Underenumeration of the Jewish population in the UK 2001 Census

Underenumeration of the JewishPopulation in the UK 2001 CensusDavid Graham* and Stanley Waterman*The Institute for Jewish Policy Research, 79 Wimpole Street, London W1G 9RY, UK

INTRODUCTION

On 29 April 2001 a census was carried out in the UK by three parallel agencies: the Office for National Statistics (ONS) (in

England & Wales), the General Register Office for Scotland (GRO), and the Northern IrelandStatistics and Research Agency (NISRA). Withthe exception of Northern Ireland, this censuswas unique in that it asked a question on religion.

This paper discusses the considerable debatewhich occurred during the 1990s concerning theintroduction of this question, and examines someof the issues involved in analysing the results inrespect of people who stated their religion to beJewish. It uses survey evidence and other censusrecords to show that there was a probable under-count of the Jewish population in 2001.

THE PURPOSE OF A CENSUS

Censuses do more than simply count people.According to the United Nations, the knowledgecreated by the collection of such data is ‘the cor-nerstone of democracy’ (UN Command Paper3882 1998, Annex C, para 1.1, cited in South-worth, 2001: 10). In the UK, the concept of official data collection is so important that it isenshrined in law in the Census Act of 1920(HMSO, 1993) of which part six of the Schedulestates that the census should be used to ‘obtainstatistical information with a view to ascertain-ing the social and civil condition of the popula-tion’. Although this aim is relatively benign,censuses are used extensively in political andsocial contexts in order to ‘determine access toresources, political and social visibility . . . aswell as defining how we conceive, understandand relate to people as regards race, ethnicity andreligion’ (Southworth, 2001: 6). In the US, thecensus is seen as a basic resource in the

POPULATION, SPACE AND PLACEPopul. Space Place 11, 89–102 (2005)Published online in Wiley InterScience (www.interscience.wiley.com). DOI: 10.1002/psp.362

Copyright © 2005 John Wiley & Sons, Ltd.

ABSTRACT

The size of the UK Jewish population hasalways been a source of uncertainty fordemographers. Following considerablediscussion and testing, a voluntary questionon religion was introduced into the 2001Census, which afforded the first opportunityto provide definitive answers to the socio-demographic make-up of Jews in Britain.However, examination of the 2001 Censusfigures and data from several large surveyssuggests that the census population of 266,740British Jews by religion is probably aconsiderable undercount. Jews areincreasingly defining themselves in ethnicrather than religious terms, so there is reasonto question the efficacy of the data derivedfrom the current format of the census question on religion and identity in general.With growing demands for comprehensiveplanning of social service needs, the necessityfor accurate data is more important than ever.Although much of this can be derived fromthe Census, there continues to be a key rolefor community-wide surveys. Copyright ©2005 John Wiley & Sons, Ltd.

Received 25 February 2004; revised 20 October 2004; accepted 22October 2004

Keywords: census; question on religion;underenumeration; Jewish; identity; ethnicity

* Correspondence to: D. Graham or S. Waterman, The Insti-tute for Jewish Policy Research, 79 Wimpole Street, London,W1G 9RY, UK.E-mail: [email protected] or [email protected]

redistricting and redistribution of political powerin Congress (Morrill, 1973).

The census is the only survey that provides a‘snapshot’ of the entire population in fine detail(Holt et al., 2001), further raising its status. More-over, with no population register in the UK, theGovernment, as well as voluntary and privatesector agencies, must rely heavily on its data asan authoritative source of social and economicindicators. Furthermore, the analysis of com-parative data from different censuses allowsresearchers to view changes in society over time,thus offering potential for longitudinal analysisand historical comparability (Sillitoe and White,1992).

On a practical level it is an efficacious meansof monetary distribution used ‘for the decade fol-lowing its collection as the base from which £600billion are allocated [by the Treasury] to localauthorities and health authorities alone’ (Holt et al., 2001: 455). Census data are used for localplanning and the delivery of services, as well asinitiating various commercial and social projects(Sillitoe and White, 1992; Holt et al., 2001). Localsubcommunities also benefit since it allows ‘insti-tutions and agencies . . . to provide effective andefficient services’ (Waterman and Kosmin, 1986a:483). They can also use the census to plan forfuture service provision and the development oflocal investment initiatives. For all these reasons,it is important that the data are accurate and,importantly, believed to be so by those who willuse them.

Since ‘population estimation is not an exactscience’ (Holt et al., 2001: 452) and no universallyacceptable criterion for classifying a nation’spopulation exists (Sillitoe and White, 1992),census data are often subject to lengthy delibera-tion in the months and years prior to and fol-lowing initial publication. This was especiallytrue in the debate that occurred over the inclu-sion of a census question on religion.

A CENSUS QUESTION ON RELIGION: THE DEBATE

In the mid-1990s, the Office for National Statis-tics (ONS) commenced a large consultation andtesting exercise over whether or not to include aquestion on religion in the UK’s censuses, takingsubmissions from representatives of the usercommunity via the Census Advisory Group

(Moss, 1999). The results of this work weredescribed in Parliament by Jonathan Sayeed MP,who noted that ‘[t]he wording of the proposedcensus question has been subject to detailed con-sultation with not only the users of census statis-tics, but a wide range of religious organisations. . .’ (Hansard, 2000b: Clm1159–60; HMSO, 2000c);ONS has ‘carried out two major tests of censusquestions: the 1997 census test and the 1999census rehearsal. Both exercises showed that thepublic are generally prepared to answer such aquestion . . . and that the quality of the responses[would be] sufficient to provide the informationthat would meet users’ express needs’ (Hansard,2000b: Clm1159–60; Hansard, 2000c: Clm308; see Weller, 2004: 9–12). In 1996, the ReligiousAffiliation Sub-Group had been set up,1 com-posed of representatives from various religiousorganisations, academics and other interestedparties, and consulted by the ONS. It was ‘unani-mously in favour of including a question on religious affiliation’ (Weller and Andrews, 1998;our italics), based on a strong body of opinion infavour of its inclusion. However, the debate grewincreasingly intense towards the end of the 1990s.

A CENSUS QUESTION ON RELIGION: THE NOES

There were many arguments against the inclu-sion of a census religion question, some morepersuasive than others. As Census Yearapproached, this debate was nowhere more vig-orous than in Parliament itself. For example, EricForth MP expressed concern that Governmentcould use the religion data to direct taxpayers’money towards preferred religious groups(Hansard, 2000c: Clm276). Further trepidationwas expressed by both Douglas Hogg MP andEdward Davey MP, who made strong represen-tations to the effect that such a question was‘intrusive’ (Hansard, 2000c: Clm271, 273 and278). This philosophical theme, that religionshould be considered a private matter of con-science, was elucidated by Graham Zellick whofelt strongly that the state had no business oblig-ing the general public to reveal their religion. Heurged both Jews and the wider UK population torefuse to answer such a question since it was ‘. . . wholly inconsistent with our traditions offreedom and personal privacy to ask a questionabout a person’s religious beliefs’ (Zellick, 1998).

90 D. Graham and S. Waterman

Copyright © 2005 John Wiley & Sons, Ltd. Popul. Space Place 11, 89–102 (2005)

On a practical level, Zellick also expressedconcern from an academic perspective, specifi-cally that the responses would not accuratelyreflect the religious make-up of society, since thequestion would only capture ‘affiliation’ or‘membership’, data already held by religiousbodies (Zellick, 1998). Others pointed to theproblem that low levels of literacy might inhibitcorrect responses amongst some religiousminorities (Weller and Andrews, 1998; Gonen,2005). Some Buddhists noted that a ‘question asto religious adherence is very personal . . . wesimply never ask [it]’ (Weller and Andrews,1998). Even the legal implications were usedagainst its inclusion, with some considering reli-gion to be so personal that a census questionmight contravene European Human Rights lawsby representing ‘an unprecedented exercise ofstate power’ (Zellick, 1998; Aspinall, 2000b;Hansard, 2000b: Clm1145; Hansard, 2000c:Clm271).

These concerns were aired with good reason.Some Muslim organisations noted that there was‘a fear of victimisation, particularly amongMuslims who may be branded as fundamental-ists’ (Weller and Andrews, 1998). Meanwhile,some Jews expressed concerns related to experi-ences in Nazi-occupied Europe, where in theNetherlands, for example, efficient use had beenmade of a comprehensive population registerwhich included religious identity in contributingto the annihilation of some 75% of the commu-nity there (Brasz, 2001).

A CENSUS QUESTION ON RELIGION: THE AYES

Although data on religion have routinely beencollected for many years in the UK – in prisons,the armed forces and NHS hospitals (Aspinall,2000b) – with no obvious ill effects, this was nota major factor in the argument for inclusion.Therefore, it is important to clarify why the deci-sion was made to include a question on religionin 2001, in light of the fears noted.

The first reasons were economic. Highlightingthe size and residential location of minoritygroups ameliorates the task of local and healthauthorities in allocating scarce resources effi-ciently (Comenetz, 2003). This improves theeffectiveness of Treasury monetary allocationand elevates the role of the census to that of a

‘public good’; like street lighting, it is somethingfrom which everyone gains but which is prohib-itively expensive for any individual or group toprovide alone.

Important social benefits were also foreseen inthe capacity of such a question to render ‘visible’certain groups whose identity was otherwisehidden. For example, ‘Some respondents saw thequestion as an opportunity to further define theircultural identity, or considered that religion wasa better indicator of their ethnicity and culturethan ethnic group alone’ (Moss, 1999: 32).Aspinall (2000b: 586) argued that it would sup-plement ethnicity data by ‘identifying ethnicminority subgroups, particularly those originat-ing from the Indian subcontinent (which containsHindus, Muslims, Sikhs and Christians), in termsof their religion’. Jews would also benefit fromthis unveiling because in the 1991 Census, theymostly fell into the very broad category of ‘WhiteBritish’ and had no other means to express a sep-arate identity. By revealing these ‘invisible’ sub-groups, Aspinall (2000a) argued that the questionon religion enhances the identity of religiousminority groups, since it provides for officialrecognition of the importance of religion insociety. Another ‘public policy’ benefit is relatedto the health and welfare of minority groups.Schmool notes that ‘[o]ne major reason why theJewish community supported the call for a reli-gion question was its own need for firm statisticson which to plan provision for social welfare,provision that would relate to both the age andthe state of health of the community’ (Schmool,2004: xxviii).

Perhaps the most persuasive argument for theinclusion of a question on religion was that itwould augment government baselines used tomonitor social disadvantage, inequalities andexclusion (Aspinall, 2000b; Hansard, 2000b:Clm1158; Hansard, 2000c).

‘The successful introduction of a question onethnic group in the 1991 Census underlined thevalue of the Census as a source for under-standing the patterns and processes of integra-tion, identifying and evaluating the impact ofinequalities, targeting services and assessingdisadvantage.’ (Moss, 1999: 31–2; see also Sillitoe and White, 1992)

There were also many practical, or ‘technical’potential benefits from its inclusion. Demo-

UK Jewish Population 91

Copyright © 2005 John Wiley & Sons, Ltd. Popul. Space Place 11, 89–102 (2005)

graphic researchers working on minority groupshave suffered from a serious paucity of data,effectively inhibiting communal strategic plan-ning and policy formation (Kosmin, 1998;Aspinall, 2000b). For instance, the 2002 BritishSocial Attitudes Survey was only able to sample15 Jewish people (NatCen, 2004), clearly an insuf-ficient number from which to draw any mean-ingful conclusions. In fact, this problem was themain reason why the Board of Deputies of BritishJews altered its position on the issue of the ques-tion’s insertion to one of support during the1990s (Schmool, 1998).

Not only were demographic data on religiousminorities scarce, but also their accuracy couldinvariably be questioned. Estimating the sizeand location of the Jewish population was onlypossible using ‘sophisticated guesswork’ andused several methods, one of which was the col-lection of synagogue membership records. Thiscreated institutional estimates, which inflatedpopulation estimates in towns and cities, butalso tended to miss unaffiliated Jews (Schmool,1998).2 (See further Waterman (1986), who dis-cusses the inadequacies of such data.) Anotherwidely used method for monitoring the Jewishpopulation is based on mortality or ‘acknowl-edgement of Jewishness at the time of death’(Waterman and Kosmin, 1986b; Schmool, 1998),which assumes that anyone likely to want to beconsidered Jewish will be buried (or cremated)as a Jew. However, such estimates became unre-liable when used in smaller areas (Schmool,1998). In 1997, the most recent year for whichdata were produced using these methods, theUK Jewish population was estimated at between283,000 and 285,000 people (Schmool and Cohen,1998).

Another technical problem, which census datawould ultimately make redundant, was the needby researchers to combine data obtained fromseveral small-scale surveys. For example, in theUS, where the census does not include a questionon religion, Glenn (1987) found that it was nec-essary to compare 110 separate studies to assessthe extent to which Americans report ‘No reli-gion’. Such complex methodologies are dramati-cally simplified when the same single censusquestion is asked of everyone at the same time.In addition, a unified and consistent data-setaffords cross-tabulation which substantiallyenhances the possibility of in-depth analyses of

religious minority groups. Within the Jewishpopulation, for example, the data have the poten-tial to establish, for the first time, the extent towhich Jews cohabit with and marry non-Jews –an important communal issue virtually bereft ofreliable data (Comenetz, 2003; Schmool, 2004).

Finally, it was noted that in taking this step,Britain would by no means be alone; severalother countries including Australia, Canada,India, Ireland and New Zealand have asked aquestion on religion for many years. None hassought to remove the question because of per-ceived problems with its inclusion. Ultimately,the decision to include the question was politicaland based on the fact that religion sits within the‘politics of identity [as a] highly importantmarker for self definition’ (Weller, 2004: 12). In2000 the British Government passed legislationso that a question on religion was included in the2001 Census (HMSO, 2000a). However, uniquely,in light of these various concerns, it was madevoluntary (HMSO, 2000b; Hansard 2000a,b).

SOURCES OF CENSUSUNDERENUMERATION: IN GENERAL

The census is the only survey in Britain thatattempts to include every individual, but in prac-tice coverage inevitably falls short by about 2%of the total population (ONS, 2003a,b). In the1991 Census, for instance, underenumerationwas 2.2%, but crucially, this is ‘not uniform acrossall areas or age-sex groups’ (Pereira, 2002: 18;ONS, 2003b; Holt et al., 2001). In 2001 the ONSemployed a technique called the One NumberCensus based on the Census Coverage Survey tocalculate the underenumeration (ONS, 2003b),which in turn is based on the knowledge thatcertain groups are more likely than others to beunderenumerated, for example, those living inareas with high rates of multiple occupancy andpoverty (Holt et al., 2001). In 2001 ‘[response]rates were lowest for persons in their twenties,particularly men, and for inner city areas’ (ONS,2003b). Simpson (2001: 424) notes that ‘Responseswere difficult to gain from locked blocks of flats,student housing, and single person households,and there are many more [people] than everbefore in each of these categories.’ Pereira (2002)has commented that the most common source ofunderenumeration is not the household unit –although this does happen – but individuals,

92 D. Graham and S. Waterman

Copyright © 2005 John Wiley & Sons, Ltd. Popul. Space Place 11, 89–102 (2005)

specifically, people who work away from home,who serve in the armed forces, or are in hospitalor on holiday on census day. Also, certain demo-graphic groups are more likely than others to beunderenumerated, including young males, thevery old, very recent births, and students (Holt et al., 2001). Still others include internal and international migrants and – linked to this – those whose first language is not English.

SOURCES OF CENSUSUNDERENUMERATION: FOR RELIGION

Compounding these potential sources of generalerror is a further set of risks specifically relatedto the question on religion. As noted, it was vol-untary and the Census stated this clearly (seeFigure 1). As the only non-mandatory questionon the Census, this was always likely to reducethe numbers responding. It should be noted thatalthough the intention was to capture informa-tion about religious affiliation (Moss, 1999: 32),and that Parliament understood this to be thecase (Hansard, 2000c: Clm308), Figure 1 showsthat the question wording ‘What is your reli-gion?’ in fact gave no indication that any distinc-tion had been made between this (affiliation) andreligious belief and religious practice. This was toavoid ‘confusion’, but it is unclear how ambigu-ity is removed without explanatory notes of anykind.

Because of the controversy surrounding itsinclusion, there is evidence that some peopletreated the question mockingly and possibly

used it as an opportunity to register protest, suchas, for example, the 390,000 people recorded as‘Jedi’ (ONS, 2003c). This was encouraged bymeans of an orchestrated e-mail campaign whichpromoted the forwarding of a hoax chain e-mailas well as negative publicity prior to the Census.This also occurred in Australia (70,000), NewZealand (54,000) and Canada (20,000) (Griggs,2001; CBC News, 2003).

Other problems related to fear and suspicionas to how the data might eventually be used. Forreasons already noted, many Jews were appre-hensive, but ultimately there was no official rejec-tion of the question (Weller and Andrews, 1998).Historically, many Jews have been suspicious ofgovernment-sponsored questionnaires requiringthem to identify their religion. This can be relateddirectly to a history of persecution of whichmany older members of the community still havepersonal memories. A final point relates to thereligious legality or tolerance of being counted.Orthodox Jewish law recognises the biblicalinjunction in Samuel II, Chapter 24, against thephysical counting of Jews. This will have affectedthe propensity of some religious Jews to answerthe question.

EVIDENCE OF JEWISHUNDERENUMERATION

Surveys

In March 2003, the ONS released non-responsedata for all 376 Local Authority Districts (LADs)in England and Wales. The non-response rate foreach question is defined as the sum of all blank,missing, incorrect and other spurious entriesdivided by the size of the eligible population (seefurther ONS, 2003e). It was found that for the(voluntary) question on religion, non-responsewas 7.6%. By comparison, non-response for thequestion on ethnicity was 2.9% and for age only0.5%. The Local Authority District (LAD) withthe highest religion non-response was theLondon Borough of Hackney at 12.1%, followedby neighbouring Haringey at 11.7%, two areaswith large orthodox Jewish populations. TheLondon Borough of Camden, the LAD with thefourth largest Jewish population, had the seventhhighest non-response rate (10%) and the LondonBorough of Barnet, with by far the largestnumber of Jews nationally, had the tenth highest

UK Jewish Population 93

Copyright © 2005 John Wiley & Sons, Ltd. Popul. Space Place 11, 89–102 (2005)

What is your religion?

∑ This question is voluntary ∑ Tick one box only

None Christian (including Church of England, Catholic,

Protestant and all other Christian denominations) Buddhist Hindu Jewish Muslim Sikh Any other religion, please write in

___________________________________

Figure 1. The 2001 Census question on religion as itappeared in England and Wales.

non-response rate (9.7%) of the 376 LADs (ONS,2003e). It should be noted that because the ques-tion on religion was voluntary, the ONS did notimpute the results (unlike for other censusresults) but instead published tables with ‘notstated’ data (ONS, 2003f).

Further evidence for a Jewish undercount isprovided by data from two substantial surveyscarried out shortly after 29 April 2001 – CensusDay. The first, in Leeds, was conducted in Julyand August 2001 by the Institute for JewishPolicy Research (JPR) (Waterman, 2003); thesecond, in London and the South East, was con-ducted by JPR together with the National Centrefor Social Research (NatCen) (Becher et al., 2002).Both surveys asked people about the way inwhich they responded to the question on religionin the 2001 Census.

Whilst accommodating practical constraintssuch as budget limitations and the lack of anational population register, an unprecedentedtotal of 2965 and 1496 questionnaires werereturned for London and Leeds respectively,making London alone ‘the largest direct surveyof British Jewry ever’ (Becher et al., 2002: 9). Evenso, as with any sample surveys, decisions weretaken to maximise the likelihood of reachingJewish households, meaning that some sectionsof the population were inevitably under-sampledto varying degrees. For example, ‘the extremesecular and religious fringes of the [Jewish] spec-trum’ (ibid.: 10), as well as singles, renters andconsequently those under 30 years old, the poorand women. In Leeds, a spatially compact Jewishpopulation made sampling relatively straightfor-ward (the minimum ‘best estimate’ response ratewas 30% (Waterman, 2003: 37)). But in Londonthe sample had to be split into three: North-West(‘best estimate’ response rate 32%), North-East(‘best estimate’ response rate 33%) and SouthLondon (‘best estimate’ response rate 21%)(Becher et al., 2002: 69). In North-East and North-West London, sampling was based on data fromthe Electoral Register, whereas in both SouthLondon and Leeds the Distinctive Jewish Namestechnique was utilised (see further Watermanand Kosmin, 1986a), using lists selected from192.com’s UK-Info database. Detailed descrip-tions of the sample design for the surveys havebeen described in Thomson and Waterman(2003), Becher et al. (2002: 66–9) and Waterman(2003: 35–7).

The results provide a clear indication of anational Jewish undercount. In both surveys, thecovering letter asked ‘anyone aged 18 or above. . . who is Jewish . . . who most recently had abirthday’ to complete the questionnaire(Thomson and Waterman, 2003: Appendix A).(Note that this differs, subtly, from the Census,which identifies those who said that their religionwas Jewish.) However, in Leeds only 87%, and inLondon only 84%, said that they had definitelyreported ‘Jewish’ in the 2001 Census (see Table 1).The remainder either ‘opted out’ by definitely notreporting ‘Jewish’ – 9% in both London andLeeds – or could not remember what answer theyhad given – 7% in London and 5% in Leeds.

From the point of view of the publishedCensus count for Jews, it is therefore appropriateto impute a new figure which accounts for these9% of British Jews that did not report their reli-gion as Jewish (as noted above, imputation wasnot carried out for Religion results by the ONS).Imputation, based on this evidence, places theJewish population size at a shade above 293,000,an increase of 26,000 people over the publishedfigure. This of course is a minimum imputation,since a further 7% were unable to rememberwhether they had reported being Jewish by reli-gion or not. Unfortunately, it is impossible toassess what proportion of this group ought alsoto have been imputed.

94 D. Graham and S. Waterman

Copyright © 2005 John Wiley & Sons, Ltd. Popul. Space Place 11, 89–102 (2005)

Table 1. In the national Census of 29 April 2001, therewas a voluntary question on religion. Did you answer‘Jewish’ for this question?

London survey Leeds survey(%) (%)

Response given: (n = 2936) (n = 1417)

Yes (I chose Jewish) 83.7 86.6No – I chose not to 5.3 6.4

answer that question

No – I gave a 1.1 N/Adifferent answer

No – I did not fill 2.5 2.2in a Census form

I cannot remember 7.3 4.9Total 100.0* 100.0*

* Columns do not sum to 100 due to rounding.Sources: Becher et al. (2002); JPR/NatCen (2003) unpublisheddata.

The survey data permitted cross-tabulations toassess whether there were any specific attributesaffecting the propensity of an individual toanswer the question on religion. For example, thesurveys found no difference in the propensity ofmales and females to respond ‘Yes, [I choseJewish]’. This is interesting, given that amongstthe population at large, women tend to be morereligious than men (Glenn, 1987; Stark, 2002), andyoung males are a potentially underenumeratedgroup. However, there was a statistically signifi-cant difference for age. The London data showedthat 74% of those under 30 (n = 91) said that theywere Jewish, compared with 84% of those aged30 and above (n = 2354) (Pearson chi-square sig-nificant at the 0.001 level).

Marital status was another important demo-graphic variable where differences wereobserved. The surveys found a significantlyreduced propensity on the part of cohabitingcouples, who defined themselves as Jewish, tohave said that they were Jewish by religion in theCensus (see Table 2). Although 85% (n = 1856) ofmarried couples in London reported that theywere Jewish, only 68% (n = 49) of cohabitingcouples said that they were (Pearson chi-squarewas significant at the 0.001 level). A statisticallysignificant result was also obtained in Leeds forthe marital status variable.

Furthermore, it was found that people whowere affiliated to a synagogue (regardless ofdenomination) showed a higher propensity toreport Jewish by religion in the Census comparedwith those who were not. In London, 86% (n =2083) of respondents who reported that theywere members of a synagogue said that they

were Jewish in the Census, compared with 68%(n = 322) of those reporting that they were notmembers of any synagogue (Pearson chi-squarewas significant at the 0.001 level). A statisticallysignificant result was also obtained in Leeds. Thistendency to under-report among unaffiliatedJews is very important as national levels of synagogue affiliation decline.

In terms of current Jewish practice, similarresults were obtained. It was found that amongpeople who described themselves as ‘Just Jewish’or ‘Reform/ Progressive’ or ‘Traditional (notstrictly Orthodox)’ or ‘Orthodox (e.g. would notturn on a light on Sabbath)’, 86% (n = 2168)reported ‘Jewish’ in the Census. This comparedwith only 69% (n = 229) for respondents describ-ing themselves as ‘Non-practising (i.e. secular/cultural)’ Jews. Similar results were found forquestions on Jewish upbringing and Jewish consciousness.

Whilst the sampling pools from the Londonand Leeds surveys tended to concentrate on‘middle-of the-road’ Jews, there is considerableevidence to suggest that it was Jews at bothextremes of the religious spectrum – those with a secular/cultural outlook and the strictly Orthodox – who were at greatest risk of beingunderenumerated in the Census. These twogroups will now be examined.

Seculars

Within the Jewish population, there has for sometime been evidence that secular Jews – those whoare unaffiliated and have little, if any, religiousbelief – identify less by faith than by culture anda ‘shared ethnicity’ (Kosmin, 1999a). As Miller et al. (1996: 3) have noted, ‘unlike other religiousgroups, in the Jewish community, levels of ritualobservance are far more closely related to ethnicidentity than to strength of belief’. As if to under-line this point, the Race Relations Act 1976defines Jews as both a religious and an ethnicgroup (Waterman and Kosmin, 1986a; Aspinall,2000b). The crucial difference between these twoconcepts is that of personal choice. Whereas ortho-dox Jewish law uses matrilineal descent or offi-cially recognised orthodox conversion as the coredefinition of being Jewish, emancipation intoWestern societies has encouraged many Jews toembrace a more liberal understanding of theirown identities.

UK Jewish Population 95

Copyright © 2005 John Wiley & Sons, Ltd. Popul. Space Place 11, 89–102 (2005)

Table 2. Proportion of survey respondents stating ‘Yes,(I chose Jewish)’ in the 2001 Census, by marital status.

London Leeds survey survey

Marital status % n % n

Married 85 (1856) 88 (775)Cohabiting 68 (49) 77 (23)Divorced/separated 74 (112) 96 (67)Widowed 85 (277) 82 (244)Single/never married 77 (158) 84 (112)Mean 84 – 87 –

Source: JPR/NatCen (2003) unpublished data.

To highlight the importance of this issue, com-ments made during the Parliamentary debateabout the question on religion are pertinent.Evan Harris MP told the House ‘I was bornJewish, of a Jewish mother, and . . . I considermyself Jewish culturally, but I do not have reli-gious belief. I would be worried, as wouldothers, about giving an inaccurate answer [to acensus question on religion]’ (Hansard, 2000b:Clm1141–2). Further, John Bercow MP expressedsimilar concerns ‘. . . although my [Jewish]parentage is different from that of [Evan Harris]I identify passionately with the Jewish peopleand the State of Israel, but I confess that thesedays I go to synagogue only for funerals’(Hansard, 2000b: Clm1142).

This tendency was analysed by Glenn (1987)in the US as early as the mid-1980s. He showsthat in surveys, Americans raised as Jewish weremore likely to report ‘no religious preference’than were those raised as Protestants orCatholics (Glenn, 1987; Kosmin and Lachman,1993; Kosmin, 2002). Interestingly, the concept ofa dual ethno-religious identity is not necessarilyunique to Jews. For example, Jacobson notes thatIslam is described by some Muslims not simplyas a religion but as a ‘way of life’, and that for less observant Muslims it was ‘feelings of dif-ference . . . and distance’ that mattered most(Jacobson, 1997: 249–51, italics in original). Theissue of definition is also prevalent amongstSikhs and Hindus (Weller, 2004). But of course,in the Census these groups have the option tostate their identity in ethno-national terms if theyso choose (for example, by ticking ‘Indian’within the question on ethnicity). This option isnot available to most ‘ethnic’ Jews who mustwrite ‘Jewish’ in the relevant section. Such‘ethnic-others’ are numerous. In fact, the ONSacknowledges 80 subcategories of ‘perceivedethnic group and cultural background’ (ONS,2004: 56–7), including all the large religions aswell as Israelis and Arabs. Unfortunately, thecoloured/nationalistic flavour of the Censusquestion on ethnicity means that the accom-panying note on the form to respondents regarding one’s ‘cultural background’ couldeasily be overlooked.

Evidence from the results of the 2001 ScottishCensus illuminates where the Census fromEngland and Wales is shady. In Scotland, respon-dents were asked to report their ‘current religion’

and their ‘religion of upbringing’. The questionwording was different too: ‘What religion, reli-gious denomination or body do you belong to?’followed by ‘What religion, religious denomina-tion or body were you brought up in?’. Theoption to select ‘Jewish’ was available in bothquestions. Note that the notion of belonging usedhere is absent in England and Wales. All told,25% of those reporting a Jewish upbringing didnot report that they were currently Jewish (byreligion) (Graham, 2003a). In this way the qualityof the Scottish data on religion was substantiallymore accurate than that for England and Walessince it presented additional, temporal infor-mation on accessions ( joiners) and secessions(leavers). It hints, at the very least, at substantialunderenumeration, possibly of many tens ofthousands of Jews in the UK. But it would cer-tainly be imprudent to impute a UK populationfigure based solely on these data for two reasons.Firstly, the Scottish data are a relatively small andnot necessarily representative sample, and sec-ondly, the limited options available in Englandand Wales may have meant that some people,who would otherwise have only reported theirupbringing as Jewish, were left with no optionbut to report current religion Jewish if they wereto be enumerated at all. Such people cannot beconsidered inconsequential for practical pur-poses such as marriage and burial. At the veryleast, these issues mean that the two data-sets forScotland and England and Wales cannot easily becombined to produce one meaningful figure ofthe total number of UK Jews.

One way of measuring the tendency of Jews toidentify in an ethnic rather than a religious wayis the concept of outlook (Graham, 2003b). Return-ing to the JPR/NatCen survey data, respondentswere also asked to say whether they saw them-selves as being ‘Secular’ or ‘Somewhat secular’or ‘Somewhat religious’ or ‘Religious.’ Twenty-five per cent considered their outlook to besecular, and these respondents were less likely tohave answered ‘Jewish’ in the Census than thosewith a more religious outlook – 77% (n = 536)compared with 87% (n = 1845) respectively(London survey data) (see Table 3).

An important and interesting point now arises.Had the 2001 Census explicitly offered a Jewishoption within the ethnicity question, in additionto the religion option, would these ‘missing’secular Jews have been correctly enumerated? At

96 D. Graham and S. Waterman

Copyright © 2005 John Wiley & Sons, Ltd. Popul. Space Place 11, 89–102 (2005)

this juncture the Canadian census offers a usefulinsight, where the concept of being ‘Jewish byethnicity’ is recognised. The Canadian ethnicityquestion (based on ‘ancestral origin’ (StatisticsCanada, 2003a: 38)) is explicit; ‘Jewish’ is one ofseveral possible examples (including mixed eth-nicity). The 2001 Canadian Census reported329,995 ‘Jews by religion’ but 348,605 ‘Jews byethnicity’ (Statistics Canada, 2003b,c). The differ-ence of 5.6% is, however, apparent, and does notmean there are 18,610 more ethnic than religiousJews. Some people will have selected bothoptions, whilst some would have selected one orthe other (see Kosmin, 1999b, with reference to1991 data). In the case of the UK, it is thereforeargued that the 2001 Census failed to recordmany Jews who self-identify by ethnicity ratherthan by religion.

Linked to, but independent of, the seculargroup are Israelis, an increasingly significant pro-portion of the UK’s Jewish population (Schmooland Cohen, 1998). Not only is this group pre-dominantly secular, but it also exhibits severalunique features that will probably have causedunderenumeration. Although there are manylong-term Israeli residents living in the UK,many others are visiting students and recently-arrived migrants living in rented accommoda-tion, and, in terms of underenumeration, aretherefore a higher-risk group. Unfortunately,only rough estimates are available of the numberof Israelis living in the UK, and at the time ofwriting (October 2004), data from the 2001Census on Israel as the country of birth had notbeen made available. However, even if such datawere available, it would not reveal the true sizeof the UK’s Israeli population because not allpeople who are Israeli by nationality, or whosemother tongue is Hebrew, were necessarily bornin Israel (Davids, 2003).

Strictly Orthodox Jews

At the other extreme of the Jewish spectrum arethe strictly Orthodox (Haredi or Hassidic) Jews,and here too there is evidence of an undercount.Around 10–15% of British Jews are strictly Orthodox (Valins, 2003) and most live in theLondon Borough of Hackney and adjacentHaringey (forming the ‘Stamford Hill commu-nity’). The other principal centres are in theLondon Borough of Barnet (within the wards ofTemple Fortune, Hendon and Golders Green),Broughton Park in Greater Manchester, andGateshead in northeast England.

Until the 2001 Census, two data sources hadbeen particularly useful in estimating the com-munity’s size: synagogue membership records,useful because of the high propensity amongstmales to be members of a synagogue, and the‘North London Shomrei Shabbos [lit. SabbathObservers] Directory’, which contains addressdetails of most strictly Orthodox Jewish house-holds in the Stamford Hill vicinity (Gonen, 2005).Yet despite this knowledge, estimating the size of the strictly Orthodox community with anydegree of accuracy has proved difficult. Forexample, the Board of Deputies estimatedHackney’s Jewish population in 1996 to be 17,900(Schmool and Cohen, 2002), whereas Holman(2001) suggested more than ten different figures,ranging from below 10,000 to above 22,000,depending on who was included and how thefigure was calculated. However, in a separatestudy Holman and Holman (2002) estimated thepopulation to be between 17,780 and 22,225. Afurther confusing factor is that not every Jewishperson in Stamford Hill is necessarily strictlyOrthodox, especially some older residents wholived there prior to the growth of the Orthodoxgroup as well as some secular Jews with littlecommunal attachment. In addition, there is anOrthodox community of Yemenite origin notlinked to the strictly Orthodox population, whomay or may not appear in the various estimatesand orthodox ‘communal’ lists.

Notwithstanding these grey areas, the 2001Census recorded just 10,732 Jewish people in thewhole of Hackney. For the five wards in whichthe strictly Orthodox community is predomi-nantly located (Cazenove, Lordship, New Riverand Springfield in Hackney, and Seven Sisters inadjacent Haringey), the Census recorded just

UK Jewish Population 97

Copyright © 2005 John Wiley & Sons, Ltd. Popul. Space Place 11, 89–102 (2005)

Table 3. Proportion of survey respondents stating ‘Yes,(I chose Jewish)’ in the 2001 Census, by outlook(London).

Somewhat SomewhatSecular secular religious Religious

% 77 87 88 86n 976 1295 1563 368

Source: JPR/NatCen (2003) unpublished data.

9552 Jews, a surprisingly low figure given previ-ous estimates.

In attempting to reconcile this low total, it isinstructive to note additional factors which con-tribute to the difficulty of enumerating strictlyOrthodox Jews. Residentially highly concen-trated, they tend to have demographic and socio-economic characteristics that differ from otherBritish Jews. They have a much younger ageprofile and far higher birth rates, with very largehouseholds. Significant proportions are in thelower NS-SeC (National Statistics Socio-economic Classification) categories, many do notspeak English as a first language and were notborn in the UK (Holman, 2001; Holman andHolman, 2002). In addition, strictly OrthodoxJews are more likely to be influenced by the bib-lical prohibition on counting Jews noted above,and generally adhere to the concept of da’at torah which refers to ‘the totality of rabbinicknowledge and interpretation’ (Kellner, 2003), aprinciple of significance if a rabbi decrees his congregants not to answer the voluntary Censusquestion on religion.

Together, these factors suggest that underenu-meration was inevitable. As already noted, non-response in Hackney and Haringey was thehighest out of all 376 LADs in England andWales. By analysing ward-level data in theseareas, an indication of the extent to which therewas underenumeration can be gleaned. This isachieved by comparing the proportion of thepopulation of a ward that reported ‘Jewish’ in theCensus with the RNS (Religion Not Specified)group, i.e. the proportion of the ward that did not respond to the religion question at all, choosing in effect to ‘opt out’.

Table 4 shows the proportion of Jews in each ofHackney’s 19 wards plus Seven Sisters ward inHaringey. It also shows the proportion of eachward’s population that did not specify a religiouspreference (RNS) in the 2001 Census. A Pearsoncorrelation was carried out to examine the rela-tionship between the proportion of each ward thatwas Jewish and the proportion that did notspecify any religion (RNS). The correlation resultwas 0.89 (significant at the P < 0.01 level). No othercategory (Muslim, Hindu, No Religion, etc.) pos-itively correlated with RNS to such an extreme.An imperfect yet clear positive relationship there-fore exists between the proportion of Jews in award and its RNS proportion. However, it is not

clear why Kings Park ward, of which less than 1%is Jewish, has an RNS of 14%. Other idiosyncrasiesin Hackney include 30% of Clissold ward (1.6%Jewish) reporting No Religion: unusual given themean for Hackney is 12% (ONS, 2003d).3

Table 5 shows that the proportion of RNSresponses in the five ‘Stamford Hill’ wards is16.1%, which is over twice the national level of7.7%. This result was corroborated by a similaranalysis of the strictly Orthodox population inGreater Manchester. The majority of that com-munity lives in Salford, which recorded 5173Jews in the 2001 Census, of which 91% live in justtwo out of 20 wards: Kersal and Broughton. Theproportions of RNS for these two wards were15.9% and 11.2% respectively. The mean RNS for the remaining Salford wards was 7.8%. It istherefore clear that in areas with high densities of strictly Orthodox Jews, large proportions of people chose to opt out of answering the voluntary question on religion.

98 D. Graham and S. Waterman

Copyright © 2005 John Wiley & Sons, Ltd. Popul. Space Place 11, 89–102 (2005)

Table 4. Jewish and RNS proportions by ward inHackney and Seven Sisters ward in Haringey (%).

Proportion Proportion of ward of ward that where religion was

Ward is Jewish not specified (RNS)

Lordship 17.2 19.0Springfield 23.5 17.8New River 20.4 16.1Cazenove 13.2 14.6King’s Park 0.8 14.0Seven Sisters (LB 10.3 13.3

Haringey)Wick 1.4 12.5Victoria 1.6 11.4De Beauvoir 1.2 10.6Stoke Newington 2.3 10.6

CentralQueensbridge 1.0 10.6Brownswood 3.7 10.5Dalston 1.8 10.2Hackney Downs 2.1 10.2Haggerston 0.9 10.0Chatham 0.7 10.0Leabridge 1.5 10.0Hoxton 0.8 9.6Hackney Central 2.0 9.5Clissold 1.6 9.4

Source: ONS (2003d).

LESSONS TO BE LEARNED

People define themselves in numerous ways, andthe census rightly deems it fit to attempt tomeasure how they do this. As Jonathan SayeedMP said to Parliament ‘. . . Many people intoday’s society choose to identify themselves interms of their religion or culture, [as well as theirethnicity] and the 2001 Census provides a once-in-a-decade opportunity to collect informationfrom groups that increasingly prefer primarily toidentify themselves in such a way’ (Hansard,2000b: Clm1160). Consequently, the inclusion ofa question on religion in the UK 2001 Census islaudable and has produced a wealth of valuabledata. But in doing so, its inclusion, in conjunctionwith the previously existing question on ethnic-ity, takes the census into new territory: that of thecomplex and subjective realm of identity ratherthan the more straightforward realm of tangible,domestic facts. We have argued here that, in thecase of the Jewish ‘snapshot’ at least, where theconcept of religion and its relationship with ethnicity is not clear-cut, the results of the 2001Census were inaccurate and were subject tounderenumeration. In order to improve theoverall level of accuracy and reliability of the religion counts in future censuses, we suggest the incorporation of several amendments to thecensus form.

Firstly, given the survey evidence showing thatBritish Jews increasingly identify on ethnic ratherthan religious grounds, the 2011 Census shouldmake it far clearer that the ethnicity question con-siders cultural background to be as valid as skincolour and should include an example such as‘Jewish’.

Secondly, the question on religion must high-light the fact that this is in relation to religiousaffiliation only. Its current format does not removethe possibility of interpreting it as a question onreligious belief or practice.

Thirdly, the current format of the question onreligion in England and Wales should beamended to match the format used in Scotland,so that it includes a separate question on religionof upbringing. The question format resulted inthe quality of the data in Scotland being superiorto that in England and Wales.

Although these measures would undoubtedlyimprove the accuracy and quality of the censusdata on religion in 2011, it is nevertheless essen-tial for religious (minority) groups to continue tocarry out local-level surveys. Bespoke surveyscan collect data that censuses are simply notdesigned to do, such as asking about levels ofreligiosity and religious commitment, levels ofreligious education, schooling experienced andschooling preferences of parents, details aboutcharitable giving and voluntary work done,social pastimes and activities, as well as ex-panding on existing census questions to obtainmore detailed information about health, caringand family circumstances. The optimum socialresearch solution, therefore, is sample surveys inconjunction with the comprehensive coverage ofa well-drafted census form.

NOTES

(1) The group produced a report called the ‘ReligiousAffiliation Sub-Group of the Census ContentWorking Group of the Office for National Statistics1997 Indicative Business Case for the Religion

UK Jewish Population 99

Copyright © 2005 John Wiley & Sons, Ltd. Popul. Space Place 11, 89–102 (2005)

Table 5. RNS proportions at various levels for the 2001 Census question on religion.

Number Religion Not Total population % Religion Not Area Specified (RNS) of area Specified (RNS) (%)

Hackney wards of: Springfield, 9,220 57,341 16.1%New River, Lordship and Cazenove. Plus Seven Sisters ward in Haringey

Remainder of Hackney wards 16,852 158,662 10.6%Inner London 270,940 2,766,114 9.8%England and Wales 4,010,658 52,041,916 7.7%

Source: ONS (2003d).

Question in the 2001 Census’. They later becamethe Census 2001: Religious Affiliation Group,which was in effect a lobbying group for the ques-tion on religion (Weller, 2004: 11).

(2) Unpublished analysis by Graham (2003c) revealedthat 14% of the UK’s Jewish population did not livein the previously ‘known’ urban concentrations.These people were distributed thinly across ruraldistricts of the UK.

(3) A second Pearson correlation was carried out toexamine the relationship between the proportionof the ward that was Jewish and the proportionthat reported No Religion. The relationship wasfound to be -0.55 (significant at P < 0.05), i.e. neg-ative and less strong than the RNS relationship.This suggests that this (Orthodox) Jewish popula-tion was not inclined towards selecting No Religion,which other, more secular and less concentratedJewish groups may have been.

REFERENCES

Aspinall P. 2000a. The new 2001 Census Question seton cultural characteristics: is it useful for the moni-toring of the health status of people from ethnicgroups in Britain? Ethnicity and Health 5(1): 33–40.

Aspinall P. 2000b. Should a question on ‘religion’ beasked in the 2001 British Census? A public policycase in favour. Social Policy and Administration 34(5):584–600.

Becher H, Waterman S, Kosmin B, Thomson K. 2002.A Portrait of Jews in London and the South-east: A Com-munity Study. JPR: London.

Brasz C. 2001. After the Shoah: Continuity and changein the post-War Jewish community of the Nether-lands. Jewish History 15: 149–168.

CBC News. 2003. Canada has 20,000 Jedis: Census.http://www.cbc.ca/stories/2003/05/13/census_religion030513 [accessed 13 May 2003].

Comenetz J. 2003. Stand up and be counted in NationalCensus. Forward. http://www.forward.com/issues/2003/03.11.07/oped2.html [accessed 7November 2003].

Davids L. 2003. Israelis in Canada today: the censuspicture. Canadian Jewish News 27 February 2003.

Glenn ND. 1987. The trend in ‘No Religion’ respon-dents to US national surveys, late 1950s to early1980s. Public Opinion Quarterly 51(3): 293–314.

Gonen A. 2005. Strictly Orthodox Jews Negotiating theModern Metropolis: Stamford Hill in London. Floer-sheimer Institute for Policy Studies: Jerusalem.

Graham D. 2003a. So how many Jews are there in the UK? The UK Census and the size of the Jewishpopulation. JPR News Spring 2003. www.jpr.org.uk/Newsletter/index_spring_2003.htm

Graham D. 2003b. Secular or Religious? The Outlook ofLondon’s Jews. Report 3, JPR: London.

Graham D. 2003c. The Geographical Distribution of theUK’s Jewish Population in 2001. Unpublished workingpaper. JPR: London.

Griggs K. 2001. Jedis Stand for the Census Count.http://www.wired.com/news/culture/0,1284,44141-2,00.html [accessed 31 May 2001].

Hansard. 2000a. Census (Amendment) Bill [Lords] Asamended in the Standing Committee, considered.Clause 1. Particulars in respect of religion may berequired. http://www.publicwhip.org.uk/divi-sion.php?date=2000-07-26&number=293 [accessed23 September 2004].

Hansard. 2000b. Census (Amendment) Bill [Lords]Order for Third Reading. Columns 1133 to 1176.http://www.publications.parliament.uk/pa/cm199900/cmhansrd/vo000726/debtext/00726-10.htm [accessed 24 September 2004].

Hansard. 2000c. Census (Amendment) Bill [Lords]Order for Second Reading. Columns 266 to 311.http://www.parliament.the-stationery-office.co.uk/pa/cm199900/cmhansrd/vo000620/debtext/00620–34.htm [accessed 24 September 2004].

HMSO. 1993. Statutes in Force, Official RevisedEdition. Census Act 1920. HMSO: London.http://www.statistics.gov.uk/census2001/pdfs/censusact1920.pdf [accessed 21 September 2004].

HMSO. 2000a. Census (Amendment) Act 2000.Chapter 24. Crown Copyright. http://www.hmso.gov.uk/acts/acts2000/20000024.htm [accessed 24September 2004].

HMSO. 2000b. Statutory Instrument 2000 No. 3249 The Census (Amendment) Order 2000. CrownCopyright. http://www.legislation.hmso.gov.uk/si/si2000/20003249.htm [accessed 9 September2004].

HMSO. 2000c. Explanatory Notes to Census (Amendment) Act 2000. Chapter 24. Crown Copyright. http://www.hmso.gov.uk/acts/en2000/2000en24.htm [accessed 27 September 2004].

Holman C. 2001. Orthodox Jewish Housing Need in Stamford Hill. Unpublished paper. De Montfort University: Leicester.

Holman C, Holman N. 2002. Torah, worship and acts of loving kindness: Baseline indicators for the Charedicommunity in Stamford Hill. De Montfort University:Leicester.

Holt T, Diamond I, Cruddas M. 2001. Risk in officialstatistics: a case-study of the 2001 one-numbercensus project. The Statistician 50(4): 441–456.

Jacobson J. 1997. Religion and ethnicity: dual and alternative sources of identity among young BritishPakistanis. Ethnic and Racial Studies 20: 238–256.

Kellner M. 2003. JPR initiates Jewish enrichment sem-inars. JPR News Autumn 2003. http://www.jpr.org.

100 D. Graham and S. Waterman

Copyright © 2005 John Wiley & Sons, Ltd. Popul. Space Place 11, 89–102 (2005)

uk/Newsletter/index_autumn_2003.htm [accessed29 December 2003].

Kosmin B. 1998. A religious question in the BritishCensus? Patterns of Prejudice 32: 9–46.

Kosmin B. 1999a. Ethnic and Religious Questions in the2001 UK Census of Population: Policy Recommenda-tions. Policy Paper 2, March. JPR: London.

Kosmin B. 1999b. The Jewish Market in NorthAmerica. In Creating the Jewish Future, Brown M,Lightman B (eds). Sage: London; 219–233.

Kosmin B. 2002. As secular as they come. Moment June:44–49.

Kosmin B, Lachman S. 1993. One Nation Under God.Harmony Books: New York.

Miller S, Schmool M, Lerman A. 1996. Social and Polit-ical Attitudes of British Jews: Some Key Findings of theJPR Survey. Report 1. JPR: London.

Morrill RL. 1973. Ideal and reality in reapportionment.Annals of the Association of American Geographers 63:463–477.

Moss C. 1999. Selection of topics and questions for the2001 Census. Population Trends 97: 28–36.http://www.statistics.gov.uk/census2001/pdfs/celiamos.pdf [accessed 14 September 2004].

NatCen. 2004. (National Centre for Social Research)British Social Attitudes Survey 2002 [computer file]UK Data Archive [distributor]: Colchester, Essex.SN: 4838.

ONS. 2003a. Census 2001: Methodology: Response Rates:Annex B Better methods. http://www.statistics.gov.uk/census2001/annexb.asp [accessed 13 May 2003].

ONS. 2003b. A One Number Census: Census 2001 was themost accurate count of the UK population ever achieved.http://www.statistics.gov.uk/census2001/onc.asp[accessed 21 September 2004].

ONS. 2003c. 390,000 Jedis There Are: But did hoax cam-paign boost response in teens and 20s? http://www.statistics.gov.uk/CCI/nugget.asp?ID=297&Pos=&ColRank=2&Rank=1000 [accessed 14 October2004].

ONS. 2003d. Census 2001 CAS for Output Areas, Wards,and higher administrative geographies: DVD 07 Wards,and higher administrative geographies in England andWales, and Output Areas in London. DVD 7. ONS:London.

ONS. 2003e. Census 2001: review and evaluation: Part 1 –Question non-response rates. http://www.statistics.gov.uk/census2001/dataqualityevrep.asp [accessed13 May 2003].

ONS. 2003f. National Statistics 2003. Edit and Imputation– Evaluation Report. http://www.statistics.gov.uk/census2001/editimputevrep.asp [accessed 9 Sep-tember 2004].

ONS. 2004. National Statistics 2004. Census 2001 Defini-tions SE/2004/106. http://www.statistics.gov.uk/StatBase/Product.asp?vlnk=12951.

Pereira R. 2002. The Census Coverage Survey – Thekey element of a One Number Census. PopulationTrends 108: 16–31.

Schmool M. 1998. The ethnic question on the BritishCensus: a Jewish perspective. Patterns of Prejudice32(2): 65–71.

Schmool M. 2004. British Jewry in 2001: first impres-sions from the censuses. In The Jewish Year Book 2004,Massil SW (ed.). Vallentine Mitchell: London;xx–xxxi.

Schmool M, Cohen F. 1998. A Profile of British Jewry: Patterns and Trends at the Turn of the Century. Boardof Deputies of British Jews: London.

Schmool M, Cohen F. 2002. British Synagogue Member-ship in 2001. Board of Deputies of British Jews:London.

Sillitoe K, White PH. 1992. Ethnic Group and theBritish Census: the search for a question. Journal ofthe Royal Statistical Society, Series A: Statistics inSociety 155: 141–163.

Simpson L. 2001. Editorial: UK census 2001 – progressand promise. Journal of the Royal Statistical Series A:Statistics in Society 164(3): 423–425.

Southworth J. 2001. Extracts from: The changing natureof the Census? Counting race, ethnicity and religion inBritain and America. PhD thesis, University of Bristol.http://www.gees.bham.ac.uk/downloads/gesdraftpapers/josouthworththesisextracts.pdf[accessed 3 February 2004].

Stark R. 2002. Physiology and faith: addressing the‘universal’ gender difference in religious commit-ment. Journal for the Scientific Study of Religion 41(3):495–507.

Statistics Canada. 2003a. 2001 Census: Analysis Series: Canada’s Ethnocultural Portrait: The ChangingMosaic. Ministry of Industry: Ottawa. Cat No:96F0030XIE2001008.

Statistics Canada. 2003b. 2001 Census: Population bySelected Ethnic Origins. http://www.statcan.ca/english/Pgdb/demo28a.htm [accessed 14 October2004].

Statistics Canada. 2003c. 2001 Census: Selected religions,provinces and territories. http://www.statcan.ca/english/Pgdb/demo30a.htm [accessed 14 October2004].

Thomson K, Waterman S. 2003. London and South EastJewish Community Study Technical Report. NatCen:London.

Valins O. 2003. Stubborn identities and the construc-tion of socio-spatial boundaries: ultra-orthodox Jewsliving in contemporary Britain. Transactions of theInstitute of British Geographers NS28: 158–175.

Waterman S. 1986. The distribution of Jews in theUnited Kingdom. Geography 71: 60–65.

Waterman S. 2003. The Jews of Leeds in 2001: Portrait ofa Community. Report 4. JPR: London.

UK Jewish Population 101

Copyright © 2005 John Wiley & Sons, Ltd. Popul. Space Place 11, 89–102 (2005)

Waterman S, Kosmin B. 1986a. Mapping an unenu-merated ethnic population: Jews in London. Ethnicand Racial Studies 9: 484–501.

Waterman S, Kosmin B. 1986b. British Jewry in the Eight-ies: A Statistical and Geographical Study. Board ofDeputies of British Jews: London.

Weller P. 2004. Identity, politics and the future(s) ofreligion in the UK: the case of religion questions inthe 2001 decennial Census. Journal of ContemporaryReligion 19(1): 3–21.

Weller P, Andrews A. 1998. Counting Religion: Religion,Statistics and the 2001 Census. World FaithsEncounter, 21 November 1998, 23–34. http://www.multifaithnet.org/images/content/seminarpapers/CountingReligionReligion,Statisticsandthe2001Census.htm [accessed 15 September 2004].

Zellick G. 1998. Letters to the Editor. The Times 16October 1998.

102 D. Graham and S. Waterman

Copyright © 2005 John Wiley & Sons, Ltd. Popul. Space Place 11, 89–102 (2005)