Back to home page


Food Desert Metrics


Like their physical counterparts, food deserts may be said to have both extent and intensity.  Their extent is the physical area they cover – that is, the area of urban or rural settlement where difficulties of access to fresh fruit and vegetables may be experienced.

Their intensity is a measure of how many people within the food desert area actually face major difficulties in accessing a healthy diet.  Some food desert areas may be quite extensive, in that large tracts of residential land in cities, or large urban areas, are devoid of fresh fruit and vegetable shops.  However, for some of these extensive food desert areas, the majority of the population may be relatively affluent and have cars, and possess good knowledge of what constitutes a healthy diet, they have the time to cook and prepare such foods.  In other areas, quite a small area, such as a single edge of town housing estate, may constitute a ‘food desert’; however if a large majority of the households on that estate face barriers to healthy eating this is an intense food desert.


Measuring the extent and intensity of food deserts can be done in several ways. 


The first type of analysis involves measuring the distance between fresh fruit and vegetable outlets and residential areas.


The second type of analysis extends the study of food deserts to include ‘psychological distance’, ‘financial distance’, and ‘psychological distance’.. 


The third type of analysis extends the basic spatial quantification to include socio-economic factors that have an impact on healthy eating even though these factors are not of themselves distance-dependent.


The fourth type of analysis involves a more sophisticated spatial breakdown of the socio-demographic factors, investigating the effects of these factors across selected neighbourhood types, using a kind of ‘social tomography’, rather than the whole study area.


These four analysis methodologies are more fully described below.


1) Distance between shops and residential areas


The density of all retail food shops, including fresh fruit and vegetable shops, tends to decline as one moves outwards from the historic town or city centre, beyond the early suburbs built ca. 1800 – 1914.  In large conurbations, former village centres now swallowed up by suburban sprawl will generally retain a neighbourhood shopping centre in the former village centre.  The large supermarkets will often prefer a location out on the urban periphery, where main roads and the urban ring road offer good transport links for bringing in both retail goods and customers.   At fooddeserts,org we have developed a methodology based on the 250 x 250 metre grid squares of the food retailing maps hosted on this website that facilitates the classification of an urban residential area, square by square, according to how far each square is from its nearest fresh fruit and vegetable retailer. 

The results for Greater Birmingham are shown here. 

Click here for a map of distance to fruit and vegetable shops in Nantes, France

Note that whilst each square is 250 x 250 metres, shoppers do not travel in straight lines to the shops but must use the local street pattern.  Therefore each square distance represents some 300 metres average travel distance house-to-shops for that square.  Areas 7 squares out are then some 2,100 metres from the nearest fruit and vegetable shop.


Note that this does not mean that residents of such ‘distant’ areas will have difficulties accessing a healthy diet as they may well have access to a private household car.  Equally, residents of areas close to fruit and vegetable shops do not necessarily find it easy to access the fruit and vegetables they would like to eat; if an older White less affluent pensioner is living in an area with a large south Asian ethnic monitory, e.g. Sparkhill in Birmingham, this pensioner may be faced with most local stores selling fruit and vegetables she is unfamiliar with and has no idea how to prepare.


2) Psychological, financial, and physiological distance to shops


Just because a certain food store is, say, 600 metres from a residential area does not mean that all shoppers from this residential area will experience the same effort in getting there.  For example the store may be located in a valley and the houses are further uphill.  Younger, less affluent, and car-less shoppers on foot may have little difficulty carrying their shopping home but this could be a daunting task for a pensioner.  We can say that the presence of the uphill section on the way home has increased the physiological distance the pensioner must travel to/from the shop, as opposed to if the journey was all on the level.  Effectively, a store that is 600 metres away but 30 metres below the level of the houses might be the equivalent of 900 metres flat-travel distance for a frail pensioner.   Other such physiological barriers might include a busy road crossing for a mother with children, or a main road that can only be crossed by a pedestrian subway where there is a risk of mugging.  Effective psychological distance may even vary with the time of day and season; shops that are 500 metres away across a park may be easy to get to for a woman in the daytime, but on dark winter evenings she may be cautious of using this route and prefer a longer route via well-lit roads..  At we have developed a methodology to indicate, once such physiological barriers have been determined by shopper-interviews or other research, the ‘true-physiological-distance’ to the shops, as opposed to the 2-dimensional flat distance as shown on a map.


Some features actually reduce physiological distances.  The presence of a frequent bus route may bring stores effectively closer for car-less households.  Certain stores, or store brands, may also attract or repel shoppers, which effectively reduces or increases the psychological distance of the store.  Take for example a thrifty pensioner, who dislikes very large hypermarkets because they find them confusing and it’s a long walk to the tills; however this pensioner likes the cheaper discount stores Aldi, Lidl, and Netto.  For this pensioner, the actual (map) distance to their nearest Tesco Extra and Lidl could be 700 metres and 1,000 metres respectively.  But the psychological preferences might make the pensioner shop as if the Tesco was 1,200 metres away and the Lidl, 500 metres.


However for the unemployed a return bus journey of around £3.00 represents a large slice of their disposable income and in the absence of subsidised fares for the jobless this cost will increase the ‘financial distance ‘they face in travelling to the shops and predispose them to shop at more local shops, where often only unhealthy foods such as burgers and takeaways are available.


At we have developed an algorithm that can modify actual-map distances for different store types, or for features such as bus routes and hills, to produce psychological-distance maps for various consumer types, with various shop preferences or travel capabilities.


3) Socio-economic factors


A range of socio-economic factors have been linked to the propensity to consume a poor diet, and to become obese.  These indicators range from a low level of educational achievement and suffering ill health / being a carer, to unemployment and lack of a household car.  Research for in Birmingham, Leeds, London, and other cities suggests that level of educational achievement is a strong indicator of dietary quality and obesity.  Some socio-demographic factors are linked in complex ways.  For example obesity tends to increase with age, especially for poorer people; there is also an ethnic link in that younger less-affluent Asians tend to be less obese than young less-affluent Whites, and there is also a positive correlation between poverty and obesity.


When distance-to-shops measures are combined with socio-economic factors some interesting relationships emerge.  There is evidence, at least for some large urban areas, that the tendency to be obese increases with distance from shops in less-affluent areas, but falls off with increasing distance from shops in better off areas.  Conversely, and perhaps even more counter-intuitively, whilst higher levels of unemployment are positively associated with obesity and poor diet in more affluent areas, being unemployed in a less-affluent area may actually predispose to a healthier diet.  The reason for this may be that in affluent areas, the wealthier tend to live in large houses distant from shops (and have cars); the poor may live closer to the shops but have less income to spend on healthy food, and in these affluent areas do not have access to cheap markets.  In less affluent areas the working poor have less time to prepare healthy food, whereas the unemployed (on not so much less disposable income than that possessed by a working poor person on the Minimum Wage) have time to seek out cheaper food bargains and to cook.


By combining spatial shop-distance data with census area or postcode-based data a full picture can be built up of the interplays between spatial and socio-economic factors in influencing quality of diet and obesity.


4) Spatial breakdown of socio-economic factors affecting food access, diet, and obesity


A more complex level of analysis involves a systematic analysis of partial areas selected for by degree of social deprivation.  There are a large number of ways in which urban or rural neighbourhoods can be selected in order to deliberately obtain ‘biased samples’ of such areas.  Systematic analysis of a series of such areas as done at is used to produce a kind of ‘social tomography’ along multiple dimensions of social and economic deprivation.  More information can be extracted from the manner in which the closeness of fit between various indicators changes as one moves across the social spectrum, from e.g. rich to poor.  N example of this is that in poorer parts of Birmingham, the propensity to obesity associated with having no car is similar to that associated with being on Income Support; however in wealthier areas the factor ‘Income Support’ is still associated with obesity whereas ‘no car’ is not.  A possible explanation may be that in wealthy areas many older people have no car due to health, not poverty, reasons; these people can still afford a healthy diet, and can pay for a taxi to the nearest Waitrose to get these foods.


Further ‘dimensions ‘may be added by comparison with different areas, or with the same area as it changes over time.  Areas such as Hall Green in Birmingham and Armley in Leeds have seen a shift in ethnicity over the past decade, and a change in the type of fresh fruit and vegetable retailing on offer; this will alter the interactions between age, ethnicity, poverty, and obesity.


Social tomography of food deserts produces a shifting pattern of links between social deprivation and diet; underlying social principles are overlaid by a set of social circumstances distinctive to each rural or urban area.  These factors determine the most appropriate food policies for an area.  For example a poor area with many low skilled and low paid jobs may require an intervention aimed at introducing easy and quick to prepare healthier foods, e.g. pasts, fish, and vegetables, rather than more complex nutritious dishes that require considerable time to cook.  At our socio-spatial analysis not only elicits the set of dietary deprivation circumstances that are unique to each neighbourhood, but can also suggest the most cost-effective social intervention that would improve diet and reduce obesity for that area.


Back to top