An ‘undeliberate determinacy’? The changing migration strategies of Polish migrants in the UK in times of Brexit

ABSTRACT This paper reformulates classical questions regarding the plans and strategies of Polish migrants in the UK – such as decisions to leave or remain in the host country, or be ‘deliberately indeterminate’ about future plans – from a sociologically situated ‘rights-based’ perspective. This approach considers migrants’ attitudes towards specific ‘civic integration’ measures in a medium-term time frame, as well as in the new context created by the UK’s vote to leave the EU. Based on the quantitative analysis of original survey data, we investigate the factors behind Polish migrants’ migration strategies, and we argue that basic socio-economic and demographic factors are inadequate, on their own terms, to explain future migration and civic integration plans. Instead, we find that aspects such as interest in and awareness of one’s rights, as well as anxieties about the ability to maintain one’s rights in the future are stronger determinants.


Introduction
The UK's decision to leave the EUor 'Brexit', as it became colloquially knownraises various theoretical and empirical questions for migration scholars studying what less than a decade ago was still an emerging 'new European migration system' (Favell 2008), in which the 'free movement of people' served as the main driving force behind 'horizontal Europeanisation' (Mau and Verwiebe 2014). In the UK, the largest single national group among the new arrivals were Polish citizens (Burrell 2009;White 2011), and early research into this migratory phenomenon has identified transience, fluidity and contingency as some of its main characteristics (Eade, Drinkwater, and Garapich 2007). At the same time, sustained research on Polish migrants has increasingly emphasised emergent desires of achieving a 'normal' life, alongside strengthening reasons for and signs of more permanent settlement (Burrell 2009). The UK's departure from the EU and the inevitable change in the legal status of resident Polish nationalswhatever particular shape that will takewill unavoidably impact on the two contrasting trends of 'transience' and 'settlement'.
In this paper, we provide an early assessment of how 'Brexit' affects Polish migrants in the UK, focusing specifically on the question of migration plans and strategies, which has been a core theme in the research literature from the very early post-Accession studies to the present (Burrell 2010;Drinkwater and Garapich 2015;Eade, Drinkwater, and Garapich 2007). As Burrell (2010, 299) noted, Polish migrants have long constituted 'a diverse, not entirely predictable, population, all existing within the same economic framework but formulating different strategies of migration and return' which will most likely 'change over time'. The Brexit vote has introduced an additional element of uncertainty to the complexity of Polish migrants' active and passive plans and strategies, and the aim of this paper is primarily to further explore such strategies in the context of Brexit. To this end, we present results from a quantitative analysis of data obtained from a purposefully designed and 'targeted' online survey carried out in the months leading up to the EU Referendum. The survey had the broader goal of assessing different EU migrants' opinions on the Referendum, their future plans and coping strategies in the event of a potential Brexit vote, as well as their attitudes towards British citizenship. The focus on the latter was driven by the already noticeable surge in applications for British citizenship by EU nationals, stirred by fears regarding the UK's uncertain future EU membership (McGhee and Piętka-Nykaza 2016;Ryan 2015). In this paper, we focus on a sample of 894 Polish respondents. This allows us to draw closer parallels with the existing literature, and provides more reliable grounds for extrapolation to a Polish migrant community which is currently the largest non-British national group in the UK, estimated at 916,000 individuals, and making up 29% of all EU nationals living in the country (ONS 2016).
This paper, therefore, makes a contribution to the literature on migrants' intentions and strategies, while breaking new ground in our understanding of the effects of Brexit as perceived by the UK's Polish migrant community. First, we examine more closely the literature on migration intentions, with a particular focus on the tendency to keep one's plans deliberately indeterminate, which has been repeatedly highlighted as one of the most distinguishing characteristics of mobility throughout the first decade of post-Enlargement Polish migration (Eade, Drinkwater, and Garapich 2007;McGhee, Travena, and Heath 2015). We then discuss our data and methods before moving on to our analysis and findings, and contextualising the latter in the existing scholarship.

Migration strategies and the habitus of 'indeterminacy'
Return, settlement and 'deliberate indeterminacy' A key theme in the literature on post-Accession migration to the UK is concerned with the new migrants' intended length of stay and their plans with regard to either returning to their 'home' country or settlement in the country of destination (Burrell 2010). Notwithstanding this interest, the study of return migration has remained somewhat 'problematic' (Parutis 2014, 159). The discrepancy between planned and actual return has become a truism in migration research, and it is also true of Polish migrants (Anwar 1979;Ryan 2015). Nevertheless, qualitative studies were successful in identifying some of the main factors which have an influence on migrants' strategies. The role of familial obligations with regard to children, grand-children, elderly parents and other relatives in the host and home countries (McGhee, Heath, and Trevena 2013;White 2011;White and Ryan 2008), of childbearing decisions (Janta 2013), the length of stay in the destination country (Ryan 2015) and the level of social integration and transnationalism (de Haas and Fokkema 2011) in particular have been shown to influence migrants' decisions to settle or re-migrate.
At the same time, the increasing ease of travel and communication, complemented in Europe by the policy regime of free movement and equal treatment for EU citizens, allows for more opportunities not only for travelling but also for 'living transnationally', and this necessarily engenders novel mobility habita (Heath, McGhee, and Trevena 2011;Levitt and Schiller 2004;Ryan 2011). One such 'mobility habitus' highlighted in the literature on post-Enlargement Polish migration to the UK was the 'intentional unpredictability' of many recent migrants with regard to their future plans, their tendency to 'keep their options deliberately open' (Eade, Drinkwater, and Garapich 2007, 11). Alternatively referring to it as 'intentional unpredictability', 'liquid migration', 'lasting temporariness' or 'deliberate indeterminacy', authors describing this disposition were highlighting two putative empirical characteristics of this mobilityits 'temporary' and 'unplanned' characterwith a stronger emphasis on one or the other (Eade, Drinkwater, and Garapich 2007;McGhee, Travena, and Heath 2015;Snel, Faber, and Engbersen 2015).
In this paper, we give preference to the term 'deliberate indeterminacy' for several reasons. Firstly, the more commonly used notion of 'intentional unpredictability' seems to reflect, semantically at least, the perspective of a politics of measurementby emphasising the difficulty to predict migratory movementsrather than that of the migrants themselves. Secondly, we wished to de-emphasise 'temporariness' and 'fluidity'; Polish migrants, qua EU citizens, may remain undetermined about plans for the future without necessarily becoming 'temporary' residents, given that they automatically acquire permanent resident status after five years of exercising free movement rights. Also, 'indeterminacy' can arguably be construed as supportive of the oft-described quest for a 'normal life' among those who have been resident for longer and have begun establishing families, as evading the necessity to formulate clear return or settlement plans can enhance the sense of 'normalcy' by essentially normalising transnational forms of belonging and activity across the lifecourse (Galasińska and Kozłowska 2009;Lopez Rodriguez 2010;McGhee, Heath, and Trevena 2012;Morokvasic 2004).

Operationalising indeterminacy
A few studies have recently attempted to disentangle the relationships between migration strategies and migrants' indeterminacy habitus based on primary quantitative data. Snel, Faber, and Engbersen (2015), for instance, have examined the factors determining the selfdeclared planned length of stay of three EU national groups (Polish, Romanian and Bulgarian) in the Netherlands using survey data (N = 654). The authors treat 'do not know' answers in respect to planned length of stay as a manifestation of 'intentional unpredictability'. Their findings show that involvement in transnational activities significantly reduced the chances of migrants wanting to settle permanently, while age and the level of socio-cultural integration significantly increased their intention to settle. The influence of gender, education and socio-economic status produced mixed results, while labour market participation seemed not to have any influence on migrants' plans. Drinkwater and Garapich (2015) adopt a similar study design to investigate the migration strategies of Poles in the UK (N = 700). Their specific aim was to develop a new typology of Polish migrants in light of their planned stay in the UK, factoring in changes in their intentions over time. Similar to Snel, Faber, and Engbersen (2015), the authors also treated those who did not know for how long they were planning to stay, as exponents of the habitus of deliberate indeterminacy. Overall, of the respondents in their sample, 11% intended to stay in the UK permanently upon arrival, and 18% felt the same at the time of the interview, while 32% and 44% were indeterminate at the time of arrival and at the time of the interview, respectively. Their analysis shows that the length of stay in the UK determines whether migrants change their intentions.
These two studies are significant in that they attempt to operationalise the concept of deliberate indeterminacy, and examine it in relation to other strategies and changes in plans over time. However, they remain constrained by self-declared migration intentions at a specific point in time, and the different contributions of each paper are yet to be integrated within a unifying framework. As a consequence, our knowledge of how migration plans change over time and what factors determine this change remains to be synthesised with the rather inconclusive findings regarding the determinants of particular plans and strategies at given points in time. Furthermore, there is a lack of consideration in the broader literature of the relationship between migrants' self-declared preferences and the constellation of rights and options available to them. At the same time, we now must consider 'Brexit' as a new factor which potentially affects migrants' plans and strategies. Based on sporadic evidence in this respect in the recent literature, we can assume that the outcome of the referendum on EU membership is likely to disrupt Polish migrants' reported habitus of 'deliberate indeterminacy' (McGhee and Piętka-Nykaza 2016;Ryan 2015).

A rights-based approach to migration strategies
In response to the gaps in current knowledge identified in the literature review, our aim in this paper is to reframe questions regarding migration strategies and their determinants within a sociologically situated civic-rights-based approach, while also addressing the latest political development that potentially affects the lives and future plans of Polish migrants in the UK. Instead of focusing purely on migrants' self-declared plans and strategies, we place these strategies in relation to specific actions toward 'civic integration' (such as applying for British citizenship or a document certifying permanent residence), while leaving open the possibilities for interpretation of the motivations behind such actions. We acknowledge, accordingly, that in the case of EU migrants 'civic integration' must not necessarily equate 'settlement', just as 'indeterminacy' does not necessarily reflect 'temporariness'. Under the conditions of EU citizenship, the difference between opting for a permanent residence certificate and espousing a habitus of 'indeterminacy' may come down to specific personal circumstances or individual dispositions such as one's subjective sense of insecurity and confidence in relying on 'supranationally derived' rights (Wiener 2013). We therefore consider both the constellation of civic rights legally enjoyed by Polish EU nationals in the UK, and their perceptions of them.
In terms of legal rights and civic integration options, EU citizens find themselves in an ambivalent position. On the one hand, the requirement to 'integrate civically'as 'civic integration' is commonly defined in the literature (Goodman 2012, 659;Joppke 2007;Mouritsen 2013)does not obtain in respect to securing a permanent residence certificate, and EU nationals cannot be charged more than what British nationals would be for comparable identity documents. As noted earlier, EU migrants acquire permanent residence status automatically after five years of qualified residence, and obtaining a document certifying this does not provide any additional benefits, although it can help prove one's status in certain circumstances. 1 As such, it cannot be readily interpreted as 'settlement', as it may merely reflect a desire to secure tangible proof of supranational rights. On the other hand, as of November 2015, holding a permanent residence certificate has been a requirement before applying for British citizenship, 2 and EU citizens face similar 'civic integration' and financial requirements to Third Country Nationals when applying for naturalisation.
Within this broad legal-normative framework, Polish migrants can make 'civic integration' decisions based on their subjective assessment of the rights they possess, and the EU referendum provides an additional context for such decision-making by potentially increasing the sense of uncertainty regarding supranationally derived rights. Taking into consideration the above, our aim here is therefore to find out what factors contribute towards determining the migration and 'civic integration' strategies of Polish migrants in the UK (1) over a normatively relevant medium-term time frame and (2) under the conditions created by the UK's vote to leave the EU. Our approach, therefore, acknowledges that long-term plans are changeable and hard to determine (Burrell 2010;Parutis 2014), and also that 'indeterminacy' and 'civic integration' can both be manifestations of different forms of transnational living (McGhee, Heath, and Trevena 2012;Morokvasic 2004;Ryan 2011;White and Ryan 2008).

Study design and data structure
We analyse data originating from a purposefully designed 'targeted' online survey (Polishonly sample N = 894). The purpose of the survey was to assess the attitudes of EU migrants towards the Referendum on EU membership, its effect on their lives and their strategies in case of a vote for Brexit, and thus contribute to the public debate on the issue. The online survey data collection method was chosen with the aim to obtain a geographically dispersed large sample under strict time constraints (see Sue and Ritter 2012, 10-11). In order to achieve greater representativeness among the respondents, the survey was translated into Polish. Data collection was undertaken in the four months leading up to the EU Referendum; the questionnaire was launched on 11 March 2016 and closed at midnight on 23 June 2016 (i.e. the date of the EU Referendum). The main body of the questionnaire contained a total of 194 items, and it took respondents 25 minutes on average to complete (min = 6, max = 111, standard deviation = 13.9).
To avoid the respondent and coverage biases often considered as 'innate' to online surveys (Couper 2000;Sue and Ritter 2012), we used an active data collection method, directly targeting members of Polish online communities and readers of UK-based Polish-language online media portals. The social networking site Facebook served as the largest participant recruitment platform, where we advertised the survey in 21 Polish-language 'groups'. Two large UK-based Polish online newspapers have also helped to distribute the survey through their websites and official Facebook pages. Such strategies of 'appropriate targeting' proved particularly successful in research with hardto-reach populations, and were generally recommended as good practice for online data collection (Miller and Sønderlund 2010;Temple and Brown 2011). Furthermore, recent migration research has shown that migrants are increasingly less affected by 'coverage bias', as access to the internet is broadening while their involvement with the so-called online migration industry is extensive (Dekker and Engbersen 2014;Moreh 2014).
In order to assess the representativeness of our collected data, Table 1 compares the personal characteristics of our sample with those of the 2015-2016 Annual Population Survey (APS), which is itself based on the Labour Force Survey, and is the main source of information regarding the number and attributes of migrants in the UK.
As detailed in Table 1, females are clearly overrepresented in our sample (66%) compared to the APS, as are migrants in the 40-49 age group, those who had been living in the UK for less than 5 years, and those with secondary and postgraduate educational qualifications. However, in respect to educational qualifications and length of residence in the UK the APS itself is unrepresentative of the Polish migrant population, often failing to record foreign qualifications adequately and not accounting for migrants with less than one year of residence. In respect to the core personal and socio-economic characteristics of the respondents in our sample, we can further see that the average age is 38, and the average time spent in the UK is 7 years and 2 months, with 47% of our participants having been in the UK for 5-10 years. Those living in England constitute 80% of the sample, with a very broad distribution by counties and those in London making up less than 6.5% of the entire sample (not displayed in Table 1). About 40% had a secondary educational degree and 42% some postsecondary degree; 61% were employed full-time, 17% part-time, 9% were self-employed, and 13% were inactive; of those in employment, 18% worked in professional occupations in the highest two groups of the National Statistics Socio-Economic Classification, 34% in technical and semi-routine occupations, and 33% in routine occupations reflecting lower socio-economic status. As seen in Table 1, the major differences compared to the APS sample are in respect to those inactive (19% in the APS), presumably due to the higher proportion of elderly respondents, and those in higher managerial, administrative and professional occupations (8% in our sample compared to 3% in the APS), which parallels the already mentioned difference in the share of those with postgraduate qualifications.

Methods
We undertake a combination of bivariate and multivariate statistical analyses on the survey data described above, with the aim to identify and explain the factors which potentially determine the migration and civic integration plans of Polish migrants in the medium-term and as a short-term reaction to the Brexit vote.
We first examine bivariate associations between migrants' plans for the next five years and various demographic, socio-economic, interpersonal, migration-and civic-statusrelated, and attitudinal variables. Some relevant results of this analysis are included in Table A1 in the Appendix. This preliminary analysis enabled us to identify the potentially most relevant predictor variables and the internal structure of associations. Based on this, we undertake multinomial logistic regression analyses on two polytomous variables describing our participants' five-year plans and planned actions in case of a Brexit outcome. This analysis examines the factors determining plans not in a time frame confined to five years, but as an indication of how Polish migrants confront their 'immediate context of action' (Stones 2005, 166). The statistical analysis was performed using the IBM SPSS 23 software package.
Decisions regarding the variables to be included and the comparative structure of the regression model were made on the basis of both theoretical and data-driven considerations. Thus, some basic demographic and socio-economic variables were included even though they did not significantly correlate with our dependent variable or contributed to the model, yet they needed to be accounted for in relation to other variables (e.g. gender, due to the overrepresentation of women in our data). Some variables were excluded or recoded due to low counts in certain categories, andwhere it also aided explanationinitial variables were recoded based on the internal structure of the adjusted standardised residuals in their association with the dependent variable (see Table A1 in the Appendix). For our theoretical aims, since we were mostly interested in understanding how migrants formulate their actions in respect to their current status and their options within a rights-based framework, we chose as the reference category of our dependent variable the one which can be most closely associated with the habitus of 'indeterminacy', as we detail in the following section describing the structure of our measurement variables.

Measurement variables
The two main survey questions that organise our analysis relate to the respondents' future plans in respect to migration and civic integration (1) in a five-year time frame (five-year plan) and (2) in the more immediate context of a Leave vote in the EU Referendum (Brexit action). The two questions shared six similar response options, which in our analyses are coded to reflect four broader options for future action: (1) to leave the UK (re-migrate to a third country or return to Poland), (2) to remain in the UK and apply for a permanent residence certificate (PRC), (3) apply for British citizenship or (4) remain in the UK without any concrete plan/action (for a detailed list of the survey questions and derived variables, see Appendix Table A2).
Confining the approach taken by Snel, Faber, and Engbersen (2015) and Drinkwater and Garapich (2015) to a more time-restricted and rights-based framework, we consider those who choose to 'remain in the UK without any concrete plan' to represent the habitus of 'deliberate indeterminacy'. The focus of this paper required that those in the sample who already hold a PRC be distinguished from those who do not, given that their 'indeterminacy' has a different interpretation in respect to civic integration options. To highlight this latter case, a new category called 'No plan/action (with PRC)' was created and is presented in Figure 1, but was excluded from all bivariate and multivariate analyses, reducing the valid sample size by 61. The two variables 'five-year plan' and 'Brexit action' will serve as our dependent variables. The independent predictor variables are a combination of core socio-demographic characteristics that have been singled out in the literature as influencing some aspects of migrant behaviour related to our focus in this paper, and other characteristics concerning family relationships, perceptions on civic rights and opinions related to the context of Brexit that were specific to our survey. Some of the main socio-demographic characteristics were already presented in Table 1, and other survey-specific variables not presented there due to a lack of comparability with the APS are: children (in the UK), partner's economic activity (if not single); the language spoken in a workplace environment (English or other); whether the respondent is planning to apply for British citizenship 'at any time in the future', and whether they were eligible to do so at the time of the survey or within one year; whether they have close relatives who are eligible for a PRC within one year; whether they feel 'insecure' about the rights they currently hold as EU citizens; whether they feel 'anxious' about the possibility of Brexit; and their preference for the outcome of the EU Referendum.
These above variables were all considered in the bivariate analysis (see Table A1), but some were changed for the multivariate analyses. The variables relating to relationship status and children were excluded due to a lack of statistically significant influence; instead, the category of 'single' was included within the variable coding the 'partner's economic activity', while a dichotomous variable describing having 'close relatives in the UK' was created to include those who were not single and/or had children and/or parents in the UK. Accounting for the partner's economic activity status was deemed important due to the overrepresentation of women in the sample and the significant correlation between gender and economic activity (coded as in Table 1, χ 2 = 106.22 (3, N = 838), p = 0.000; Cramer's V = .356; 3.8% of the men in the sample were 'inactive', compared to 18% of the women). By controlling for the partner's economic status, we hoped to further reduce any bias caused by the underrepresentation of men in our sample.
Considering intentions to naturalise as a British citizen 'at any time in the future' enables us to grasp the relationship between subjective intentions and the rights-based framework we have introduced. It can indicate, for instance, what migrants' general disposition is towards acquiring British citizenship, and whether their attitudes translate into concrete actions in a five-year time frame or as a reaction to the vote for Brexit.
The variable measuring self-reported eligibility for British citizenship not only indicates the respondents' status, but also awareness of the legal requirements for naturalisation, by distinguishing the answer option 'Do not know (the eligibility requirements).' The two variables referring to feelings of 'insecurity' and 'anxiety' are meant to assess the influence of the respondents' state of mind on their planned actions. 'Brexit anxiety' was derived from two more general five-level Likert items regarding the perceived likelihood of a Brexit vote, and whether it would have a negative effect. 'Rights insecurity' derives from two Likert-type scales measuring the perceived likelihood that certain concrete entitlements held as EU citizens would become limited in the future, and whether losing that particular entitlement would have a negative effect; each scale consisted of nine similar items listing concrete social, work-related and mobility rights. Therefore, while 'Brexit anxiety' denotes a more general unease in respect to the EU referendum, 'rights uncertainty' has a more specific character, and it estimates whether the rights granted by EU citizenship are deemed to be in jeopardy regardless of the outcome of the Referendum.
Finally, the variable referring to the preferred outcome for the EU Referendum checks the influence of outcome favourability on migration strategies. A detailed description of the original survey questions and the ways in which our variables were derived is presented in Table A2 in the Appendix.

Analysis and findings
We first examine the distribution of responses to the two main survey questions examining migration plans in a five-year medium-term time frame and in case of Brexit. As shown in Figure 1, in the next five years 40% of the respondents were planning to apply for a PRC, 32% for citizenship, 5% were planning to return to Poland while merely 1% wished to move to a third country, and 22% were 'indeterminate', choosing to remain in the UK without any concrete plan. However, 7% were 'indeterminate' while already holding a PRC, and only 15% chose to remain in the UK relying solely on their EU free movement rights. 3 We can also see that in the case of Brexit a much higher proportion of the respondents would apply for permanent residence (51%), while in the share of those who are planning to naturalise we observe a decrease of a comparable magnitude (down to 21%). This change, however, is primarily attributable to those who would not qualify for naturalisation immediately after Brexit, and would therefore apply for PRC instead, which is nevertheless a required step before a potential future application for British citizenship [reference withheld]. The number of those planning to stay in the UK without taking any further action also decreases to 11%, while more respondents would choose to migrate to a third country (6%). There is only a negligible difference in the percentage of those who would return to Poland. In broad terms, we find that in case of a vote for Brexit 10% of the Poles in the UK would leave the country (compared to 6% otherwise), an overwhelming majority (72%) wished to remain and formalise their residence through civic integration measures regardless of the referendum result, and Brexit does seem to somewhat reduce the level of 'indeterminacy', giving support to a hypothesis that Brexit pushes Polish migrants towards having to formulate more concrete plans for the future.
The bivariate analysis examining the association between different variables and fiveyear plans discloses that age, education, socio-economic status, the language spoken in a workplace environment, wanting to apply for British citizenship 'at any time in the future' and eligibility to do so within one year, having close relatives eligible for PRC, and the respondent's preference as to the outcome of the EU Referendum are all significantly correlated (at the p < .05 level) with stated five-year plans ( Table A1 in the Appendix, which also includes descriptive statistics for the discussed variables). On the other hand, key socio-demographic and migration-related variables such as sex, economic activity, time spent in the UK, relationship status or having children do not seem to be on their own significantly related to migration plans. Apart from having general future naturalisation intentions, however, the strength of the significant associations is small (Cramer's V statistic is below .2 in all cases) and the examined factors each explain less than 2% of the variance in five-year plans (based on the Goodman and Kruskal τ test). As we might expect, having plans to apply for British citizenship 'at any time in the future' is very strongly associated with medium-term plans (Cramer's V = .542, p = .000)which includes plans to apply for citizenship in the next five yearsand explains 11% of the variance in migration strategies (Goodman and Kruskal τ = .109, p = .000). From the distribution of adjusted standardised residuals, we also find that those open to naturalisation in general were significantly more likely than expected to be planning to apply for British citizenship in the next five years (z = 13.8), and significantly less likely to consider leaving the UK, applying for PRC only, or taking no action (z = −6.6, −3.4 and −9, respectively).
Some other highly significant associations worth highlighting here are: speaking English in a workplace environment, which significantly increases the likelihood of naturalisation planning (z = 4.2) and decreases 'indeterminacy' (z = −4.7); perceived eligibility to apply for British citizenship within one year has a similar effect, with those eligible significantly more likely to be planning naturalisation within the next five years (z = 5.2)notably, however, not being eligible does not have a significant effect in the opposite direction, but 'not knowing the requirements' for naturalisation does (z = −4.6) and it also significantly increases the likelihood of having no concrete plans (z = 3.6); the same pattern of effects applies in respect to having close relatives who are eligible to apply for PRC, which increases the likelihood of naturalisation plans (z = 3) and decreases both intentions to leave (z = −3.1) and of being 'indeterminate' (z = −3)again, not knowing the requirements for permanent residence increases the likelihood of remaining in the UK without any concrete plans (z = 4.3) (for other results see Table A1, in the Appendix).
To account for the effect of all these variables, however, we also examine their interaction in a logistic regression model. Table 2 reports results from two separately run models, one only accounting for six core socio-demographic variables, namely sex, age, education, employment status, time spent in the UK and having close relatives in the UK (Model I, N = 746, Nagelkerke R 2 = .105), and one also including the survey-specific variables (Model II, N = 683, Nagelkerke R 2 = .473). The purpose of Model I is primarily to highlight the change in some of the effects following the inclusion of more specific variables.
As we find, gender has a very marginal role in indicating migration plans, while the effect of age and education proved more significant. Compared to those under 30, migrants aged 40 and over are twice to three times as likely to be planning civic integration (PRC or citizenship) as opposed to not having concrete plans, and less likely to leave the UK. Those in their thirties are also twice as likely as those in the younger age group to be planning permanent residence or citizenship (in Model II). Those whose highest achieved educational qualification is at the secondary school level have proven significantly less likely to be planning either to leave or to become British citizens than those holding post-secondary educational qualifications. Highly qualified individuals are four times as likely as those educated at the secondary level to be planning to leave the UK in the next five years, while at the same time twice as likely to apply for citizenship as those with secondary education, and 2.7 times as likely as those with vocational qualifications. Employment status, the length of time spent in the UK and having close relatives in the UK present more complex effects. For instance, employment status is not at all significant when accounting only for core socio-demographic characteristics (Model I), but it gains significance when also holding constant other attitudinal and interpersonal variables (Model II). Interestinglyand somewhat counterintuitivelyhaving any kind of employment (compared to being economically inactive) appears to very significantly reduce the odds of planning citizenship as opposed to not having concrete plans. Apparently, those economically inactive are six to twelve times more likely to be planning to apply for citizenship than those in any form of employment, and also three times more likely to apply for PRC than those employed part-time. However, these effects gain significance only when controlling for the language spoken at the workplace. As we can see from Table 2, being exposed to English language in a workplace environment very significantly increases the odds of planning to apply for citizenship as opposed to having no concrete plans (those speaking English at work are in fact seven times more likely to have such plans). What this shows is that it is not the fact of being in employment per se, but rather having a job in which they do not speak English which significantly decreases the likelihood of civic integration plans.
The length of time spent in the UK, on the other hand, correlates with one's eligibility for civic integration, and we can see how it gains statistical significance when controlling for the latter in Model II. Interestingly, those who had been in the UK for more than 10 years are significantly (and over three times more likely) to plan leaving compared to being undetermined than those with shorter residence, but only when knowledge about one's civic status is also considered; simultaneously, those who are aware that they are not eligible for British citizenship within one year are more likely to have plans to leave the UK within five years than those who are either eligible or who do not know whether they qualify for naturalisation or not. At the same time, being unaware of one's eligibility significantly increases the odds of 'indeterminacy' as opposed to having plans of naturalisation for the next five years, compared to knowing that one is not eligible (Exp(B) = .435); put differently, those aware of their short-term ineligibility are 2.3 times more likely to still intend to apply for citizenship within a 5-year time frame than those who had not shown interest in acquiring this knowledge about the requirements for naturalisation. In conjunction with the finding that awareness of having close family members who are eligible for PRC makes one three times more likely to be planning PRC oneself, and 2.6 times more likely to be planning naturalisation (while having close relatives in the UK, in itself, is not significant, see Table 2), the conclusion to be drawn is that prior interest in and awareness of one's rights and those of other close family members, are the factors that more strongly determine civic integration decisions, rather than simply the length of residence in the UK.
The strong statistical significance of having general future naturalisation intentions in determining medium-term plans is not the least surprising, yet the odds ratios that it generates are noteworthy; as we can see in Table 2, planning to apply for British citizenship 'at any time in the future' understandably decreases the odds of planning to leave the UK in the next five years as opposed to planning to remain without any concrete plans of civic integration in the medium-term, while increasing the odds of applying for permanent residence. We can also see that those with a general intention to naturalise are 106 times more likely to apply for British citizenship in the next five years rather than in the more distant future. 4 This finding may have considerable policy implications, especially since in our overall sample 66% of the Polish respondents have expressed an intention to apply for British citizenship 'at any time in the future'.
In respect to the remaining independent variables that we have not yet discussed, we find that having a partnerwho is either in employment or inactivesignificantly increases the odds of 'indeterminacy' compared to all other optional future plans (although there is no statistically significant difference in respect to planning citizenship when the partner is in employment). As mentioned in the previous section, the variable regarding the partner's economic status was mainly included as a control due to the overrepresentation of women in the sample, who were themselves significantly overrepresented among those in part-time employment and the economically inactive. By itself, however, the effect of the partnership variable is hard to interpret, although it may suggest that for those who have partners any decision regarding the future has to be made in agreement, which makes the decision-making process more difficult.
As we may have expected, feeling 'anxious' about a possible 'Brexit' was not significant in determining five-year plans 'regardless of the EU referendum', but feeling 'insecure' about the durability of currently enjoyed rights did make Polish migrants twice as likely to be planning naturalisation than continuing to rely on EU citizenship entitlements. Finally, respondents' preference in respect to the outcome of the Referendum was significant: those who preferred that the UK left the EU or remained a member only conditionally have significantly higher odds of being among those with a concrete plan for the future, be they intentions to leave or plans of civic integration.
We also ran the full model on another dependent variable referring to planned actions in case of a Brexit vote. The results are shown in Table 3, and we will highlight here some of the main differences compared to the results discussed above. Generally, the determinants of planned actions in case of Brexit are more difficult to reduce to any socio-economic or demographic factors; while the independent variables selected for analysis explained 47.3% of the variance in '5-year plans', they only explain 32.6% of the variance in 'Brexit actions' (Nagelkerke R 2 ). As we can see, age, education, partnership status, and having close relatives eligible for PRC all lose significance in some aspects, while speaking English in a workplace environment maintains its influence on naturalisation intentions. Length of residence in the UK loses its significance in respect to plans to leave, but gains significance in respect to permanent residence: those who had been in the UK for less than 5 years are 2.6 times more likely to apply for PRC rather than to take no action compared to those who had lived in the UK for over 10 years.
General intent to apply for British citizenship 'at any time in the future' also loses its significance in respect to plans to leave following Brexit, and the odds by which it determines plans to apply for citizenship in the 'immediate context of action' created by Brexit is also reduced to a nevertheless still considerable level (Exp(B) = 43.248). The changes in the explanatory power of eligibility for citizenship are, however, more interesting. While in case of Brexit it no longer explains plans to leave, it gains significance in respect to post-Brexit plans to naturalise. Following a Brexit vote, those eligible for British citizenship are seven times more likely to apply for citizenship than not to take any action, compared to those who are not eligible; furthermore, while unawareness of one's eligibility acted to reduce one's likelihood to contemplate naturalisation as a 5-year plan, following Brexit those not knowing whether they are eligible are 3.3 times as likely to want to apply for citizenshipas opposed to not taking any actionas those aware of their immediate ineligibility for citizenship. Again, these results highlight the relevance of rights-awareness, and at the same time reveal one of the central effects of Brexit on Polish migrants' strategies, namely its emotionally 'mobilising' effect.
In a Brexit scenario, the outcome favourability of the EU Referendum still significantly albeit more weaklydetermines plans to apply for PRC, but not other planned actions. Lastly, while 'rights insecurity' loses its significance, feeling 'anxious' about Brexit emerges as a highly significant determinant of post-Brexit migration strategies. As mentioned in the previous section, this derived variable combines opinions regarding the likelihood of a Brexit vote with opinions on what the personal effects of a Brexit would be (see also Table A2 in the Appendix). We can therefore assume that those who had not only thought that Brexit would have a negative effect on their lives, but also that it was something likely to occur (i.e. they felt 'anxious' about it) would have given more consideration to their options and planned actions following Brexit. As shown in Table 3, those 'anxious' about Brexit were 3.2 times more likely to plan to leave the UK than to stay without taking any action, and 2.5 times more likely to want to apply for either PRC or British citizenship following a Brexit vote. It seems, therefore, that it is the combination of rights-awareness, active consideration of one's options, and the emotional stress caused by the EU Referendum that most powerfully determines the migration strategies of Polish migrants in the UK following the Brexit vote.

Discussion and conclusions
The above analysis allows us to draw several conclusions regarding the medium-term plans and migration strategies of Polish nationals living in the UK, within the framework of civic and mobility rights available to them and in relation to the UK's prospective departure from the EU. As the descriptive results have shown, the overwhelming majority of our respondents (72%) were aiming for civic integration over the next five years, thus tying their legal status to UK law rather than relying solely on the constellation of rights provided by EU citizenship. We could also see that Brexit has little effect on this outcome on the 'aggregate' level [cf. reference withheld].
The results from the bivariate analysis broadly indicate that Polish migrants in the UK are not only a very diverse population in respect to socio-economic and demographic characteristics (see Burrell 2010, 299), but that these characteristics play a less important role in determining their future migration strategies under the terms discussed here. More meaningful is their general attitude towards British citizenship and the UK, their configuration of rights and knowledge thereof, and their exposure to the local cultural environment (as partially and indirectly measured by the use of English at work). This finding is also supported by the regression results: we could see that neither the length of residence nor having close family members in the UK significantly determined one's plans, but awareness of one's civic integration options (which largely depend on residence) or the rights of family members do have a significant influence. To some extent, this finding corresponds with that of Snel, Faber, and Engbersen (2015, 17), who find that 'it is not so much the duration of their stay as such that affects migrants' return intentions as their level of socio-cultural integration'. While we do not have comparable measurements of 'integration' to include in our model, and we have reframed our analysis in terms of 'rights' rather than the intended length of stay in the destination country, the underlying message behind our findings shares obvious similarities.
The fact that those with a declared preference for the UK to leave the EU (or remain a member only conditionally), were also more likely to have formulated clear plans for the medium-term future rather than being 'indeterminate', was another interesting finding which opens up to various interpretations. To understand the mechanisms behind this association would require further qualitative research, but we can think of several plausible hypotheses: one would be that already conceived plans of civic integration are the ones that determine attitudes towards the UK's EU membership, which may be shaped by an assimilation of mainstream majority attitudes, fears and insecurities about future immigration, or other factors that played a role in determining the result of the Referendum overall. This, however, would not explain mobility plans. Another possibility is the influence of general Euro-sceptic sentiments, and a further one could be, on the contrary, disillusionment with life in the UK and support for Britain's exit from the EU as a vindictive attitude coupled with intensions to leave the UK within five years. Examining these hypotheses, however, falls outside the scope of the present study.
In this paper, we were particularly interested in assessing the continued validity of Polish migrants' 'deliberate indeterminacy' (Drinkwater and Garapich 2015;Eade, Drinkwater, and Garapich 2007;Snel, Faber, and Engbersen 2015). Within the rights-based framework that we adopted, the habitus of 'deliberate indeterminacy' is still relevant, yet it is this opportunity to keep one's 'options deliberately open' as EU citizens (Eade, Drinkwater, and Garapich 2007, 11)which, from a rights-based perspective, could most closely be described by a lack of plans to leave in the medium-term coupled with a lack of intentions of civic integrationthat is most jeopardised as a consequence of Brexit. As our analysis has highlighted, the sense of insecurity regarding rights derived from EU citizenship was significant in determining naturalisation intentions as opposed to keeping options deliberately open in respect to civic integration. Unfortunately, our research cannot determine whether such insecurities had existed at the time of migration, or whether they were the consequence of the heated public debates on EU free movement which have taken place over the past couple of years, or of the UK Government's attempts before the EU Referendum to secure the right to restrict certain entitlements held by EU migrants. In either case, two-thirds of our respondents held such fears (see Table A1).
On the other hand, 'Brexit anxiety' was the main factor pushing Polish migrants away from 'indeterminacy' and towards having to formulate more concrete plans for the future in the eventuality of a Leave vote in the EU Referendum, whereas the simple fact of being eligible for naturalisation was the strongest factor explaining preferences for British citizenship. We have also seen that in case of Brexit those who had been in the UK for less than 5 years were more likely to apply for PRC than those who had been resident for longer than 10 years (while no other dimensions of length of stay proved significant in respect to 'Brexit actions'). This may reflect that those who have arrived to the UK more recently and may not yet qualify for naturalisation, but at the same time would prefer to remain in the UK for at least another five years, are drawn to permanent residence in greater numbers than some of those who migrated more than a decade ago and, who in case of Brexit, would rather opt for British citizenship or conclude that the time to leave Britain has come. Based on our findings, therefore, it may be legitimate to ask whether one important effect of Brexit is to engender an antithetical habitus of 'undeliberate determinacy'. Following from the above discussion, the main argument that we put forward in this paper is that the socio-economic and demographic variables which have emerged in qualitative research as significant in determining settlement or re-migration planssuch as relationship status, children, length of stay or employment among others (Janta 2013;McGhee, Travena, and Heath 2015;Ryan 2015;White 2011) are not on their own significant in determining the plans and strategies in respect to rights and options of civic integration and mobility in the case of as diverse a migrant group as that of Polish nationals in the UK. Rather, we argue that it is the additional elements of rights-awareness, interest and social proximity to the available options of civic integration, as well as anxieties and insecurities concerning supranationally derived rights, which carry more weight in explaining both medium-term plans and actions under the circumstances created by the UK's vote to exit the EU. This paper, nevertheless, opens up a series of questions that require separate in-depth treatment and need further research. We have put forward a case for a sociologically situated 'rights-based' approach to examining migration strategies, which takes account of the different constellation of rights that EU migrants possess or have access to, while at the same time reflecting upon the subjective interpretation of their rights by migrants. In taking forward such an approach, more qualitative research is needed in order to understand the social-psychological mechanisms behind the subjective interpretation of one's rights and the related proactive decision-making processes. By reconceptualising 'deliberate indeterminacy' and 'civic integration' within a temporally constrained rights-based approach, the need to (re)theorise them became more evident. As acknowledged in the introduction to our conceptual framework, civic integration could very well be employed as a strategy to expand one's options for remaining 'indeterminate' in the future, while indeterminacy should not necessarily reflect temporariness. In other words, neither of the two strategies may be signalling the 'settlement' or otherwise of Polish migrants, but rather new legal-structural moorings for transnational living, for being able to lead 'normal' lives while 'settled in mobility' (Galasińska and Kozłowska 2009;Morokvasic 2004). This assumption is also supported by our findings indicating that insecurities and anxieties about potentially losing currently possessed rights are driving forces behind civic integration actions. Understanding such processes and the ways in which the Brexit process will impact on Polish migrants' future strategies, however, will require a focused qualitative treatment.