design effective interventions (Michie and others, 2017).

The BCW brings together 19 behaviour change frameworks to present a single, systematic, theory-based framework for intervention design (Michie and others, 2014). At the BCW’s core is the COM-B model of behaviour which sets out that capability, opportunity, and motivation (and their sub-domains, defined in Table 1) combine to generate behaviour (see Figure 2; Michie and others, 2011). Of note, capability and opportunity influence the relationship between motivation and behaviour (whereas motivation influences behaviour directly), acting like gates where the more open they are, the easier it is for motivation to generate behaviour. Opportunity or capability can also directly influence motivation since people are more likely to want to perform a behaviour if the environment is supportive and they believe themselves capable (West & Michie, 2020).

Table 1: Definitions of COM-B Domains (adapted from Michie and others, 2014)

Figure 2: The COM-B Model of Behaviour (taken from West & Michie, 2020)

Figure 2 is a diagram showing how the COM-B domains listed in Table 1 relate to each other and influence capability, motivation, opportunity and behaviour.

The BCW links COM-B domains to intervention functions, the broad means through which an intervention acts (for example, education or persuasion) as defined in Table 2. The BCW also links to Behaviour Change Techniques (BCTs), drawn from the BCT Taxonomy v1 (Michie and others, 2013), these are the active components of the intervention that are likely to be more effective in achieving change in the target behaviour (Michie, Atkins & West, 2014). APEASE criteria, as defined in the BCW (see Table 3), are applied to the choices of BCT and intervention function the BCW offers to support selection of those most appropriate to the context.

Table 2: Definitions of Intervention Functions (adapted from Michie and others, 2014)

Table 3: Definitions of APEASE Criteria (adapted from Michie and others, 2014)

Where an intervention already exists, a process evaluation (how the intervention is working) can aid understanding of why it is not achieving its intended effect (Craig and others, 2008). A process evaluation might consider the intervention’s delivery (fidelity and/or dose), its reach, any contextual factors, and/or the mechanisms of action (MoAs). The MoAs state how the intervention brings about the change (Moore and others, 2014; Carey and others, 2019). Understanding MoAs within an intervention can be particularly helpful when policymakers wish to understand how an intervention might work in other contexts (Moore and others, 2015).

As well as being used to design behaviour change interventions, the BCW can be used retrospectively as part of a process evaluation to refine existing interventions (Michie and others, 2014). By providing a shared terminology for intervention components, the 93 BCTs of the BCT taxonomy enable a consistent approach to identifying the active components in the intervention (Michie and others, 2013). The BCW’s linkages to intervention functions and COM-B domains then supports hypothecation of the likely MoAs. If the intervention functions are not targeting relevant COM-B domains (for example, those predicted to influence the target behaviour), the BCW process can be used to suggest theory-based improvements.

Research aims

The need to understand not just what behaviour personnel perform but why they perform it when evaluating FAM was recognised by Walker and others (2020) in an independent report. As a systematic, theory-based tool that can be used to evaluate an existing intervention and suggest improvements (even to interventions not designed using behaviour change theory (Steinmo and others, 2015)), the BCW can underpin the research questions (RQ) which deliver the study’s overall aim – to improve the effectiveness of FAM in increasing renting amongst UK SP.

RQ1: What are the active ingredients and hypothesised MoAs in the FAM intervention?

RQ2: What are the likely influences on UK Service personnel renting private accommodation?

RQ3: How might the FAM intervention be improved to address the identified influences more effectively?

Methodology

Overall design

Qualitative methods were used throughout this research. FAM’s active ingredients and hypothesised MoAs were identified through documentary analysis and retrospective application of the BCW process (RQ1). Focus groups with service personnel then identified potential influences on renting (RQ2). Based on the hypothesised MoAs and influences identified, the BCW was used to assess the likely effectiveness of FAM and suggest improvements (RQ3).

Ethics

The research proposal was approved by the MOD Research Ethics Committee in August 2021 (Ref:2064/MODREC/21).

Methodology: RQ1: What are the active ingredients and hypothesised MoAs in the FAM intervention?

Sample

When eligible personnel are assigned to a FAM pilot site, their Assignment Order (the letter informing personnel of their next role) instructs them to visit a website for information before submitting an ‘accommodation preference form’. The assignment order, website, the videos embedded within that website and a presentation delivered by each site’s FAM Cell (the team that administers the intervention) form the dataset for documentary analysis. Personnel do not have to engage with the information to be part of FAM (they can submit an accommodation preference form regardless), but discussions with the MOD FAM team identified these as the intervention elements most frequently accessed.

Procedure

To identify active ingredients, the dataset was reviewed to identify embedded BCTs using the BCT taxonomy (v1) as a framework. To hypothesise the MoAs through which the active ingredients were likely to work, the BCTs were mapped to the intervention functions they most commonly serve, based on BCW linkages. The intervention functions were mapped to the COM-B domains in which they were likely to be effective in bringing about change, again based on BCW linkages. Where a BCT could be mapped to multiple intervention functions, consideration was given to context and content then judgement applied based on discussion with the MOD FAM team and the author’s knowledge of the subject. No independent reliability checks were conducted. The mapping was summarised in a logic model hypothesising the likely MoAs in FAM.

Methodology: RQ2: What are the likely influences on UK Service personnel renting private accommodation?

Sample and recruitment

The sample was drawn from the 6,000 SP assigned to a FAM pilot site. Personnel under the age of 18 were excluded. The opportunity to participate was advertised for three weeks via posters, chain of command briefings and FAM Cell interactions with personnel. Key characteristics of service (Royal Navy, Army, or Royal Air Force), rank, partnership status, and accommodation type were collected from those expressing an interest. From this, a sample of 36 personnel covering a broad range of the characteristics were invited to attend a focus group. The number of participants was revised down from an initial target of 72 due to poor response rates and the need to assign personnel to both this study and another research project. All consented to participate in this study with no incentives offered.

Achieved sample

Of 36 participants invited to focus groups, 14 attended across five sessions with one conducted as an interview (only one participant). Service and rank group of participants are in Table 4. Partnership status and current accommodation characteristics are summarised to maintain confidentiality. Across the sample, seven participants were married or in a civil partnership, five were in a long-term relationship, one was single with child visitation, and one was single. Five participants lived in private rental, eight in service family accommodation (SFA), none in single living accommodation (SLA) and one in their own home. Although not a prerequisite, all participants were familiar with the FAM intervention.

Table 4: Demographic characteristics of participants

The digit in the participant ID relates to the focus group participated in.

Procedure

Personnel participated in one of five semi-structured focus groups lasting 60-90 minutes. Groups were stratified by rank groups (junior ranks, senior ranks, or Officers) to reduce the impact of hierarchy on responses. Focus groups were conducted remotely using Microsoft Teams during the working day. All participants joined from their workplace (remote working during the COVID-19 pandemic meant this was often their home). The use of video was encouraged to facilitate non-verbal communication, but most participants chose to use only voice. Focus groups were recorded and transcribed automatically using the video-conferencing software. Transcripts were checked against the recording, edited for accuracy and pseudonymised by removing names and identifying references. The transcription approach captured voiced pauses but not non-verbal communication (pauses, body language, etc.). Comments in the chat bar were added to the transcript. Groups were facilitated by the author, supported by a Government Social Researcher from the MOD FAM team.

The focus groups began with an ‘ice-breaker’ information gathering exercise to relax participants and get them thinking about their accommodation behaviour to-date and intentions. Questions were then grouped by COM-B domain, covering participants’ motivations, opportunities, and capabilities to rent, drawing on the example questions set out in the BCW guide (Michie and others, 2014). A sample of questions is in Table 5 and full discussion guide at in the Technical Annex.

Table 5: Sample Focus Group Questions

Subject-matter knowledge enabled the author to take an informed line of questioning and understand some of the underlying, latent meanings in participants’ responses. Use of a discussion guide provided a handrail to focus on influences and avoid getting drawn into debates on the policy itself.

Analyses

Transcripts were first analysed deductively, using a framework analysis approach (Gale and others, 2013) to code data fragments on influences (including whether they were a facilitator or barrier) from the first two transcripts to a COM-B domain, or multiple domains if appropriate. A coding guide (see Technical Annex) was then developed to support consistency in applying the framework to the remaining transcripts. Framework analysis simplifies presentation of large quantities of data which supports effective engagement with policymakers not involved in the analysis itself (Gale and others, 2013), aligning neatly with this study’s aim.

To check coding reliability, a 20% sample from across the rank groups was independently coded for COM-B domains by a social researcher from the MOD FAM team and checked for alignment with the author’s coding. The second coder was practiced in qualitative research, familiar with the subject-matter, and familiarised themself with COM-B ahead of coding. Discrepancies were discussed and related changes made to the coding guide and applied across the dataset.

An inductive thematic analysis approach, following that set out by Braun and Clarke (2006, 2021) was then applied to the whole dataset to generate themes. Each data fragment was given a code summarising the subject covered, then similar data were grouped together to create sub-themes which were subsequently rationalised into overarching themes (promoting the most significant themes and nesting others within them). The overarching themes were next reviewed for quality, boundaries, and a clear central organising concept. Borrowing methodology from framework analysis, these were again defined in a coding guide (see Technical Annex) to check their application across the dataset. Once complete, consideration was given to the overall story the themes told.

The inductive approach was aware of the COM-B domains within which each data fragment sat but was intentionally not constrained to generating themes only within a COM-B domain (instead considering potential themes across the dataset). This mitigated the risk that rigidity of the deductive process constrains the inductive approach and enabled themes to be generated where the same issue could be seen to influence different participants via different routes of the COM-B model. For example, opportunity and motivation influences within one theme might be because a lack of resources (an opportunity barrier) is influencing motivation before behaviour, as well as acting as a gate on the path of influence from motivation to behaviour. Constraining themes to only one domain could miss this relationship and impair clarity of narrative by creating separate but insufficiently distinct themes within opportunity and motivation domains. Such clarity is particularly important when complex analysis must be easily intelligible for policymakers (Braun & Clarke, 2014), as is the aim here. Nonetheless, the awareness of the COM-B domains woven into the analysis mitigated the risk that the lack of framework leads to a simple summary description of the dataset (Braun & Clarke, 2006).

To identify which themes and domains may be most important, two main factors were considered – frequency in the dataset and strength of participants’ expression (measured qualitatively by the author). Where views were mixed as to whether an influence was primarily a barrier or facilitator, the number of participants contributing each perspective was also considered. Strength was given greatest weighting to acknowledge that frequency can be inappropriate when a sample is unrepresentative (Gale and others, 2013) and because the semi-structured approach meant some influences arose primarily in response to facilitator prompting (Atkins and others, 2017). Elaboration (the number of themes or sub-themes in each domain (Graham-Rowe and others, 2018)) was discounted as a measure since some themes presented in multiple COM-B domains.

Methodology: RQ3: How might the FAM intervention be improved to address the identified influences more effectively?

Procedure

The hypothesised MoAs from the logic model generated in RQ1 were assessed against the likely important influences on behaviour identified in RQ2 to understand whether they might be effective in changing behaviour. To identify potential improvements to FAM, the likely important influences from RQ2 were mapped to the intervention functions identified by the BCW as most likely to be effective in changing behaviour. These were then assessed for appropriateness to the context of FAM using the APEASE criteria, based on the author’s knowledge and experience of MOD policy formulation and delivery. BCW linkages were used to map the selected intervention functions to the BCTs most frequently used within them. The BCTs were again assessed for appropriateness using APEASE and suggestions made as to how FAM might incorporate the selected BCTs.

Results

RQ1: What are the active ingredients and hypothesised MoAs in the FAM intervention?

No active ingredients (BCTs) were identified in the assignment order. The website contains multiple instances of BCT 4.1 instruction on how to perform the behaviour and information about the availability of payments to support personnel with the costs of renting, coded as BCT 10.1 material incentive (behaviour). One video embedded within the website featured personnel who are already renting, coded as BCT 6.1 demonstration of the behaviour. The presentation delivered to personnel contain further instances of BCTs 4.1 and 10.1.

BCT 4.1 (instruction on how to perform the behaviour) is frequently used within the training intervention function which the BCW suggests is likely to be effective in bringing about change through the physical capability, psychological capability, physical opportunity and automatic motivation domains. Since the “instruction” in FAM relates to an administrative (rather than physical) process, the MoAs are hypothesised as psychological capability and automatic motivation.

BCT 10.1 (material incentive (behaviour)) is not frequently used in any intervention functions (according to BCW linkages) but has similarities with other BCTs occasionally used in the incentivisation intervention function. Incentivisation is likely to be effective in bringing about change through automatic and reflective motivation domains.

BCT 6.1 (demonstration of the behaviour) is frequently used in the training and modelling intervention functions. Since the video had limited instructional content, the training function is likely to be minor. The BCW suggests that modelling is likely to be effective in bringing about change in social opportunity and automatic motivation domains. However, the video is a relatively small part of the intervention and its overall contribution to the MoAs is likely limited. Hypothesised MoAs are summarised in the logic model at Figure 3.

Figure 3: Logic Model of FAM’s Active Ingredients and Hypothesised MoAs

Figure 3 is a diagram of a logic model showing how intervention components lead to active ingredients, lead to intervention functions lead to hypothesised mechanisms of action.

Arrows and text in bold denote the elements likely to play a more active part in the intervention.

RQ2: What are the likely influences on UK Service personnel renting private accommodation?

Deductive analysis suggested all COM-B domains except physical capability influenced behaviour with physical opportunity and reflective motivation most important. Inductive analysis identified five overarching themes: Availability of affordable rental properties in area, administrative burden of renting, characteristics of rental properties, draw of the service community, and cultural and institutional beliefs. “Affordable availability” and “administration” themes were primarily physical opportunity influences, the “characteristics” and “service community” themes were reflective motivation influences, and the “cultural beliefs” theme was an automatic motivation influence.

Deductive analysis by COM-B domain

Data fragments were found to most frequently capture reflective motivation (135) and physical opportunity (115) influences, with social opportunity (24), automatic motivation (19), and psychological capability (13) less frequently cited influences on behaviour. No data fragments captured physical capability influences. With strength overlaid, the importance of psychological capability and social opportunity diminishes as frequency was driven by questions in the discussion guide. Automatic motivation grows in importance as some participants expressed this influence strongly without prompting.

Reliability checks on deductive coding found 75% alignment with two main areas of discrepancy. First, disagreement over whether some data fragments about rental cost were reflective motivation or physical opportunity influences highlighted that deciding when a lack of resources prevents behaviour directly (unaffordability) and when it decreases motivation (preference to spend money on other things) is subjective and shaped by the coder’s background. Second, disagreement over whether some data fragments about the administrative burden were psychological capability or reflective motivation influences illustrated a similar challenge in understanding whether personnel lacked mental stamina to administer frequent moves or that the administration was demotivating them. Data fragments affected by these disagreements were reconsidered and re-coded until agreement was reached.

Inductive generation of themes

Inductive generation of themes was conducted across the data set, not just within COM-B domains. As such, data fragments relating to each overarching theme can be influences in more than one domain. Of the overarching themes, data fragments most frequently related to availability of affordable rental properties in area (105) followed by administrative burden of renting (87); in both themes these were primarily physical opportunity influences although these often influenced behaviour via reflective motivation. The characteristics of rental properties (55) and draw of the service community (35) themes contained primarily reflective motivation influences. The cultural and institutional beliefs (25) theme contained data fragments coded as both automatic motivation and social opportunity influences, but the latter have low importance as they were generated by a specific question and participants generally confirmed their absence: “I don’t think … [the views of others] would have any difference on how I’d act” (participant 1D).

Table 6 sets out the frequency (but not strength) with which data fragments were found relating to each of the themes and COM-B domains: Figure 4 then sets out the relationship between domains, themes and sub-themes. The text description thereafter only covers the influences and themes considered most important (i.e., after strength is overlaid).

Table 6: Frequency of data fragments by COM-B domain and theme

Figure 4: Influences on Renting by Theme and Sub-Theme

Figure 4 is a diagram showing the themes and sub-themes that were identified in the research.

Physical Opportunity Influences

Availability of affordable rental properties in area.

All but one participant expressed the view that rental properties were often too small or far away to be appealing: “[I’ve] put three-to-seven-bedroom properties available within 15 miles of me … there’s only four that are under £2,000 a month” (1C), or more expensive than the alternative of service accommodation: “SFA was always a cheaper version” (1B); “[compared with renting] we are saving circa a grand a month” (5A). Participants who were renting acknowledged an evaluative process was involved when considering the higher cost of renting: “Obviously you have to factor it into your budget, but we could be flexible” (3B). Eleven participants suggested that cost and/or availability was an outright barrier: “Is there actually any space to move people into private rental?” (1C); “Two to three months’ rent in advance … that’s quite prohibitive” (1B); “Renting would have … left us impoverished” (5B).

Whilst this theme is primarily a physical opportunity influence stemming from a lack of financial resources, the frequency of data fragments relating to reflective motivation and the evaluative processes acknowledged by participants suggest it influenced the relationship between motivation and behaviour (unaffordability preventing participants from renting, however high the motivation), and motivation itself (the additional cost or lack of choice negatively influencing evaluations of renting; see figure 5). Individual finances and variations in rental markets across the three sites likely contributed to which route this theme influenced through.

Administrative burden of renting

Although sometimes raised as a facilitator (by eight participants) there was a higher frequency and strength of barriers in this theme (raised by 11 participants). As a barrier, the most common concerns were related to finding and moving into a rental property: “Moving from private rented, it was far easier to apply for SFA” (5A); “You just cannot deal with the rental market … [especially] when you come back from an [operational] tour” (5C); “I had, you know, 20 minutes [to look at properties online] … but I was never going to be able to travel down Aldershot and start looking” (5B). Online services were the key facilitator: “Searching for a place was, it was [the] easiest thing … you can just do everything online” (4B). The strength of the influence was also affected by a participant’s mobility (how frequently they are assigned) with the more mobile perceiving a greater barrier: “When you’re moving anything between one year or three years … [and only have] two or three months to try and do all that administration … Rental, I just can’t see working” (5C).  

Whilst this theme is primarily a physical opportunity influence stemming from lack of time and/or means to visit a distant location, some participants did have opportunity but still evaluated that renting was not worth the hassle in comparison with occupying service accommodation. As such, this theme influenced the relationship between motivation and behaviour (administration unachievable thus closing the gate to renting), and motivation itself (the additional administration negatively influencing evaluations of renting; see figure 5). The demands of participants’ jobs and physical distance from the next assignment likely affected which route the theme influenced through, with high mobility amplifying barriers.

Figure 5: Route of physical opportunity influences on renting

Figure 5 is a diagram illustrating the link between the availability of rental properties in an area and the administrative burden of renting on the physical opportunity of renting.

Reflective motivation influences

Characteristics of rental properties.

All participants expressed finding some characteristics appealing, suggesting an evaluative process and reflective motivation facilitators. Some were motivated by physical characteristics: “The rented comes with all the appliances” (4A); “I wanted an area that I could have my own office” (4B) or distance from amenities “That’s why I picked it, because it’s in the Town Centre I can walk to everything” (1A). For others, renting provided stability for their family as they could live in a central location between likely future assignments: “My son, who is [age] has lived in seven SFA … I’ve lost count how many jobs my wife has had so it was all about continuity and stability at this point” (2A). In summary, it was the choice the rental market offers, and thus the ability to meet individual priorities, that led to positive evaluations: “We did want to privately rent … to choose the size house and have a garden and a garage” (4A); “You have way more choice … you can almost hand pick” (2A).

The service community theme could be considered a sub-theme of characteristics of rental properties (rental properties characteristically lack a service community), but it was defined separately to reflect the strength and divergence of opinion. Though closely related to social opportunity, most participants consciously evaluated the service community’s attractiveness to them, thus a reflective motivation influence. All single participants saw getting away from the service community as a facilitator: “[Renting] was the quickest and easiest option of getting off [base]” (3A); “The further away [from work] the better really” (4A), as did five participants with families: “After a while I used to prefer to be away from the patch life … everybody wants to be in everybody’s business and nothing is private” (1B). Five participants with families also saw distance from the service community as a barrier: “From a partner’s perspective to be within that vicinity of other like-minded people, it is useful” (3B); “In the military community you’ve got an automatic neighbourhood watch” (5C). As with the administration theme, mobility was a key variable with the more mobile most likely to value the service community.

Automatic motivation influences

Cultural and institutional beliefs

This theme was most prevalent in officer groups who expressed concerns that renting would degrade a unit’s operational role and welfare: “You disperse people, then the welfare bit starts to fall away … culture and ethos does change” (5C), and undermine the deal (the moral deal between the government and SP): “What the government had calculated was the private rental going rate … was frankly insulting” (5B); “The really basic unwritten rule is you expect to be at the whim and behest of the military because the admin will be supported” (5C). In these instances, participants did not evaluate renting for themselves, but exhibited an emotional reaction setting them against a concept they instinctively perceived as a degradation of the deal. This suggests an automatic motivation barrier which was strongest amongst the most mobile. Conversely, one senior rank, whilst not evaluating renting positively for themself, supported the concept due to a concern that “protecting” personnel from the real-world of renting harmed longer-term welfare, suggesting an automatic motivation facilitator: “So many people struggle with accommodation once they finish the military” (1C).

RQ3: How might the FAM intervention be improved to address the identified influences more effectively?

For most participants, physical opportunity and reflective motivation influences were most significant. For officers, automatic motivation was also important but, since this cohort is relatively small, this was omitted from the BCW process below.

Does the current intervention address the influences?

As set out in the logic model at Figure 3, BCT 4.1 (instruction on how to perform the behaviour) within FAM primarily acts through psychological capability but this was not an important influence amongst participants thus its effectiveness is questionable. BCT 6.1 (demonstration of the behaviour) similarly fails to address the most important influences on behaviour identified through the focus groups (physical opportunity and reflective motivation), even if it were a more prominent part of the intervention. Based on the BCW linkages, BCT 10.1 (material incentive (behaviour)) may be effective in addressing reflective motivation influences but none of FAM’s hypothesised MoAs address the physical opportunity influences.

Intervention functions likely to be effective

The intervention functions that serve physical opportunity and reflective motivation influences were assessed using APEASE criteria (outcome summarised in Table 7). Training and education were dismissed as unlikely to be effective since a knowledge deficit was not apparent amongst participants. Environmental restructuring is impracticable since the behaviour is performed in spaces outside of the MOD’s control, and coercion is unlikely to be acceptable to service personnel. Although restriction raises equitability and related acceptability concerns, it would likely be practicable and effective so was taken on for further consideration alongside the other remaining relevant functions of enablement, persuasion and incentivisation, as shown in Figure 6.

Table 7: APEASE assessment of intervention functions serving relevant COM-B domains

Figure 6: Appropriate intervention functions and BCTs to influence renting

Figure 6 is a diagram showing the intervention functions that could be used to influence renting. Arrows and text in bold denote the elements likely to be more effective and/or appropriate for FAM.

Restriction

Whilst restriction does not have any BCTs specified in the BCW (BCTs focus on changing feelings, thoughts and reactions rather than imposing external limitations (Michie, Atkins & West, 2014)), restrictions on service accommodation did act as a significant facilitator for some participants who could only live with whom or where they wanted by renting: “[I] wasn’t gonna [sic] rush my marriage to get [SFA]” (4A). “I could not retain [SFA] which was in the same mileage threshold” (2A). Imposing further restrictions on service accommodation is practicable and could be effective in increasing renting but is contrary to the MOD’s aim to widen choice in accommodation. As such, it is unlikely to be acceptable (particularly if perceived inequity between the service and rental accommodation offers remains).

Enablement

Many BCTs can be considered appropriate for enablement, and it can address influences in most COM-B domains. In the context of renting, BCT 3.2 social support (practical) is perhaps most likely to be effective and meets the other APEASE criteria. The administrative burden of renting was a significant barrier for many participants therefore practical support should help facilitate renting. By reducing the time required for administration, this BCT addresses physical opportunity barriers to renting which can open the gate to existing motivation blocked by lack of resources, and/or increase motivation where the requirement to invest time is negatively influencing an evaluation of renting.

Persuasion

FAM was designed not to persuade personnel to rent due to concerns this could be perceived as the MOD “pushing” personnel out of service accommodation. Since the MOD needs more personnel to rent voluntarily to avoid restricting access to service accommodation, using BCTs in support of the persuasion intervention function would seem sensible. Giving information about the positive emotional consequences of renting (BCT 5.6), linked to the reflective motivation facilitators identified (e.g., the beneficial characteristics found in the rental market), could be effective. A potential obstacle is the distrust personnel expressed in messages from “MOD Head Office”, therefore delivering the message through a credible source (BCT 9.1) will be important.

Incentivisation

The most common BCTs specified for incentivisation are related to ongoing monitoring or feedback which is unlikely to be effective in the context of a behaviour that only tends to be reassessed every few years (at the point of assignment). Though less frequently used, incentivisation also covers reward BCTs and, in this context, a financial incentive delivered through the intervention may be effective. BCT 10.1 (material incentive (behaviour)) is already an active ingredient in FAM but some participants felt the payment’s value was insufficient: “If you’re really intent on getting people to rent, then you gotta [sic] give them more money.” (5B). Whilst the BCW only links incentivisation to motivation, in a similar way to social support described above, greater financial support should also act on physical opportunity with the additional resources opening the gate where unaffordability is blocking existing motivation, and/or increasing motivation where the requirement to spend more money is negatively influencing evaluations of renting.

As noted in the APEASE assessment, a significant increase in the incentive may be unaffordable for the MOD. If so, a reduced incentive for occupying service accommodation (as an alternate behaviour) may have a similar effect in influencing evaluations but will not address physical opportunity barriers to renting and may newly create them in service accommodation.

Timing of Intervention

Although one participant did consider changing behaviour mid-assignment, none did: “Once we were in and we were unpacked … do we really want to go through this again” (5E). Nine participants also highlighted that awareness and understanding of FAM was limited prior to making an accommodation choice: “take up hasn’t really taken off because people don’t know enough about it” (1B). Deploying the above BCTs before accommodation intentions are implemented for the next assignment is therefore key in providing opportunity to change behaviour.

Discussion

This study set out to identify the active ingredients (BCTs) and hypothesise the MoAs in the FAM intervention; identify the influences on SP renting; and propose improvements to FAM. The BCTs and MoAs identified were found to go some way in addressing identified reflective motivation influences on behaviour, but do not address the physical opportunity barriers preventing some personnel from renting. FAM might be improved by enhancing the material incentive and offering practical social support for renting, backed up by persuasive information about the benefits of renting delivered by a credible source. If these measures are insufficient to change behaviour (enhancement to the material incentive is likely to be constrained by affordability for the MOD), restricting access to service accommodation would likely change behaviour, but with the acceptability and equitability challenges that such a blunt tool brings.

Integration with Literature

Since there is no literature in the same field as this study, no direct comparison can be made. The studies identified in the introduction do, however, offer some reinforcement of this study’s findings.

Studies into moving from care to rented accommodation (Biehal & Wade, 1999) and when moving from less to more affluent neighbourhoods using voucher schemes (Varady and others, 2001) found, as this study also suggests, that practical support could be effective in promoting and maintaining a change in accommodation behaviour. Significantly, this has also found to be an influence on accommodation behaviour amongst service personnel at transition (Fleuty and others, 2021).

The importance of practical support in the contexts above adds weight to the suggestion that the MOD should consider increasing the practical rental support offered to personnel. However, studies considering accommodation behaviour at transition have noted the risk of longer-term harm if too much support prevents exposure to the “real world” of renting that personnel encounter on transition (Rolfe, 2020). This tension arose in this study through the differing automatic motivation influences between officers and senior ranks who were concerned about a negative impact of accommodation administration on delivery of a unit’s role and a positive impact on individuals’ preparedness for transition respectively.

Since it was availability of support (rather than uptake) that was found to be effective across voucher schemes (Varady and others, 2001), it may be that the offer of support is sufficient for reflective motivation facilitators linked to the positive characteristics of renting to overcome barriers linked to the administrative burden, even if that support is ultimately not used. It may also go some way to addressing automatic motivation barriers amongst officers where a lack of support is seen to undermine the moral deal. Where the administration is a physical opportunity barrier then such support may have to be used for behaviour to change.

Availability of affordable accommodation was not cited as an influence in the care-leaver or voucher scheme studies but was referenced in the context of transition from service. The suggestion that some personnel do not start financial preparations sufficiently early in service (Rolfe, 2020; Scottish Veterans Commissioner, 2021) adds weight to the suggestion in this study that there is a cost barrier. However, the same tension as with practical support is found here where excessive subsidising of accommodation whilst in service contributes to accommodation difficulties for some personnel at transition (Rolfe, 2020). Reducing the subsidy for service accommodation (rather than increasing the incentive for renting) may change behaviour amongst some personnel whilst also addressing the transition issues identified.

Finally, research into voucher schemes suggests that developing social ties with the community families move into can play a role in supporting success (Boyd, 2008; Boyd and others, 2010). Social opportunity was not found to be a significant influence in this study, but it is unclear as to whether, like with smoking, alcohol and sexual misconduct behaviours, this is due to cultural difference between service personnel and civilians, or because the current low prevalence of renting in the armed forces means there was limited experience of social ties with civilian communities for participants to draw on.

Implications

As set out in the results and integration with the literature above, the MOD should consider amending the FAM intervention to offer greater practical support with renting accommodation. If affordable, greater financial support should also be offered (or, alternatively, current financial support for service accommodation reduced). Even so, there will likely still be personnel with poor credit ratings or relocating from overseas who will be unable to secure a rental property for themselves, and personnel will still be assigned to remote locations where there simply is no rental market. In these instances, behaviour change may not be achievable and thus some service accommodation will need to be retained.

Across many of the themes, this study identified mobility as a key variable that affects the strength and direction of influences on renting, summarised in Figure 7. Reducing service mobility is beyond the scope of the FAM programme but, if practicable for the MOD, this would go some way in addressing many of the reflective motivation barriers, and potentially physical opportunity barriers around the administrative burden.

Notwithstanding the limitations on the scope of the intervention, the improvements identified above should reduce physical opportunity barriers around administration and cost, whilst improved communication of the beneficial characteristics of renting should strengthen this reflective motivation facilitator. If, however, these measures are insufficient to change behaviour at the scale and in the time required, restrictions on service accommodation may need to be considered.

Figure 7: How mobility affects strength and direction of influences

Figure 7 is a diagram illustrating how the increased mobility of a SP affects renting.

Looking to the field of behaviour change, this study provides an example of how the BCW process can be used retrospectively to identify MoAs in an existing intervention and propose theory-based improvements. It also shows how the BCW process can be applied to fields outside its more common use for health and social behaviours. The study further strengthens the linkages set out in the COM-B model (Figure 2) as a lack of time or financial resources (physical opportunity) influenced evaluations of the behaviour (reflective motivation) as well as sometimes acting as a closed gate blocking existing motivation. If implemented, the suggested changes to the intervention offer an opportunity to evaluate the extent to which BCTs deployed to address physical opportunity barriers can also influence reflective motivation.

Areas for Further Research

If increased practical support and financial incentives are the proposed adaptations for the intervention, policymakers’ next question may be “how much?”. Research into the sensitivity of behaviour to increased support would be valuable and could be tested through the existing pilots to provide revealed rather than predicted behaviour (the limitations of intention in predicting renting behaviour already being well-demonstrated by the pilots). Since the overall intended outcome is not necessarily to increase renting but rather to decrease service accommodation demand, other behaviours within this system should be considered and research into influences on homeownership (as the other main alternative to service accommodation) would be valuable.

To focus influences relevant to the largest population, this study did not propose intervention modification to address the automatic motivation influences on officers within the cultural and institutional beliefs theme. However, further research with a larger, tri-service, officer sample could help tailor the intervention to address these influences more effectively. Some participants also highlighted their partner’s role in accommodation (also highlighted by research into influences on transition (Heal and others, 2019)) therefore understanding the influences on renting amongst family members could also be valuable.

Finally, if the suggested amendments to FAM are adopted, a process evaluation should be conducted to understand how the revised intervention is working, and an outcome evaluation used to assess its effectiveness. The process evaluation will be made easier by this study’s use of a systematic approach to design the modifications, with the intended active ingredients and MoAs clearly identified, allowing the evaluation to pay greater attention to delivery, reach and context.

Conclusion

The FAM policy as piloted is unlikely to be particularly effective in increasing renting behaviour amongst UK SP since it does not address the main influences. The amendments proposed above should be incorporated into ongoing policy development and tested (outcome and process evaluations). FAM’s poor effectiveness likely stems from the lack of a systematic, theory-based design. The BCW-based approach used by this study to propose improvements will enable a more straightforward process evaluation of the revised intervention and the clear description of intervention components will allow it to contribute to the growing evidence base of what behaviour change interventions work in what context.

References

Atkins L, Francis J, Islam R, O’Connor D, Patey A, Ivers N, Foy R, Duncan EM, Colquhon H, Grimshaw JM, Lawton R and Michie S (2017), ‘A guide to using the Theoretical Domains Framework of behaviour change to investigate implementation problems’, Implementation Science, Volume 12, Issue 1, pages 1 to 18

Bauer MS, Damschroder L, Hagedorn H, Smith J and Kilbourne AM (2015), ‘An introduction to implementation science for the non-specialist’, BMC Psychology, Volume 3, Issue 1, pages 1 to 12

Biehal N and Wade J (1999), ‘The early housing careers of young people leaving care.’ In J Rugg (Editor), ‘Young people, housing and social policy’, Routledge, pages 79 to 92

Boyd ML (2008), ‘The role of social networks in making housing choices: The experience of the Gautreaux Two Residential Mobility Program’, Cityscape, pages 41 to 63

Boyd ML, Edin K, Clampet-Lundquist S, and Duncan GJ (2010), ‘The durability of gains from the Gautreaux Two residential mobility program: a qualitative analysis of who stays and who moves from low-poverty neighborhoods’, Housing Policy Debate, Volume 20, Issue 1, pages 119 to 146

Braun V and Clarke V (2006) ‘Using thematic analysis in psychology. Qualitative research in psychology’, Volume 3, Issue 2, pages 77 to 101

Braun V and Clarke V (2014), ‘What can “thematic analysis” offer health and wellbeing researchers?’, International Journal of Qualitative Studies on Health and Well-being, Volume 9, Issue 1

Braun V and Clarke V (2021), ‘Thematic analysis: A practical guide’, Sage

Bray I, Richardson P and Harrison K (2013), ‘Smoking prevalence amongst UK Armed Forces recruits: changes in behaviour after 3 years follow-up and factors affecting smoking behaviour’, BMJ Military Health, Volume 159, Issue 1, pages 44 to 50

Carey RN, Connell LE, Johnston M, Rothman AJ, De Bruin M, Kelly MP and Michie S (2019), ‘Behavior change techniques and their mechanisms of action: a synthesis of links described in published intervention literature’, Annals of Behavioral Medicine, Volume 53, Issue 8, pages 693 to 707

Craig P, Dieppe P, Macintyre S, Michie S, Nazareth I and Petticrew M (2008), ‘Developing and evaluating complex interventions: the new Medical Research Council guidance’, BMJ, page 337

Doherty AM, Mason C, Fear NT, Rona R, Greenberg N and Goodwin L (2017), ‘Are brief alcohol interventions targeting alcohol use efficacious in military and veteran populations? A meta-analysis’, Drug and alcohol dependence, Volume 178, pages 571 to 578

Fleuty K, Cooper A and Almond M (2021), ‘Armed forces and veteran housing policies: the United Kingdom 2021 vision’, Journal of Veterans Studies, Volume 7 Issue 1, pages 232 to 240

Forces in Mind Trust (2013), ‘The transition mapping study: Understanding the transition process for service personnel returning to civilian life’

Gale NK, Heath G, Cameron E, Rashid S and Redwood S (2013), ‘Using the framework method for the analysis of qualitative data in multi-disciplinary health research’, BMC medical research methodology, Volume 13, Issue 1, pages 1 to 8

Gaskarth G (2016), ‘Commuter Cops: Helping our police to live in the city they serve’, Policy Exchange

Gibb K (2015), ‘The multiple policy failures of the UK bedroom tax’, International Journal of Housing Policy, Volume 15, Issue 2, pages 148 to 166

Graham‐Rowe E, Lorencatto F, Lawrenson JG, Burr JM, Grimshaw JM, Ivers NM, Presseau J, Vale L, Peto C, Bunce J and Francis J (2018), ‘Barriers to and enablers of diabetic retinopathy screening attendance: a systematic review of published and grey literature’, Diabetic Medicine, Volume 35, Issue 10, pages 1308 to 1319

Hagger MS (2010), ‘Self-regulation: an important construct in health psychology research and practice’, Health Psychology Review, Volume 4, Issue 2, pages 57 to 65

Heal J, Crouch L, Halkiopoulos S, Fussey V and Kirkman E (2019), ‘Applying behavioural insights to successful transition’, Forces in Mind Trust

Michie S, Fixsen D, Grimshaw JM and Eccles MP (2009), ‘Specifying and reporting complex behaviour change interventions: the need for a scientific method’, Implementation science, Volume 4, Issue 1, pages 1 to 6

Michie S, Van Stralen MM and West R (2011), ‘The behaviour change wheel: a new method for characterising and designing behaviour change interventions’, Implementation science, Volume 6, Issue 1, page 42

Michie S, Richardson M, Johnston M, Abraham C, Francis J, Hardeman W and Wood CE (2013), ‘The behavior change technique taxonomy (v1) of 93 hierarchically clustered techniques: building an international consensus for the reporting of behavior change interventions’, Annals of behavioral medicine, Volume 46, Issue 1, pages 81 to 95

Michie S, Atkins L and West R (2014) ‘The behaviour change wheel: a guide to designing interventions’, Silverback Publishing

Michie S, Thomas J, Johnston M, Mac Aonghusa P, Shawe-Taylor J, Kelly MP and & West R (2017) ‘The Human Behaviour-Change Project: harnessing the power of artificial intelligence and machine learning for evidence synthesis and interpretation’, Implementation Science, Volume 12, Issue 1, pages 1 to 12 

Ministry of Defence (2017), ‘Future Accommodation Model Survey 2016: Main Report’

Ministry of Defence (2020) ‘JSP 464 Tri-Service Accommodation Regulations Volume 1, Part 1’

Ministry of Defence (2021a) ‘Annex B to AFCAS Main Report – Reference Tables (Excel), Armed Forces Continuous Attitudes Survey 2021’

Moore G, Audrey S, Barker M, Bond L, Bonell C, Hardeman W and Baird J (2014), ‘Process evaluation of complex interventions’, UK medical Research Council (MRC) guidance, pages 1 to 133

Moore GF, Audrey S, Barker M, Bond L, Bonell C, Hardeman W and Baird J (2015), ‘Process evaluation of complex interventions: Medical Research Council guidance’, BMJ, 350

New Zealand Defence Force (2021), ‘Life in Uniform: Accommodation and Food’

Pashup J, Edin K, Duncan GJ and Burke K (2005), ‘Participation in a residential mobility program from the client’s perspective: Findings from Gautreaux Two’, Housing Policy Debate, Volume 16, Issue 3 to 4, pages 361 to 392

Rolfe S (2020) ‘Working Together to Meet the Housing Needs of Ex-Service Personnel: Examining the Challenges of Transition and Collaboration,’ Forces in Mind Trust. 

Scottish Veterans Commissioner (2021), ‘Positive Futures, Getting Transition Right in Scotland’

Steinmo S, Fuller C, Stone SP and Michie S (2015) ‘Characterising an implementation intervention in terms of behaviour change techniques and theory: the ‘Sepsis Six’ clinical care bundle’, Implementation Science, Volume 10, Issue 1, pages 1 to 9  

Stewart J (2005), ‘A review of UK housing policy: ideology and public health’, Public Health, Volume 119, Issue 6, pages 525 to 534

Sutton S (2001), ‘Health behavior: Psychosocial theories. International encyclopaedia of the social and behavioral sciences’, pages 6499 to 6506

Taber N (2017), ‘The Canadian armed forces: Battling between operation HONOUR and operation Hop on Her’, Critical Military Studies, Volume 6, Issue 1, pages 19 to 40

Varady DP, Walker CC and Wang X (2001), ‘Voucher recipient achievement of improved housing conditions in the US: Do moving distance and relocation services matter?’ Urban Studies, Volume 38, Issue 8, pages 1273 to 1304

Walker J, Selous A and Misca G. (2020), ‘Living in our shoes: understanding the needs of UK armed forces families’, Report of a review commissioned by the Ministry of Defence

West R and Michie S (2020) ‘A brief introduction to the COM-B Model of behaviour and the PRIME Theory of motivation’