Food Deserts from the Consumer Perspective

Vol. 149, No. 1 · Research article

‘Real’ versus ‘mental’ food deserts from the consumer perspective – DIE ERDE concepts and quantitative methods Journal of the Geographical Society applied to rural areas of Germany of Berlin

Ulrich Jürgens Department of Geography, University of Kiel, Ludewig-Meyn-Str. 14, 24118 Kiel, Germany, [email protected]

Manuscript submitted: 1 April 2017 / Accepted for publication: 25 June 2017 / Published online: 30 March 2018

Abstract Developments in food retail in Germany have for decades tended to lead to ever larger retail units, the filling of these units with ever broader and deeper product ranges, and an increasingly oligopolistic market dominated by chain stores. However, as the large chain stores only choose the ‘best possible’ sites according to population density, absolute purchasing power and transport networks, there has been a dramatic thinning out of food-retail facilities in large, particularly rural areas. Has this made it possible to detect supply gaps or, more polemically expressed, food deserts? The term ‘food deserts’, in particular, has achieved a certain amount of acclaim in the Anglo-American context since the 2000s. However, the concept has neither been transferred to nor empirically verified for the German context. In this paper quantitative and qualitative methods are applied to investigate the situation in the rural regions of the most northerly state of Germany (Schleswig-Holstein), in order that food deserts no longer be understood only as ‘real’, tangible and bounded patterns arising from the thinning out of infrastructure but rather as cognitive, perceived patterns (mental food deserts). It is suggested that customer (groups) have long-term and varied shopping predispositions so that diverse groups no longer perceive the loss of supply options and actually create local supply gaps for others through their shopping behaviour. Cluster, discriminant and network analyses are used to complement an inventory of tangible retail facilities with ‘lived’ mental attitudes and shopping behaviour, distinctly broadening the present understanding of food deserts.

Zusammenfassung Die Entwicklungen im deutschen Lebensmitteleinzelhandel hin zu tendenziell immer noch größer werdenden Ladeneinheiten, die Ausfüllung dieser Einheiten mit immer breiteren und tieferen Sortimenten und die fort- schreitende Oligopolisierung unter den Filialketten schreiten seit Jahrzehnten voran. Weil sich aber die Großket- ten nur die „bestmöglichen“ Standorte nach Bevölkerungsdichte, absoluter Kaufkraft und verkehrlicher Logistik Vorsichern, allem dünnen der Begriff großflächige, der food vor deserts allem hat ländliche seit den Territorien 2000er Jahren in ihrer für Lebensmittelversorgung den angloamerikanischen dramatisch Kontext eineaus. Lassen sich hierdurch Versorgungslücken oder, semantisch verschärft, gar Versorgungswüsten konstatieren?- - gewisse Berühmtheit erlangt. Ein Transfer von Konzeption und empirischem Nachweisfood deserts auf dienicht deutsche mehr nur Situa als tion steht noch aus. Anhand quantitativer und qualitativer Methoden, die auf ländliche Regionen im nördlichs ten Bundesland Deutschlands, Schleswig-Holstein, Anwendung finden, sollen

Ulrich Jürgens - 149 (1): 25-43 2018: ‘Real’ versus ‘mental’ food deserts from the consumer perspective – concepts and quantitative meth ods applied to rural areas of Germany. – DIE ERDE DOI: 10.12854/erde-149-56

DIE ERDE · Vol. 149 · 1/2018 25 ‘Real’ versus ‘mental’ food deserts from the consumer perspective – concepts and quantitative methods applied to rural areas of Germany

„reale“ und abgrenzbare Muster infrastruktureller Ausdünnung, sondern als kognitive, wahrgenommene Aus- dünnungsmuster verstanden werden (mentale food deserts). Unterstellt wird, dass Kunden(gruppen) in ihrem

Einkaufsverhalten langfristig und unterschiedlich prädisponiert sind, sodass diverse Gruppen den Verlust an AusstattungsinventareVersorgungsangeboten mitnicht „gelebten“ mehr wahrnehmen mentalen Einstellungen und durch ihr und Einkaufsverhalten Einkaufsverhalten Nahversorgungslücken zu spiegeln und das bisfür- andere erst schaffen. Hierfürfood deserts werden deutlich Cluster-, zu erweiternDiskriminanz-. und Netzwerkanalysen durchgeführt, um reale herige Verständnis von Keywords food deserts, local shopping, food retail, cognition, network, rural areas, Germany

1. Introduction compatible. Is the focus on areas where there are no

diversity of providers is limited to one well-known interdisciplinary, involving subjects such as a) busi- more shops due to closures? Or on regions where the Theness academicadministration, field of concerned retail research with marketing is particularly and - ofsupermarket? people have veryOr is differentthe range subjective of products perceptions within one of retail system too limited or ‘unhealthy’? Or do groups the internal processes of firms; b) politics and plan ecotrophology,ning, which influence which focus the expansion on the composition and locational and the thinning out of retail facilities? healthinessdecisions of ofbusinesses; food products c) food and science, through medicine their speand- following paper. First, the state of research is re- cialist knowledge positively or negatively impact the viewed,These deliberations tracing the different determine perspectives the structure taken of theon duct marketing and market research, concerned with market research is presented, widening the present sales results of individual shops; d) psychology, pro- food deserts. A concept from the field of psychological research, which takes a spatial perspective on vary- description of the methodology and implementation ingcustomer scales (fromtypes theand micro attitudes; to the e) macro geographical level) and retail thus ofunderstanding an empirical ofsurvey food undertakendeserts. This by is the followed author byand a focuses not on individual shops or products but on identifying spatial-temporal patterns of retail owner- - ship and analysing their consequences, considering sionthe deconstruction and discussion ofof ‘real’future food developments deserts (via focus GIS) onto the diverse actors involved. In comparison with other mental food deserts and cognitive maps. The conclu disciplines, geographical retail research is particularly distinguished by its systematic consideration of possible applications of ‘mental’ food deserts. mutually determining groups of actors in their inter- 2. State of research acting (spatial) networks, subjecting them to diverse methodological approaches such as GIS, statistical - considerable diversity in current research on food ofStrand so-called 1 – The healthy health and discourse: nonetheless Food affordabledeserts are food pri supplies,investigation convenience and network shopping analysis. and the There thinning is clearly out formarily residents identified in the where catchment there is area no sufficient of a shop, supply usu- of retail facilities that results from processes of con- - centration on the supply side and the focus on private metres (Leete - characterisedally normatively by defineda varied in supply terms of of fruit minutes and orvegeta kilo- tive taken by geographical retail research competes bles (Farley et al. 2012: 207). Healthy food is, e.g., withcars andthe mobilityperspectives on the of customer many other side. disciplines The perspec for whether potential customers can also afford, know dominance of the discourse about the most appro- about or prepareet al. such2009), supplies although (Pettygrove it remains and unclear Ghose priate assessment criteria and approaches to local Wright food retail, of relevance both now and in the future. US-American context, the methodology of the Chicago It follows that over the last two decades the concepts, 2016:School 271; of Sociology et al. has been2016: used 176). to Especially analyse big in data the - lation structures (ethnicity, social status, automo- desertsdefinitions, have operationalisation, been diverse and appliednot always methods mutually and bility,from censuses,age, etc.), findingand (disadvantageous) correlations between food supply popu empirical findings concerned with so-called food 26 DIE ERDE · Vol. 149 · 1/2018 ‘Real’ versus ‘mental’ food deserts from the consumer perspective – concepts and quantitative methods applied to rural areas of Germany

(Thibodeaux 2016). Such correlations may then be produces spatial models, potential studies (potential manifested in striking patterns of disease such as Eckert and Shetty 2011: 1222) and catchment obesity and diabetes (Morland Moore and Roux 2006). Food deserts have been declared dysto- beaccess; scrutinised and elaborated using empirical data from pian spaces (especially in urban et al. areas 2006; and inner cit- area scenarios for individual groups; these then need to- ies) in which people are not starving but are rather outside the GIS-cosmos; 3. The GIS literature itself iden- cesstifies of types potential of accessibility customers that to a canshop. no Subjective longer be simplyspaces leading ‘unhealthy’ and, at worst, through personal equated with the (quantitative ‘objective’) physical ac negligence, ‘undisciplined’ lives. spaces as mental availability (Goodman and Remaud analyses: Spatial representations of food deserts cre- 2015:of perception 118) and and push information these GIS overlaprepresentations these ‘objective’ of food atedStrand using 2 – computerised GIS (Geographical systems Information and based Systems)on retail deserts to their limits. occupancy and accessibility networks have repeat- edly triggered questions about whether more realistic GIS models could not be created, with the aim of - breaking down the container-like and spatially and Strand 3 – Discussion of concepts: More recent work- temporally static image of food deserts (Widener riescontradicts of which the demographic original definition criteria of food can desertsbe correlated as cov 2013: 1). It has been suggested that: withering the fixed accessibility ‘pre-existing and areal levels units’ of use within of food the retail. bounda Ac- et al. ceptance is growing for the notion that accessibility rep- a) All shops may not be equally weighted by customers in terms of perception, recognition, acceptance, Choi and Suzuki image, diversity of supply and accessibility (physi- resents only a ‘measure of supply’ and ‘not a descriptor - locationalof behavior’ accessibility. of customers Choi (quoted and Suzuki from (2013: 88) de- propriate access) and may indeed not all actually 2013: 87); individual accessibility can thus differ from representcal, financial, food-supply informational options access, (Eckert culturally and Shetty ap Chen and Clark fine individual accessibility as the ‘proximity or number- b) Spatial accessibility may vary over time (“deserts cationof opportunities of food deserts that one as perceives compact onand a personalhomogeneous level’. 2011: 1216; 2013:Chen 82); and Clark 2013: The focus is thus no longer on the area-based identifi- 84), whether this refers to opening times set by the ual level of groups of people with different propensities supplierappear in or the to periodthe travel of a timesday”; and mobility of vari- andspaces. problems Rather relatedattention to hasfood also shopping, turned whoto the thus individ have ous customer groups on a daily or even life-cycle scale (Kestens Chen and Clark some of which overlap (Ver Ploeg theirfew investigations own and group-specific have thus far food-desert been able to experiences, integrate a c) Food-shopping et al.trips 2010:may not 1095; necessarily always et al. 2015: 206). Very- start2016: at 176); home and may take the form of chain trips Widener (Widener temporaland Shannon component to reflect the changeability of ‘spa or food deserts are located not only in local resi- tial’ or ‘mental’ food deserts (“atemporalSadler data”; dential environments et al. 2013: 2) but so inthat the food much environments more com- secondly to discover 2014: 1), whether firstly to intervention prove that they in retail actually oc- plex activity spaces of each individual (Kestens cupancyexist (“fabricate can trigger food new deserts”; patterns of shoppinget al. 2016: and 445), satisfaction. d) Food deserts cannot be represented as bounded et al. 2010); - els of personal disadvantages (age, mobility, travel few exceptions (Jürgens Augustin Sperk and and ‘absolute’ spatial islands. Due to different lev KistemannStrand 4 – The2012), German neither discourse the term onnor local the concept supply: ofWith food a levels of relative hardship in accessing food sourc- deserts has thus far had a2016; lasting impact 2014;on German-lan - es”speed, (Russell finances, and Heidkamp etc.) they 2011: rather 1201). reflect “different guage retail research. At the same time, there has been increased academic interest in local food supplies in dis- considerations: 1. With their complexity and sup- larger cities or popular tourist districts but areas threat- posedThree accuracy, fundamental GIS maps challenges make a strong emerge visual from impact these enedtinctly by rural demographic areas – not trends the so-called leading commuter-beltsto the decline and of and can develop a powerful momentum of their own, aging of the population (Warburg Zibell only IEL Helmle and Kuczera Hahne 2016). 2011; et al. 2015; particularlyDIE ERDE · Vol. with 149 planners· 1/2018 and politicians; 2. GIS 2015; 2015;

27 ‘Real’ versus ‘mental’ food deserts from the consumer perspective – concepts and quantitative methods applied to rural areas of Germany

Food retail is seen (from the planning point of view) 3. Extending the food desert discussion and in- as an anchor for other service-oriented institutions, vestigation and as long as local grocery stores are in operation it It is clear that the research strands presented above (Zibell Helmle isand said Kuczera that the settlement’sKüpper and ‘heart’ Scheibe continues 2015: 49) to beatcast (technical) methods to discuss the nature of food de- et al. 2015: 149). Long-term studies ( utilise very different perspectives, definitions, and the disappearance2015; of food retail is linked to subjec- doubt on this finding: it is not possible to prove that- serts. They have in common that they focus on the- tures (Hahne ‘real’ food desert in its spatial extent, facilities, and- tiveof satisfaction satisfaction among levels. the Residents population adjust are tomuch the betterstruc gibleaccessibility. food deserts The deeper-lying, have not yet been‘hidden’ embraced mental bystruc the than the objective2016: 177)conditions. or accept Steinröx them, so(2013: that levels166f) well-knowntures that influence literature reality on food and deserts. exist behind However, the othtan- believes that individual residents ignore the danger er disciplines such as environmental psychology have that the diminishing attractiveness of the focal point already adopted such subject matter. of everyday life may accelerate population decline, - travel to neighbouring towns for other public and pri- fering levels of knowledge and different motives vateespecially services. as people find it increasingly necessary to peopleThe psychological perceive their approach environment assumes in thatdifferent due to ways dif and store it as a context-dependent or -independent - Baaser and Zehner 2014) as ries or units of meaning (Swoboda 1996: 320), seman- Muchpotential research food crisis thus regions identifies although, rural municipalitiesin contrast to ticmemory. knowledge This canthat takeis responsible the form ofmore abstract for a categoconsis- (andthe Anglo-American also urban districts; discussion, attention is directed more towards the existence of supplies than to the knowledge is stored in the long-term memory, becom- ingtency part of attitudeof the short-term than for concrete memory satisfaction. after activation This to the world of discount stores and supermarkets are (Holzmann and Wührer 2000: 431), and can thus be ‘healthiness’ of those supplies. Best-case alternatives- involved in relevant mental processes of perception rope (Schaloske Küpper and Tautz Schenk and memory, e.g. in a shop (Swoboda thus2016). debated Quantitative – also inaccessibility comparison tostudies the rest (Segerer of Eu literature discusses semantic networks consisting of 2014 Neumeier 2015)2013; on the one hand, and2015; strategy- a multitude of propositions that each1996: emerge 324). from The a oriented governance analyses of civic involvement and - private; sector initiatives to safeguard local food retail ternal or internal stimuli, a person becomes individu- on the other hand (Warburg Schaloske 2013), combinationally aware (again) of ‘argument’ of these and arguments ‘relation’. and Through reacts ex to feature prominently among recent analyses of ‘qual- them mentally or with concrete action. As this tends 2011; of the extent to which structural and constructional Maggioni 2016: 121), the semantic networks changesity’ local supplyin retail and also village impact shops. on Long-termthe behaviour analyses and ofto involveindividuals stable can ‘recallable be used toinformation predict their in consumers’ behaviour (minds’Holzmann ( and Wührer Seitz 2015). rare in German-speaking geographical retail research (‘basicMeyer emotionalMonheim attitudes’ and of Heller customers 2016) areas considera to date as- Cognitive psychology argues 2000: 431f.;that human memory is tion of individual accessibility, which goes beyond made up of a network of associations that encompass- 2005; es, on the one hand, criteria and, on the other hand, spatial and temporal dynamics of supply systems and, relations between criteria and their relative weight- onlocational the customer accessibility side, of measuredstructures by of travelperceptions time. Theand - ual or aggregated level to be derived (Wühler 2001: deserts are ‘not simply based on proximity and acces- ings; this then allows cognitive maps for the individ use can demonstrateDel Casino 2015: that 801) ‘foodscapes’ but rather or evencomprise food complex social interactions between diverse actor knowledge777). These capacities ‘maps’ are and based possibilities on a) existing for new positive learn- groups.sibility’ ( anding or negative the reinterpretation experiences; of b) old individual knowledge processing, and experience and c) imaginings or expectations that may be stored independently of experience (Swoboda 1996:

28 322f.). This gives rise to comprehensiveDIE ERDE · Vol. ‘knowledge149 · 1/2018 ‘Real’ versus ‘mental’ food deserts from the consumer perspective – concepts and quantitative methods applied to rural areas of Germany

extent unconsciously in human memory and that, e.g., as long as the premises show no subsequent use, at structures’ that exist both consciously and to a large- and thus ‘closed premises’ is only held in the system - - marketinglisation of network planning analysis aims to or expose associative and then networks influ minedwhich pointfor all it premises is overwritten using Google in the MapsNielsen and Company entered ence through external communication tools. The uti indatabank. the databank. The GIS In coordinates addition, all were bakeries manually (in particu deter- takes the form, e.g., of investigations into customer loyaltyis relatively or brand new marketing in the field (Hellmann of retail and research Marschall and andlar their the web branch pages networks) of the bakery were chains,identified as bakersthrough (un an- likeinternet butchers) search provide of the bakers’ an alternative guild, the food-retail Yellow Pages net- In2010: the 647).literature to date there is no indication that this work thanks to their continued high numbers, spatial approach has ever been used for the investigation of convenience and additional food offerings.

GIS to analyse the development of retail facilities in tangiblefood deserts. areas Thisbut also paper investigates therefore ‘mental not only represen applies- stationary food supply networks and accessibility Teichert and Schöntag gapsThis that approach is more aims realistic to provide than that a representationachieved by many of Anglo-American analyses, which are limited by a fre- networkstations’ ( of associations between 2010: retail 370) formats of supply and placesareas fromto be theuncovered consumer and perspective.thus patterns This of experi allows- with food departments. GIS analyses using structural ences, knowledge, uses, alternatives and features of dataquent are fixation thus an on initialsupermarkets technical and approach department towards stores a) - identifying areal patterns of the distribution, concen- gous to Holzmann and Wührer 2000: 438). local supply to be identified as problematic (analo using a spatial approach to identify the quantitative relevancetration and of thinning individual out patterns of stationary as the food generalisable retail; b) gaps seen as being problematic (Küpper and Scheibe Are supply gaps perceived by consumers? Are these accessibility of food supplies for various customer groupsvariables (of of differing many individual age and social cases; mobility) c) modelling by using the (2015:Broadbridge 49)? Are and peopleCalderwood interested in options that changeable speeds and transport modes (car, bicycle, could eradicate food deserts? Who uses a village shop- ferings beyond the world of supermarkets 2002)? What and types dis ofof food retail according to nameless and supposedly counters,customers what identify types themselves identify themselves with ‘alternative’ only with of pedestrian); d) no longer defining the accessibility Company dataset makes it possible to reconstruct the the center of the following discussion. entireinterchangeable spatial networks retail premises. of the supermarkets Rather, the and Nielsen dis- the dominant discount store? These questions are at count stores of various operators, and thus to identify - 4. Methods and locations scalesthe various because supply it is optionspossible (‘not to zoom just inany on shop’) large-scale avail - caseable studiesto customers; from the e) small-scale modelling areathe different of an entire spatial fed- ented perspective. Use is made of a mix of quantita- eral state. The investigation thus focuses on the demand-ori for the entire area of the federal state of Schleswig- tive and qualitative methods, applied in 2016 firstly- thinning out of food retail in rural regions to be de- The latter possibility allows detailed patterns of the usedHolstein areal (SH), data and on secondly the spatial for fivedistribution selected ruralof super mu- level are required to show whether these supply gaps markets,nicipalities. food The discount GIS investigations stores and with smaller ESRI-ArcGIS grocery arepicted. perceived Social-empirical as problematic, methods which on the alternative ‘lived’ micro- food stores (sales area under 400 m²), provided by Nielsen retail options elsewhere are used, the nature of local Company (2015) in the form of a professionally ma- discussions about these supply gaps, and whether local supply alternatives are commercially successful. operating names of the supermarkets and discount naged databank. This dataset includes the various- In 2016, quantitative empirical investigations were carried out in selected rural regions of Schleswig- stores and also the outlets’ sales area, year of open Holstein with the aim of improving the understanding ing and status (‘active’ or ‘non-active’). The data are notDIE ERDEcomplete · Vol. for 149 all · 1/2018 premises. The status ‘non-active’ 29 ‘Real’ versus ‘mental’ food deserts from the consumer perspective – concepts and quantitative methods applied to rural areas of Germany of the food desert phenomenon in the German context supermarkets or discount stores as alternatives to and advancing discussions on ways of operationalis- village retail options, and in their past experience ing food deserts. Standardised questionnaires were with village shops. As quantitative-statistical inves- distributed to all households in the case-study muni- tigations (at a given probability) on the micro-level of individual municipalities require a negligible non- all municipalities was 15.4% (385 questionnaires). cipalities via postal delivery. The response rate over with reference to all the case studies as a rural type involvedresponse in rate, a discourse the following of decline findings and out-migration. are discussed acceptanceThe survey wasof the conducted questionnaire with theand, agreement secondly, ofto theal- lowmayors anonymous or village locations shops, firstly for responses to increase to residents’be estab- generalised patterns of perceptions concerning food lished. In three cases, the questionnaire was combined This approach allows the full statistical spectrum of with a prize-winning competition. However, response (over 49 years of age) were over-represented among rates were similar to those in the two municipalities supply to be identified. Women and older individuals without the competition. Further information on the the household members mainly responsible for food structure of the questionnaire, the locations and the shoppingthe responses. and are This also clearly viewed makes by the sense, public as discourse they are technical implementation is found in Tables 1 and 2. as particularly important customers for village shops. - A political planning debate concerning local sup- tion. However, these variations are the same for all the ply facilities is currently being conducted at all loca- The analysis is thus not based on a normal distribu using non-parametric procedures. the topic among local experts and stakeholders, on case-study municipalities. Significances were tested thetions, one so hand, there andis likely among to belocal sufficient residents, sensitivity on the other for Qualitative research took the form of focus group discussions and interviews with mayors and the owners micro-contexts, i.e. in terms of size, age distribution, of shops. However, this is not further considered in proximityhand. Naturally, to rural thetowns municipalities with clusters differ of well-known in their this paper.

Section Topic Indicators Table 1 A Self-assessment as customer 40 statements on the importance of food formats, Structure of the ques- supplies and structures (Likert scale 1-5) tionnaire and technical B Food-retail structures Shopping locations, frequency of shopping implementation information. Source: own C Self-assessment as actual or potential customer 12 statements on the relevance of use (Likert scale 1-5) elaboration D Pro le of the interviewed individual and Socio-economic and socio-demographic indicators household data Locations Rural municipalities SH (Schleswig-Holstein) Sample size n=385 (April-November 2016) Procedure Postal questionnaire (2016) (standardised)

Municipality Number of Response Basic interest Date Type Table 2 postal questionnaires n (%) of location Scope and locations of distributed the quantitative sur- (n) veys conducted in 2016. Groß Vollstedt 430 64 (14.9) Saving the village shop discourse 2016 periphery Source: Jürgens (postal Holzbunge 150 55 (36.7) Threat of closure for lling station shop 2016 periphery questionnaire) 2016 Borgstedt 650 109 (16.8) New opening discourse (spring 2017) 2016 peri-urban Hollingstedt 370 61 (16.5) Independent village shop 2016 periphery Kirchbarkau/ 900 96 (10.7) New opening (October 2016) 2016 periphery Barkauer Land n total 2.500 385 (15,4)

30 DIE ERDE · Vol. 149 · 1/2018 ‘Real’ versus ‘mental’ food deserts from the consumer perspective – concepts and quantitative methods applied to rural areas of Germany

5. Finding I – GIS worlds and food deserts of these municipalities is therefore largely dependent on shopping in the rural central places, where stra- An initial depiction of food retail in a larger region is tegically planned clusters consisting of supermarket shown in Figure 1. Here supermarkets, food discount stores are mapped for the federal state of Schleswig- problemsand discount caused store by are the often large found number on greenfield of closures sites of Holstein,stores, active and their and distribution – in 2015 –underlain non-active with grocery a rep- so-calledon the outskirts. smaller The stores rural (sales municipalities area under face 400 further m²) resentation of accessibility by car showing the areas seen in recent years (Fig. 2). As of 2015, this had re- from which supermarkets and discount stores can be sulted in 802 of about 1,100 municipalities having reached in under ten minutes. Accessibility was calcu- not a single grocery store (not taking other suppliers - istically based on the various types of streets drawn fromlated usingOpenStreetMap the network (motorways, module of ESRI-ArcGIS,main roads, localreal such as bakeries or filling station shops into account).- This affects about every fifth resident of the federal- uneven distribution of supermarkets and discount state as a whole. 1,917 active retail premises (includ roads) and the average speeds defined for them. The these,ing filling 142 stationsmunicipalities and other have units only larger one point than ofsuper sale markets) are concentrated in Figureonly 291 3 settlements;clearly shows of andstores the reflects densely the developed settlement commuter structure belt ofthat the adjoins state, that many municipalities have no retail facilities or Hamburgwith only to five the larger south. cities of over 75,000 residents (includinghave lost their filling only stations). shop and thus acquired the status

Large areas of the state are of rural character with network of local retail supply very uneven, but also very low population densities (Fig. 2 thatof ‘no it localleaves supply’. much of It the is obviouspopulation that with not no only choice is the of are correspondingly uninteresting for the operators supplier in their local proximity without resorting to ). These regions longer car trips. of supermarkets and discount stores. The population

Stores grocery stores <400m² grocery stores <400m² - inactive supermarkets discount stores

Service Areas 10 Minutes

0525 0 Kilometers

Fig. 1 Service areas of supermarkets (sales area 400+ m2) and food discount stores by car accessibility in minutes in the province of Schleswig-Holstein, 2015; unconsidered are service areas from bordering provinces. Source: Own elaboration based on Nielsen Company (2015); ESRI-ArcGIS, OpenStreetMap; Landesvermessungsamt Schleswig-Holstein (2015); Cartography: Fuhrmann DIE ERDE · Vol. 149 · 1/2018 31 ‘Real’ versus ‘mental’ food deserts from the consumer perspective – concepts and quantitative methods applied to rural areas of Germany

grocery stores [m²] grocery stores <400 -active grocery stores <400 -inactive grocery stores >400 -active discount stores -active

population density [persons/km²] < 50 50 - 100 > 100 no data

0525 0 Kilometers

Fig. 2 Spatial distribution of grocery stores in municipalities with low population densities in the province of Schleswig-Holstein, 2015. Source: Own elaboration based on Nielsen Company (2015); Statistical Office SH (2011); ESRI-ArcGIS, Cartography: Johst

municipals containing grocery stores municipals without grocery stores municipals without grocery stor es - inactive stores <400m²

0525 0 Kilometers

Fig. 3 Spatial distribution of municipalities without grocery stores in the province of Schleswig-Holstein, 2015. Source: Own elaboration based on Nielsen Company (2015); ESRI-ArcGIS, Cartography: Johst 32 DIE ERDE · Vol. 149 · 1/2018 ‘Real’ versus ‘mental’ food deserts from the consumer perspective – concepts and quantitative methods applied to rural areas of Germany

6. Finding II – (quantitative) customer worlds and the case study of Groß Vollstedt (Fig. 4), demonstra- mental food deserts? Theting supplythat the situation potential is presentedof GIS networks at a larger goes scale beyond for In order to collect a comprehensive self-assessment of food customers, the survey participants were gi- of‘container-like’ supermarkets portrayals and discount based stores on showsmunicipal where boun- the ven 40 statements to grade on a gradual scale from lengthdaries. Theof a temporalone-way variationcar trip for of thefood catchment shopping areas (e.g. - spontaneous, short-term and convenience shopping) ments concerned the following areas: a) use of dif- may well reach the limits of tolerability for customers. 1 (= not important) to 5 (= very important). The state - site and a shopping cluster in the rural central place vice,fering and retail product formats; range. b) Drawinginterest onin food;Shaw c)(2006), forms the of With the exception of a supermarket on a greenfield aimmobility here andwas accessibility; to cover three d) basicawareness concepts: of price, attitude ser (what is important to me as a customer), household of Nortorf, there is no store south of Westensee – with assets (e.g. car availability, computer use, purchasing maya single seem exception to suggest in Groß that itVollstedt. has good The chances strategically of sur- power), and ability to shop (e.g. availability of a retail positive location of this shop in relation to Nortorf - ongoing discussion around saving the store, which re- rised according to whether they can strengthen or vival as a convenience store. There is, however, an weakenformat, flexiblelocal retail opening supplies. times). In the All further items areanalysis catego all to locational accessibility, subjective criteria must be takenveals theinto limits consideration of ‘objective’ when GIS judging analysis. the In success addition or undertaken allow the model to be multiply adjusted: otherwise of local retail. items are initially equally weighted. The constraints

Quarnbek # Melsdorf Flensburg Schacht-Audorf Felde #Kiel Osterrönfeld Heid# e Westensee

Emkendorf Lübeck#

Groß Vollstedt 050 100 Kilometers

grocery stores < 400 qm - active road system Bordesholm settlement areas Nortorf waterbodies

Wattenbek Service areas supermarkets & discount stores up to 5 minutes > 5 up to 10 minutes 0510 Kilometers

Fig. 4 Food supplies for the municipality of Groß Vollstedt, province of Schleswig-Holstein, in the service area of supermarkets and food discount stores by car accessibility in minutes, 2015. Source: Own elaboration based on Nielsen Company (2015); Landesvermessungsamt Schleswig-Holstein (2015); ESRI-ArcGIS, Cartography: Fuhrmann

DIE ERDE · Vol. 149 · 1/2018 33 ‘Real’ versus ‘mental’ food deserts from the consumer perspective – concepts and quantitative methods applied to rural areas of Germany

- terns. All other items have only diffuse explanatory a) by altering the number of items and ‘correct’ con tents; b) by assigning these items to the constructs Figurepower. 5 The visualises high level the extent of significance of cluster separation of the cluster us- asattitude, to whether asset items or ability; can strengthen c) by aggregating or weaken the supply items divisioning the discriminant is confirmed function. by the useTable of 4 Wilks’ shows Lambda. the dif- to other basic constructs; d) by varying judgements ferentiation of the clusters according to the means for importance for all survey participants. gaps; e) by questioning whether all items are of equal compared to the means. - each of the items and the (ANOVA) significance values nant analysis was used (with the help of the software Canonical Discriminant Functions Taking note of these constraints, cluster and discrimi- entiate the dataset not only by item contents but also 5.0 bySPSS23) survey to participants develop a model (Table to 3 compress and differ into account, are there individuals in the total sample who are more similar to one ).another Taking than all 40 to itemsother survey participants and who, thus, form a noticeable 2.5 2 randomly and a discriminant analysis used to deter- 2 minecluster? which The numberitems are of clustersrelevant is for initially maximising calculated be- 0.0 3 tween-cluster differences and the explanatory value 1 Function

assignment of the number of clusters and the relevant -2.5 of these items for the selected number of clusters. The cluster that covers age, gender, place of origin, etc. (in- items for each case results in a specific profile for each- -5.0 formation gathered in the questionnaire). These pro file groups can be used for policy and planning and retail formats), because they allow the fundamental -5.0 -2.5 0.0 2.5 5.0 for specific marketing strategies (e.g. for different openness of survey participants towards the various Function 1 retail formats and the strengths and weaknesses of Groups those formats to be deduced. In the present case three 1 Group Centroids 2 clusters were formed. 19 of 40 items had the highest 3 explanatory value here and could explain 91% of the Fig. 5 Separation of cases using 19 items and discriminant - analysis. Source: Data Jürgens 2016; SPSS23 lows us to look within the clusters and discover pat- content of the clusters. The discriminant analysis al

Methods Implementation Table 3 Technical information on Cluster analysis K-means-clustering of 40 items; pairwise deletion; number of clusters = 4 the cluster and discrimi- Cluster 1 = 148 cases Cluster 2 = 109 cases nant analyses. Source: Cluster 3 = 124 cases Data Jürgens 2016; Cluster 4 = 4 cases SPSS23 Cluster 4 is disregarded so 3 clusters and 381 cases are included Discriminant analysis Question: which items inuence cases within the clusters and are decisive for correct cluster allocation? Wilks’ method; cases included = 381 Used for classi cation = 318 cases 19 of 40 items were included in the analysis 91.2% of the originally grouped cases were classi ed correctly: Groups 1-3 = 118/96/104 cases Discriminant function 1 explains 67.1% of the group separation Discriminant function 2 explains 32.9% of the group separation Wilks’ Lambda (Test1-2) .142 (Signi cance .000) Wilks’ Lambda (Test2) .469 (Signi cance .000) 34 DIE ERDE · Vol. 149 · 1/2018 ‘Real’ versus ‘mental’ food deserts from the consumer perspective – concepts and quantitative methods applied to rural areas of Germany

Table 4 Items relevant for the discriminant analysis of the clusters (means). Source: Data collection Jürgens 2016, SPSS23. (1) Schendera (2010: 341) points out that univariate significance does not automatically also indicate discriminant relevance, as can be seen here.

Question Statement (1=not important; 5=very important) Cluster 1 Cluster 2 Cluster 3 Single factor + Food desert – favourable criteria Need satis ers Smart Alternative ANOVA - Food desert – countering criteria shoppers shoppers 1 I shop in a discount store + 3.7 4.3 2.7 .000 7 I always buy everything in one shop of my choice 2.3 2.3 1.9 .029 9 I choose the shop that is closest to my place of work + 2.0 3.0 2.3 .000 10 I would like the shop to be easily accessible by car + 4.4 4.1 3.5 .000 11 I would like the shop to be easily accessible on foot - 1.8 3.3 3.7 .000 14 I would like many dierent shops nearby so I can compare products and prices - 2.7 3.3 1.8 .000 15 When shopping for food, prices are especially important to me + 3.0 3.8 2.1 .000 21 Organic products are important to me - 2.4 2.8 3.9 .000 23 I come primarily for the special oers + 2.8 3.4 1.8 .000 24 If I can save a bit again, then especially with food + 2.2 3.1 1.5 .000 25 I want to be able to buy non-groceries too + 2.6 3.0 2.0 .000 26 I also come because of the more convenient parking + 3.6 3.4 2.3 .000 27 I like to make use of the longer opening times + 2.6 3.4 2.6 .000 28 I come because I feel comfortable in the shop - 3.4 4.1 4.0 .000 30 I like to buy food spontaneously + 2.2 3.4 3.3 .000 32 I can enjoy shopping for food - 2.3 3.3 3.5 .000 38 I also use other alternatives like a farm shop - 1.6 2.0 2.7 .000 39 I also use other alternatives like the weekly market - 2.5 2.4 2.6 .444 (1) 40 I also use other alternatives like a delivery service + 1.1 1.1 1.0 .223 (1)

the importance of feeling comfortable and enjoying character,The description because of thethe clusters number is of subjective clusters and and discri is not- shoppingings is relevant (Table for 5). this group. This does not minimise performedminant functions by SPSS. can The only statements relate to this are setmodel-like of data. in Examination of actual shopping behaviour is neces- For Cluster 1 the focus is on planned and joyless shop- - ping at a discount store or supermarket that can be ments of the survey participants, and thus whether - thesary predispositions to ascertain whether within this the reflectsclusters the are self-assess relevant to tant role in the choice of retail facility. Accessibility the market. Are Clusters 1 and 2 really a lost cause as easily reached by car. Parking thus plays an impor far as convenience shopping in a local village shop is Cluster 1. Food shopping is primarily a necessary ac- concerned, and is the importance of Cluster 3 for local tivityon foot rather and thus than local something convenience that involves is insignificant emotional for satisfaction. Cluster 2 describes itself as committed discount shoppers, who are particularly motivated Tableshopping 6 shows as great that as the the attitudes self-assessments from Table suggest? 4 have differing relevance for shopping in local retail facilities, opening times and spontaneity in order to get the and are completely independent of popular segmenta- bestby price bargains. and thusIt is no prioritise contradiction selection, that very these flexible smart tions based on age, gender or income. Surprisingly, car shoppers also want to feel comfortable in the shops. availability per household plays no role here. Institu- Cluster 3 comprises those customers who, from their tional involvement, being a member of a village-shop self-assessment, can be described as committed con- to economise nor an extreme focus on choice of offer- cooperative, is reflected in local shopping behaviour. venience and alternative customers. Neither the need

DIE ERDE · Vol. 149 · 1/2018 35 ‘Real’ versus ‘mental’ food deserts from the consumer perspective – concepts and quantitative methods applied to rural areas of Germany

Cluster value Description Summary Table 5 Cluster 1 Emphasis on car, it follows that accessibility on foot is irrelevant; Highly mobile and Characteristics of the clusters. Source: Data alternatives such as organic or farm shop of little importance; joyless food shoppers collection Jürgens neither spontaneity nor enjoyment of food shopping play a large role. Need satisers 2016, see Table 4 Cluster 2 Committed to discount stores and shopping with the car, focuses on Cheap – mobile – satised prices and special o ers; Smart shoppers product range and feeling comfortable in the shop important; not necessarily concerned with proximity to place of residence. Cluster 3 Local convenience shopper; Nearby – alternative – enjoyment interested in alternative o erings, enjoys shopping, tends not to depend Alternative shoppers on low prices, high enjoyment and satisfaction factor.

Table 6 Actual shopping behaviour in each cluster. Source: Data collection Jürgens 2016, SPSS23 - ersfilling with station their salesshop. behaviour The question cannot as tobe whetheranswered and on Cluster Cluster Cluster Signicance how quickly local shops can influence their custom 1 2 3 test the basis of this statistical dataset. Similarly unclear Make use of alternative local oerings 1.6 2.0 2.5 .000* is how successful a planned shop in Borgstedt can be (mean 1 not important to when the current predisposition of residents seems 5 very important) clearly unfavourable for local convenience offerings. Use no alternative retail formats (in %) 59.5 52.7 17.3 .000* Food shops per week (mean) 2.2 2.5 2.6 .022** Table 7 Spending per week in the village shop 30.67 32.50 52.10 .006** Socio-demographic structures within the clusters in %. (mean in euros). Source: Data collection Jürgens 2016, SPSS23 Only relevant for Kirchbarkau Cars per household (mean) 1.6 1.8 1.6 .350** Cluster 1Cluster 2Cluster 3 Co-operative member in % 35.3 22.2 66.7 .002* (only Kirchbarkau) Length of residence (years) 25.7 19.9 21.8 *Kruskal-Wallis-Test **(ANOVA) Size of household (people) 2.4 2.9 2.5 Descriptive segmentation in line with Table 7 is re- Under 35 6.2 16.1 4.9 quired to clarify the socio-demographic structures 35-49 23.9 37.6 31.1 of the clusters and thus to indicate which customer is 50-64 34.5 32.3 42.7 characterised by which attitudes and type of behav- 65-74 23.9 9.7 13.6 iour. It can be seen that Cluster 1 tends to comprise Over 74 11.5 4.3 7.8 the old-established population with longer lengths Mean household income 2,285 2,200 2,310 of residency, here the older age groups are over-rep- (per month in euros) resented and there is a large proportion of pensio- Over 3,500 euros 35.0 23.7 34.4 ners. Contrary to the notion that such individuals are Up to 2,000 euros 15.1 18.7 11.4 precisely the clientele targeted by convenience shops, Pensioners 35.1 15.6 20.8 they frequently do not use their village shop (68% Employees 36.0 60.0 41.6 Female 52.2 71.0 69.9 for all three clusters). Clusters 2 and 3 are distinctly Borgstedt 51.7 34.8 13.5 of those aged over 74 are in Cluster 1 and 2; N=25 younger. Comparing Clusters 1 and 3 reveals no link Hollingstedt 35.4 33.3 31.3 between higher incomes and interest in alternative Holzbunge 51.1 25.5 23.4 and more expensive products. A comparison between Kirchbarkau 21.3 22.5 56.3 the clusters shows clearer differences between the Groß Vollstedt 25.9 35.2 38.9 individual municipalities in terms of the basic attitudes of those questioned towards alternative offerings and local convenience shops. In Groß Vollstedt, Hollingstedt and Kirchbarkau there is currently a shop, in Borgstedt only a bakery, and in Holzbunge a

36 DIE ERDE · Vol. 149 · 1/2018 ‘Real’ versus ‘mental’ food deserts from the consumer perspective – concepts and quantitative methods applied to rural areas of Germany

7. Finding III – cognitive maps – networks of a network diagram, depicting mental food deserts qualitative knowledge The items for the three clusters can be captured in emergence of actual food deserts due to the lack of Fig. purchasesof different being extents made (Likert (shop scales). closures These following lead todisin the- 6a and 6b - terest on the part of customers) (Fig. 6a and b). Figure tualThe extentshopping to which decisions mental is shownnetworks in ofFigures attitudes 7 and ( 8. 6a visualises criteria that favour the development of ) are reflected in relational networks of ac- food deserts, because based on a focus on the car and ping locations and formats for Clusters 1 and 3, taking on prices they tend to endanger food retail close to They summarise all mentions of regular food shop the place of residence. Figure 6b covers shopping seg- leads to the depiction of networks that are independ- ments and emotions that tend to counter food deserts into account all the municipalities investigated. This outside the supermarket world. the frequency of mentions and their weighted rela- tionsent of to ‘real’ one spaceanother, and drawn GIS-space from andaggregation that represent of the Cluster1 Cluster2 Cluster3 information. A new form of generalised mapping of the behaviour of study participants is created that: Discount store 5 Spontaneity 4 Proximity to work a) demonstrates the ‘main roads and side roads’ or 3 - 2 connections of their food shopping trips; b) pinpoints Convenient with Long opening mes 1 the car shopping‘central places’ locations for shopping;similarly perceivedc) presents as niche dominant, shop 0 ping through a lack of connections or edges; d) pools

Parking Prices important varied or monotone in coloured clusters or cliques; e) visualises shopping options and local alternatives; Savings Special oﬀers importantf) presents to the customers ‘others’, and i.e. missingwhich primarily forms of occur supply. to Cluster1 Cluster2 Cluster3 These networks thus express which providers are - Within easy walking them for spontaneous food shopping. The maps can distance ping behaviour and, secondly, as the cognitive spaces 5 ofbe relevantinterpreted, shopping firstly, associations. as the activity spaces of shop 4 Organic products 3 Farm shop 2 - 1 0 The following explanations are intended to aid inter pretation of the figures. Enjoyment Weekly market a) A distinction is made between nodes and edges or connections.

Feeling comfortable mentions of all ingoing and outgoing edges. As the b) nodesThe size can of be the very nodes small is ordefined large, byvariations the number in size of Fig. 6a and 6b are nonlinear so that the nodes are not to be inter- Mental food deserts based on customer clusters. Fig. 6a preted in causal, quantitative or absolute terms but (above): favourable criteria, Fig. 6b (below): counter- rather in relation to one another. ing criteria in rural municipalities of Schleswig-Hol- - stein, 2016; 1= low importance; 5= highest importance; - Source: Jürgens (2016) c) nections The connections are nonlinearly (edges) weighted show the on complete the basis net of working of all nodes with one another. The con- With very few exceptions, the perception networks quencies (in our case) of the use of the individual of the three clusters neither intersect nor touch one nodes.mentions in the system. These edges reflect the fre instance the attitudes in Cluster 1 that favour food de- sertsanother. are Theysupplemented thus work by in the opposing logical lack directions, of interest for d) clustersThe allocation the nodes of colours and connections allows clusters and thus of nodesdomi- nantand connections mentions in to the be system identified. are Inmore each similar of these to one another than to others. in ‘alternative’ food retail. DIE ERDE · Vol. 149 · 1/2018

37 ‘Real’ versus ‘mental’ food deserts from the consumer perspective – concepts and quantitative methods applied to rural areas of Germany

- - - ily connected to larger central places and conveni- e) The aim is to present a visualisation of complex in c) ence The shopping and interest relevant of Cluster shopping 1 isassociations necessar focusesterrelations and featureswith their in mutual a data-cloud networks and, and secondly, influ ences, in order to, firstly, describe and investigate shopsare primarily that are definedeasily accessible in terms ofon accessibility foot, are men by- end.to address gaps between these focuses. Traditional car. The nearest neighbourhood stores, i.e. village cartography and statistics are insufficient for this (Table 8 - Fig. 7) and tioned five times less by Cluster 1 than by Cluster 3 Fig. 8) shows that: ). The information from Cluster 3 also con- The comparison between need satisfiers ( tersfirms 1 and more 2. sustained interest in quality, product a)‘alternative’ Food deserts customers emerge ( less due to a lack of offerings origin and ‘healthy’ food than that found in Clus than (also) due to the monotony of popular retail - d) The perceptions from the cluster-/discriminant- vealsanalysis that are the todifferences a significant between extent Clusters reflected 1 and in structures. Not only are discount stores or super- 2tangible, on the one differing hand and patterns Cluster of 3 demand.on the other This hand re wheremarkets (expressed like Aldi through and Edeka the (resp.size of Penny,the nodes), Lidl, may be explained by a long-term process of moving butFamila, are theRewe, focus Real of foodor Sky) shopping available for nearlynearly everyall the away from local suppliers, socialisation, and high study participants from Cluster 1. customer loyalty for discount stores, and a lack of - interest in discourse concerning local suppliers or ent with their ability to attract customers and this food shops. b) The monotony of the large suppliers is not inconsist-

non-chainminimises or demand independent for small-scale, retailers is ‘different’underpinned re withtail formats. dedicated The micro-clusters distinctiveness and of these colour specialist assign- ments in Figures 7 and 8, as they show no linking of nodes or edges with other formats or spaces.

Fishshop Russee

Neumünster Fahrdorf Gettorf DM

Schleswig Mildstedt Real Aldi Norma

Wankendorf Kirchbarkau Edeka Felde

Markant Markttreff Kiel Silberstedt Flintbek Preetz Meimersdorf Lütteloden Citti Home_growers Hollingstedt

Sky Bordesholm Famila Plaza Netto Kropp GrossVollstedt Eckernförde

Topkauf Market Bakery Nortorf Fig. 7 Rendsburg Penny Damendorf Tetenhusen Regular food shopping treia Butchers trips - Shopping networks Lidl Husum Jübek for Cluster 1 based on re- Holzbunge Borgstedt tail format and location.

Büdelsdorf Filling_station Dataset Jürgens 2016; Bastian et al. 2009; Cher- Rewe ven 2015

38 DIE ERDE · Vol. 149 · 1/2018 ‘Real’ versus ‘mental’ food deserts from the consumer perspective – concepts and quantitative methods applied to rural areas of Germany

Fleckeby Hollingstedt

Edeka Felde Neumünster AltDuvenstedt Ellingstedt Lütteloden Preetz Silberstedt

Friedrichsholm Market Mildstedt Penny Borgstedt Kropp Bakery Netto Butchers Rossmann Topkauf

Markant Erfde Plaza Aldi Husum

GroßVollstedt Büdelsdorf Nortorf Eckernförde

Citti Kiel DM Lidl Bordesholm Westerrönfeld Flintbek Famila Rewe Real Osterrönfeld Jübek

Filling_station Meimersdorf Fig. 8 Organic-shop Altenholz Regular food shopping Holzbunge Sky trips - Shopping networks Kirchbarkau Rendsburg for Cluster 3 based on re- Schleswig Rondo tail format and location. Home_growers Treia Dataset Jürgens 2016; IhreKette Markttreff Nindorf Brekendorf Bastian et al. 2009; Cher- Großbarkau ven 2015 Gnutz

Table 8 8. Conclusion and prospects Regular food-retail formats for Clusters 1-3 (in %). Source: Data Jürgens 2016, SPSS23; ‘other’ formats at 100% are not included in the table exclusively to GIS analyses, then food deserts can 1. Do food deserts exist? If consideration is confined Cluster 1 Cluster 2 Cluster 3 considered, then it is clear that they contribute in varyingbe ‘calculated’. degrees Ifto potentialthe prevention local customersor creation are of Number of nodes 52 60 60 food deserts. It is especially the case that inhabi- Number of edges 468 396 383 tants create food deserts for others because, due to Mentions discount stores 40.0 41.9 21.9 perceived higher prices and a more limited range Mentions supermarkets 40.6 37.4 38.4 of goods etc., a large proportion of the population Mentions village shops 2.6 4.0 12.0 does not use the village shopping alternatives to Mentions farm shops, organic shops, 0.9 2.8 13.1 weekly markets questioned can be described as ‘rejecters of the vil- Mentions specialist shops 11.8 9.3 8.6 a sufficient extent. A significant number of those (bakery, butcher, sh shop) - Average number of mentions per 4 4.3 3.7 lage shop’. survey participant dents of rural areas will in the future deal with the 2. The question then arises as to how the current resi Total number of mentions 468 (115) 396 (93) 383 (103) supply gaps for which they themselves are respon- (survey participants) aging of the current population. What kind of local marketingsible; this is by especially suppliers important and planning with a authoritiesview to the can help to change the predispositions that disad-

negative impact on local food supplies is that dis- countvantage stores local and providers? supermarkets Another are trend also tendingthat has to a thin out their branch networks in favour of focus- ing on larger stores.

DIE ERDE · Vol. 149 · 1/2018 39 ‘Real’ versus ‘mental’ food deserts from the consumer perspective – concepts and quantitative methods applied to rural areas of Germany

- Acknowledgements sional. As individual factors, aspects such as mobil- 3. Theity or discussion age play a aboutnegligible food role deserts for the is explicitmulti-dimen use of - village shops. In addition to convenience and pric- - The project was financed by the German Research Founda - hanna Fuhrmann for undertaking the cartography and Dr. tion (JU332/17-1). Many thanks go to Joschka Johst and Jo regionales, lifestyle specialities, attitudes demands – underpinned that are by completely purchas Katherine Thomas for the translation. independenting power – favour of age. organic articles, fresh food and References

Augustin, H. - deserts from the perspective of potential consumers 2014: Stadt, Ernährung und soziale Ungleich andThe demonstrating investigation focusesthe consequences on operationalising of different foodper- heit – Zur Übertragbarkeit des food desert-Konzeptes ceptual patterns for the emergence of tangible food - auf den deutschen Kontext. – Bremen (=artec-paper 197). min/user_upload/single_sites/artec/artec_Dokumente/ – Available online at: http://www.uni-bremen.de/filead sets of predispositions that correlate with tangible patternsdeserts. The of consumption latter result fromand that mentally can lead consolidated to disad- Baaser, U. and K. Zehner artec-paper/197_paper.pdf, accessed 30/03/2017 vantageous local retail supply for the individuals con- Jürgens, U. (ed.): Entwicklungen 2014: Nahversorgung im Wandel. cerned over the course of their life cycle (due to the - Das Beispiel Köln. – In: market exit of retail structures for which there is no zelhandel. Kiel (Kieler Arbeitspapiere zur Landeskunde und Perspektiven von Nahversorgung im Lebensmittelein demand) or for other groups with social and mobility disadvantages. In line with the notions of Thomas https://www.geographie.hu-berlin.de/de/abteilungen/ und Raumordnung 54): 35-45. – Available online at: Schelling - wirtschaftsgeographie/AK-Einzelhandelsforschung/ lative disadvantage of their individual behaviour for publikationen/Berichte%20des%20Arbeitskreises%20 the whole (1978), of society customers and are do unaware not realise of thethe effectscumu -%20Geographische%20Handelsforschung%20-%20 - Bericht%2035.pdf#page=8 Bastian, M., S. Heymann and M. Jacomy 2009: Gephi: An Open , accessed 30/03/2017 consequenceand the ‘victims’ of abstract of their mentalactions. or Food psychological deserts iden de- - velopmentstified in tangible that promotespace via food GIS deserts,analyses rather are thus than the a - Source Software for Exploring and Manipulating Net ence on Weblogs and Social Media, San Jose/USA works. – Proceedings of 3rd International AAAI Confer local supply. Broadbridge, A. and E. Calderwood - reflection of the result or cause of the thinning out of - 2002: Rural gro - - cery shoppers: do their attitudes reflect their ac tribution of suppliers to the development of food de- tion Management 30 https://doi. tions? – International Journal of Retail & Distribu serts,The article e.g. how does the not individual tackle the providers question ofmanage the con to org/10.1108/09590550210435282 (8): 394-406, – doi: ensure customer loyalty. What are the niches that Chen, X. and J. Clark 2013: Interactive three-dimensional

Geography 43 geovisualization of space-time access to food. – Applied oflocal local retail retail must in filllocal in ordergovernance to keep and their why patrons are the or apgeog.2013.05.012 : 81-86, – doi: http://dx.doi.org/10.1016/j. chancesregain a ofsceptical commercial clientele? success What and is social the significance acceptance Chen, X. and J. Clark 2016: Measuring space-time access to food retailers: a case of temporal access disparity in Investigation further requires a temporal, dynamic Franklin County, Ohio. 68 (2): approachmore promising in order in to one demonstrate municipality the than effects in another? that the doi: http://dx.doi.org/10.1080/00330124.201 – Professional Geographer closing or re-opening of a shop has on local percep- 175-188, – tions and discourse, how the population adjusts to the Cherven, K. 5.1032876 altered conditions, and whether their predispositions Birmingham 2015: Mastering Gephi network visualization. – are changeable (Cummins Corapi 2014). Choi, Y. and T. Suzuki 2013: Food deserts, activity patterns, - et al. 2014; Geography 43 http://dx.doi.org/10.1016/j. & social exclusion: The case of Tokyo, Japan. – Applied This would then reveal whether the threat of poten apgeog.2013.05.009 : 87-98, – doi: foodtial food deserts deserts are cannot beonly managed an Anglo-American and minimised. pheno- The Corapi, S. 2014: Why it takes more than a grocery store to menon.first step in the German context is the realisation that ht t p:// www.pbs.org/newshour/updates/takes-grocery-store- eliminate a ‘food desert’. – Available online at:

40 DIE ERDE · Vol. 149 · 1/2018 ‘Real’ versus ‘mental’ food deserts from the consumer perspective – concepts and quantitative methods applied to rural areas of Germany

eliminate-food-desert/, accessed 10/09/2016 men Deutschlands und Südtirols im Vergleich. - Cummins, S., E. Flint and S. Matthews - 73: 45-58, – Raum hood grocery store increased awareness of food access 2014: New neighbor forschung und Raumordnung – doi:10.1007/ 33 Küpper, P. and A. Tautz s13147-014-0326-3 did not alter dietary habits or obesity. – Health Affairs 2015: Sicherung der Nahversorgung Del Casino, V. 21 (3): 138-155. (2): 283-291, – doi: 10.1377/hlthaff.2013.0512 in ländlichen Räumen Europas – Strategien ausgewählter Human Geography 39 https://doi. Available online at: http://www.ssoar.info/ssoar/handle/ Länder im Vergleich. – Europa Regional – 2015: Social geography I: food. – Progress in document/43929 (6): 800-808, – doi: Eckert, J. and S. Shetty 2011: Food systems, planning and Landesvermessungsamt Schleswig-Holstein 2015: Geodata for org/10.1177/0309132514562997 , accessed 30/03/2017

Applied Geography 31 http://www.schleswigholstein.de/DE/Landesregierung/ quantifying access: using GIS to plan for food retail. – the province of Schleswig-Holstein. – Available online at: org/10.1016/j.apgeog.2011.01.011 - : 1216-1223, – doi: http://dx.doi. Farley, T., J. Rice, J. Bodor, D. Cohen, R. Bluthenthal and D. Rose enService_Geobasisdaten_mehrLesen.html LVERMGEOSH/Service/serviceGeobasisdaten/geodat 2009: Measuring the food environment: shelf space of 29/01/2018 , accessed fruits, vegetables, and snack foods in stores. - Journal of Leete, L., N. Bania and A. Sparks-Ibanga 2012: Congru- Urban Health 86 ence and coverage: alternative approaches to identi- 009-9390-3 - (5): 672-682, – doi: 10.1007/s11524- Goodman, S. and H. Remaud 2015: Store choice: how under- 32 fying urban food deserts. – Journal of Planning Educa standing consumer choice of ´where` to shop may assist tion and Research (2): 204-218, – doi: https://doi. Maggioni, I. 2016: What drives customer loyalty in grocery org/10.1177/0739456X11427145 Services 23 - the small retailer. – Journal of Retailing and Consumer jretconser.2014.12.008 - : 118-124, – doi: http://dx.doi.org/10.1016/j. retail? Exploring shoppers´ perceptions through associa Hahne, U. - vices 33 - tive networks. – Journal of Retailing and Consumer Ser ser.2016.08.012 2016: Ländliche Räume im Wandel. In: AgrarBünd : 120-126, – http://dx.doi.org/10.1016/j.jretcon Meyer, G. 2005: Untersuchungen zur Entwicklung der Jenaer nis e. V. (ed.): Der kritische Agrarbericht 2016. – Hamm: Hellmann, K. and J. Marschall - 175-181 Stegbauer, C. and R. Häußling 2010: Netzwerkanalyse in der Innenstadt. – Mainz (Reihe Geographisches Institut Uni Monheim, R. and J. Heller 2016: Die Innenstadt von Leipzig und Konsumforschung. – In: versität Mainz) 656 (eds.): Handbuch Netzwerkforschung. – Wiesbaden: 647- Helmle, S. and C. Kuczera - Moore, L. and A. Roux 2006: Associations of neighborhood die Höfe am Brühl im Wandel. – Bayreuth - characteristics with the location and type of food stores. 2015: Ländliche Lebensverhält 96 nisse im Wandel 1952, 1972, 1993, 2012: Vol. 2. - Braun https://www. schweig: Johann Heinrich von Thünen-Institut (=Thünen – American Journal of Public Health (2): 325-331, – doi: Morland, K., A. Roux and S. Wing 2006: Supermarkets, oth- Report 32, Vol. 2). – Available online at: 10.2105/AJPH.2004.058040 pdf - econstor.eu/bitstream/10419/124160/1/837830745. Holzmann, T. and G. Wührer 2000: Kognitive Landkarten zur ventive Medicine 30 , accessed 30/03/2017 er food stores, and obesity. – American Journal of Pre Entscheidungsunterstützung im Markenmanagement. org/10.1016/j.amepre.2005.11.003 (4): 333-339, – doi: http://dx.doi. Foscht, T., G. Jungwirth and P. Schnedlitz (eds.): Neumeier, S. - – In: 2015: Regional accessibility of supermarkets Frankfurt/Main: 429-443 65 Zukunftsperspektiven für das Handelsmanagement. – and discounters in Germany – a quantitative assess IEL - Institut für Entwicklungsforschung im Ländlichen Raum online at: - ment. – Landbauforschung (1): 29-46. – Available Ober- und Mittelfrankens (ed.) 2015: Einzelhandel im - http://go.galegroup.com/ps/anonymous?p=A - ONE&sw=w&issn=21943605&v=2.1&it=r&id=GALE% Jürgens, U. - , accessed ländlichen Raum. – Heiligenstadt 7CA432065009&sid=googleScholar&linkaccess=full 2016: Gibt es Food Deserts in Deutschland? Unter text&authCount=1&isAnonymousEntry=true 68 (6): 48-55 Nielsen Company suchungen in ländlichen Räumen von Schleswig-Holstein. 30/03/2017 Kestens, Y., A. Lebel, M. Daniel, M. Thériaut and R. Pampalon „small supermarkets“, supermarkets and food discount – Geographische Rundschau (2015): TradeDimensions – Dataset for 2010: Using experienced activity spaces to measure food- 16 Pettygrove, M. and R. Ghose 2016: Mapping urban geogra- stores. – Frankfurt/Main http://dx.doi.org/10.1016/j.healthplace.2010.06.016 phies of food and dietary health: a synthesized frame- scape exposure. – Health & Place : 1094-1103, – doi: Küpper, P. and C. Scheibe 10 - 10.1111/gec3.12268 2015: Steuern oder fördern? Die work. – Geography Compass (6): 268-28, – doi: Sicherung der Nahversorgung in den ländlichen Räu DIE ERDE · Vol. 149 · 1/2018 41 ‘Real’ versus ‘mental’ food deserts from the consumer perspective – concepts and quantitative methods applied to rural areas of Germany

Russell, S. and P. Heidkamp Teichert, T. and K. Schöntag 2010: Exploring consumer 2011: ‘Food desertification’: the forschung 1996/97: 315-339 31 knowledge structures using associative network ana- loss of a major supermarket in New Haven, Connecticut. org/10.1016/j.apgeog.2011.01.010 27 – Applied Geography : 1197-1209, – doi: http://dx.doi. Sadler, R., J. Gilliland and G. Arku 10.1002/mar.20332 lysis. – Psychology & Marketing (4): 369-398, – doi: Thibodeaux, J. 2016: City racial composition as a pre- 2016: Theoretical issues in 81 - the ‘food desert’ debate and ways forward. – GeoJournal Schaloske, M. 2013: Dorfentwicklung vor neuen Heraus- ban Studies 53 : 443-455, – doi:10.1007/s10708-015-9634-6 dictor of African American food deserts. – Ur (11): 2238-2252, – doi: https://doi. 75 Ver Ploeg, M., P. Dutko and V. Breneman 2015: Measuring forderungen – die wachsende Bedeutung engagierter org/10.1177/0042098015587848 - Bürger. – Flächenmanagement und Bodenordnung Schelling, T. 37 (2): 205-225, (2): 67-72 food access and food deserts for policy purposes. – Ap 1978: Micromotives and macrobehavior. – New plied Economic Perspectives and Policy Schendera, C. Warburg, F. 2011: Konsequenzen des zweiten demographi- York – doi: https://doi.org/10.1093/aepp/ppu035 Schenk, T. 2016: 2010: Clusteranalyse mit SPSS. – München - Helvetica 66 Versorgungssicherung in Nordschweden schen Übergangs für den ländlichen Raum. – Geographica 68 (6): 18-25 Widener, M, S. Farber, T. Neutens and M. Horner 2013: Using am Beispiel des Lebensmitteleinzelhandels. – Geogra (2): 132-139, – doi:10.5194/gh-66-132-2011 Segerer, M. 2014: LM-Discounter oder Supermarkt: Wer ist urban commuting data to calculate a spatiotemporal phische Rundschau

Jürgens, U. (ed.): Entwicklun- 21 der wahre Nahversorger – empirische Ergebnisse aus accessibility measure for food environment studies. – - healthplace.2013.01.004 dem Freistaat Bayern. – In: Health & Place : 1-9, – doi: http://dx.doi.org/10.1016/j. - Widener, M. and J. Shannon gen und Perspektiven von Nahversorgung im Lebensmit 4): 123-166 teleinzelhandel. – Kiel (Kieler Arbeitspapiere zu Landes 2014: When are food deserts? Shaw, H. 2006: Food deserts: towards the development of a kunde und Raumordnung 5 Integrating time into research on food accessibility. – 88 Health & Place 30: 1-3, – doi: http://dx.doi.org/10.1016/j. Wright, J., A. Donley, M. Gualtieri and S. Strickhouser 2016: classification. – Geografiska Annaler (2): 231-247, – healthplace.2014.07.011 Seitz, C. - doi: 10.1111/j.0435-3684.2006.00217.x - 53 2015: The perception of and the attitude towards Food deserts: What is the problem? What is the solu http://mediatum.ub.tum.de/ regional food products. – Diss. University of Munich, Mu tion? – Social Science and Public Policy : 171-181, – Wühler, G. Diller, H. (ed.): nich. – Available online at: doi:10.1007/s12115-016-9993-8 Sperk, C. and T. Kistemann 2012: Food Desert oder gesunde doc/1275296/file.pdf, accessed 30/03/2017 2001: Kognitive Landkarten. – In: Vahlens Großes Marketinglexikon Vol. 1. – München: 776- 86 (2): 135- Zibell, B., J. Revilla Diez, I. Heineking, P. Preuß, H. Bloem and Stadt? Eine Untersuchung von Nahrungslandschaften in 777 151 F. Sohns - Bonn. – Berichte zur deutschen Landeskunde Statistical Office SH (resp. Statistikamt Nord) 2011: Bevöl- Fachinger, U. and H. Künemund 2015: Zukunft der Nahversorgung in ländli kerungsentwicklung auf Gemeindebasis Schleswig-Hol- chen Räumen. – In: (eds.): Gerontologie und ländlicher Raum. – Wiesbaden: Steinröx, M. - stein. – Kiel und Hamburg 141-165. – Available online at: http://link.springer. 2013: Ländlicher Raum bald ohne Nahver com/chapter/10.1007/978-3-658-09005-0_8, accessed Swoboda, B. 1996: Auswirkungen der Ladenwahrneh- - sorgung? – Neues Archiv für Niedersachsen (2): 108-120 30/03/2017 anonymous?p=AONE&sw=w&issn=219436 mung auf Kaufverhalten und Einkaufszufriedenheit. - 05&v=2.1&it=r&id=GALE%7CA432065009&sid=google - mousEntry=true Scholar&linkaccess=fulltext&authCount=1&isAnony – In: Forschungsstelle für den Handel Berlin: Handels , accessed 30/03/2017

42 DIE ERDE · Vol. 149 · 1/2018 Appendix: Criteria surveyed in rural municipalities 2016. Source: Data collection Jürgens 2016

I strongly I disagree I neither I agree I strongly disagree agree nor agree disagree 1. I shop in a discount store (Aldi, Lidl, Penny). 1 2 3 4 5 2. I shop in a supermarket (Edeka, Sky, Rewe). 1 2 3 4 5 3. The choice of products in a discount store is enough for me. 1 2 3 4 5 4. The choice of products in a supermarket is enough for me. 1 2 3 4 5 5. I combine food shopping in discount stores and supermarkets. 1 2 3 4 5 6. I am a patron (regular customer) in my supermarket or discount store. 1 2 3 4 5 7. I always buy everything in one shop of my choice. 1 2 3 4 5 8. I choose the shop that is closest to home. 1 2 3 4 5 9. I choose the shop that is closest to my place of work. 1 2 3 4 5 10. I would like the shop to be easily accessible by car. 1 2 3 4 5 11. I would like the shop to be easily accessible on foot. 1 2 3 4 5 12. I would like the shop to be easily accessible by bicycle. 1 2 3 4 5 13. I would like the shop to be easily accessible by bus. 1 2 3 4 5 14. I would like many dierent shops nearby so I can compare products and prices. 1 2 3 4 5 15. When shopping for food the price is especially important to me. 1 2 3 4 5 16. I like to cook and buy the food for cooking. 1 2 3 4 5 17. Branded (food) products are especially important to me. 1 2 3 4 5 18. I don’t care about brands at all, the main thing is to get the right quality. 1 2 3 4 5 19. A large choice of food is important to me. 1 2 3 4 5 20. Fresh products are important to me. 1 2 3 4 5 21. Organic products are important to me. 1 2 3 4 5 22. Being served/advised at a shop counter is important to me. 1 2 3 4 5 23. I come primarily for the special oers. 1 2 3 4 5 24. If I can save a bit again, then especially with food. 1 2 3 4 5 25. I want to be able to buy non-groceries too 1 2 3 4 5 (e.g. textiles, computers, books, gardening articles) 26. I also come because of the more convenient parking. 1 2 3 4 5 27. I like to make use of the longer opening times. 1 2 3 4 5 28. I come because I feel comfortable in the shop. 1 2 3 4 5 29. I go shopping on a xed schedule. 1 2 3 4 5 30. I like to buy food spontaneously. 1 2 3 4 5 31. Occasionally I would also like to treat myself when food shopping. 1 2 3 4 5 32. I can enjoy shopping for food. 1 2 3 4 5 33. Shopping for food is simply something I HAVE to do. 1 2 3 4 5 34. I like to go shopping in the bakery ‘round the corner’ (a traditional baker’s shop). 1 2 3 4 5 35. I like to go shopping in the butchers (a traditional butcher’s shop) ‘round the corner’. 1 2 3 4 5 36. I also use other alternatives like online food shopping. 1 2 3 4 5 37. I also use other alternatives like a mobile supermarket. 1 2 3 4 5 38. I also use other alternatives like a farm shop. 1 2 3 4 5 39. I also use other alternatives like the weekly market. 1 2 3 4 5 40. I also use other alternatives like a delivery service e.g. from Rewe, Sky, Edeka. 1 2 3 4 5

DIE ERDE · Vol. 149 · 1/2018 43