Report No DWI0771




Background and objectives

E1 M.E.L Research was commissioned in September 1994 by the Drinking Water Inspectorate (DWI) to carry out a survey of tap water consumption in England and Wales. The purpose of the study was to provide an up-to-date estimate of average daily tap water consumption and the fraction that this comprises of the individual's total liquid consumption. This information is needed in order to quantify the risks posed to individuals by microbiological or heavy metal content in the domestic tap water supply.

E2 To date, estimates of population exposure to such risks have been based on the findings of a national survey carried out in 1978 by the Water Research Centre (WRc)(l). It was originally anticipated that M.E.L Research would replicate the methodology used in 1978 so that a statistical test of significant changes in consumption over the period 1978 - 1995 could be carried out. It became evident however that a new approach should be adopted to produce more accurate results. The new method involved adapting and updating the original survey methodology to take account of changes in liquid consumption. These changes include the increased availability of bottled water and the growth in water-mixed foods such as instant noodles, rice and pasta. Developing a new methodology also meant that an opportunity existed to make use of developments in consumer research methodology.

E3 In summary, the four central objectives of research reported here were:

    1. to estimate average total daily tap water consumption in England and Wales and the proportion of total daily liquid consumption that this represents;
    2. to estimate the relative significance of tap water substitutes consumed, with particular attention to bottled water;
    3. to assess trends in consumption of key types of drink; and
    4. to consider seasonal changes in consumption of key types of drink.

((1) Water Research Centre (1980) Drinking Water Consumption in Great Britain: A survey of drinking habits with special reference to tap-water based beverages WRc Technical Report TR137)

Research design and method

E4 Once the scope of the study was clear, consideration was given to the best methods for gathering the required information. It became clear that a sole reliance on traditional methods would not be adequate for the study, and an innovative approach was devised. This drew upon the most recent developments in diary - based market research methods to provide a data collection instrument which would be both cost-effective and provide reliable results. The methods were tested during pilot studies and several refinements were made.

E5 Three survey instruments were used to gather information:

    1. A face-to-face questionnaire, answered by the head of the household.
    2. A self completion questionnaire, answered by each member of the household.
    3. A diary of drinking 'events', completed by each member of the household.

E6 A total of 500 households were needed to take part in the survey. These were selected to be representative in terms of:

E7 A total of 10 households were selected within 50 local authority areas. The local authority areas were selected randomly within the 10 standard planning regions of England and Wales.E8 The survey was carried out between February and April 1995. In total, 476 households provided complete information, representing 1,018 individuals.

Average tap water consumption

E9 The average quantity of tap water consumed by individuals, as measured by this survey, is 1.138 litres per day.

E10 The average quantity of tap water consumed in 1978, as measured by WRc's survey, was. 0.955 litres per day.

E11 Comparison of the two figures reveals a statistically significant increase in tap water consumption, at the 1% level of significance. However, this does not necessarily mean that there has been an absolute increase in tap water consumption since the two surveys used slightly different methods.

Average total liquid consumption

E12 The average quantity of total liquid consumed by individuals, as measured by this survey is 1.561 litres per day.

E13 The average quantity of total liquid consumed by individuals in 1978, as measured by WRc's survey, was 1.589 litres per day.

E14 Comparison of the two figures reveals that there is no statistically significant difference between the two measurements, either at the 1 % or the 5% level of significance.

Effect of age on quantities consumed

E15 Individuals aged between 46 and 55 consume the most liquid in total (an average of 1.885 litres per day) and the most tap water per day (1.493 litres per day). Consumption of both total liquid and tap water increases with age, reaches a peak at 46-55 and then declines slightly.

Effect of social group on quantities consumed

E16 There is very little difference in consumption across the social groups. The clerical and administrative social group (C1) has the highest level of total liquid consumption, (1.695 litres per day) while retired people have the highest level of tap water consumption (1.315 litres per day).

Effect of gender on quantities consumed

E17 While men drink a higher total quantity of liquid than women (1.617 litres per day compared with 1.515 litres per day), women drink more tap water than men (1.149 litres per day compared with 1.127 litres per day).

Effect of geographical region on quantities consumed

E18 The highest levels of both total liquid and tap water consumption lie in the north (1.661 and 1.228 litres per day respectively). The lowest levels lie in the midlands/Wales region (1.476 and 1.087 litres per day respectively). However, there is gen erally less regional differentiation than in 1978.

Consumption of tap water

E19 Only 7 households (1%) reported that there were members of their household that do not use tap water at all for drinking or for making drinks; 99% of households therefore drink tap water or make drinks using tap water. Just less than a third of households (31%) reported using other sources of water, of which most (30% of the whole household sample) referred to bottled water. Five respondents (3%) drew water from a spring and one respondent used rain water.

Types of drink

E20 This research was not resourced to provide a comprehensive picture of the quantities of tap water consumed in the form of different drinks. To obtain an overall picture of the main drink 'intake routes' for tap water, the whole quantity of reported drinking in the sample was totalled and divided by volume into the principal drink types. Table E1 below shows the proportion of the total volume of tap water consumed by the sample during the week, according to drink type.

Table E1: Types of drink as a source of tap water

Type of drink

% of all tap water consumption





Tap water


Other cold drinks


Other hot drinks


Other sources


Methods of preparing tap water

E21 Households were asked about appliances they own for preparing water prior to its consumption. The following responses were given.

Table E2: Water preparation appliances owned

Water filtering jug



"Soda-stream" or other sparkling drinks maker



Cold water dispenser in fridge



Teasmade machine



Plumbed-in special water filtering tap






Bottled water cooler machine



"Permutit" or other water softening device




Complaints about the quality of tap water

E22 Use of tap water and trends towards other forms of liquid consumption may be linked with perceptions about the quality of tap water. This was not an issue addressed directly in this study but a short series of questions was included on whether households had experienced any problems with the quality of their tap water within the past year or two.

E23 Table E3 shows that a fifth of the sample said they had experienced problems with the colour or taste of the water, and around a tenth with the smell. Of these, the problem most likely to result in a complaint is the colour, where a fifth of those e xperiencing a problem had made a complaint. For problems of smell and taste only 2% of those with a problem had made a complaint. It should be noted that a further 8% of the sample had mentioned problems with tap water quality other than smell, colour and taste.

Table E3: Percentage of consumers experiencing problems with the quality of their tap water

Quality problem

Had problem




%of those questioned


%of those with problem


%of those complained






























Brands of bottled water consumed

E24 Questions were included on the respective consumption of still and sparkling bottled water, and the main brand names bought. In total 121 households drink still bottled water and 79 sparkling water (some drink both). While in both cases the most frequently mentioned response was 'anything/it varies' indicating little brand loyalty, the most noteworthy point is the diversity of brands mentioned and the lack of any particular dominant leader. Perrier and Sainsbury's head the sparkling water list, and Evian and Sainsbury's the still water list, but 50 different brand names were given by the 159 households responding to this question. Profile of bottled water drinkers

E25 Bottled water drinkers are most likely to be female, aged between 16 and 45, from the midlands, Wales or the south, and part of a household in which the main wage earner is in a professional/managerial (AB), clerical (Cl) or skilled manual (C2) social group. This would appear to conform to the stereotype image of bottled water drinkers, and while therefore not adding anything unexpected, the results are confirmation of a distinctive social element in drinking water consumption.

E26 It may however be wrong to assume that bottled water consumption has acted as a substitute for tap water consumption. There is a persuasive hypothesis that it has substituted for commercially produced drinks and alcohol consumption amongst the population group profiled above. While this could not be established without further analysis of the consumption of various types of drinks, it might help to explain why the consumption of bottled water has apparently increased over a period there has also been an apparent increase in tap water consumption, yet total liquid consumption has remained almost totally static.

Seasonal changes in water consumption

E27 For just over three-quarters (77%) of the sample, levels of tea drinking are reported to be broadly constant throughout the year. Nearly a fifth of the sample (18%) however, say they tend to drink more tea in winter, while 5% drink more in summer. Although there is little difference in the seasonal trend by gender, there is an interesting difference according to age. The younger age groups are more likely to say they drink more tea in winter, while for the older age groups tea drinking does not vary with season.

E28 For coffee, about the same proportions on average drink more in winter (19%) as for tea. However very few respondents said they drink more coffee in summer. As with tea, the 65 plus age band is much less likely to report any seasonal difference in coffee consumption, while the younger age groups are more likely to report increases in winter coffee consumption.

E29 Squash drinking is very much increased in the summer, with two thirds of squash drinkers saying they drink more squash in summer and a third saying it does not vary. Here however it is the older respondents who are most seasonal in their squash consumption.

E30 Similarly. consumption of both sparkling and still bottled water is reported higher in summer, but consumption of still bottled water is more seasonal than sparkling. Half those who ever drink still bottled water drink more in the summer, while for sparkling water drinkers, only 41% drink more in the summer. As with squash, the older age groups are the most likely to show seasonal variation.

Changes in drinking behaviour over the last five years

E31 The data on reported trends in tea and coffee consumption are of interest in that the evidence here points to an increase in tea consumption (17% say they drink more now, only 8% drinking less) while for coffee 14% say they drink more now, and 13% less, indicating relative stability. The reported increase in tea drinking is observed across all age ranges, but is higher in the younger groups. For coffee there is a similar tendency for younger people to report drinking more coffee these days, with older people less likely to report drinking more coffee.

E32 For squash there is a slight tendency for people to report drinking more these days, with 16% drinking more and 11 % drinking less. The under 16 age group is most likely to report an increase in squash drinking (26%), but there is also an interesting increase in the over-65s, with 14% drinking more and 7% less.

E33 For sparkling and still bottled water consumption, there is a not-unexpected pattern towards increased consumption. Over a third of the sample who drink bottled water report an increase in consumption in the past five years (36% for sparkling water, 39% for still water). Very few say they drink less these days. The reported increase is greater for women (40% report an increase in sparkling water consumption and 43% in still water consumption), than for men (30% and 35% respectively).

General conclusions and recommendations

E34 The drinking diary has proved to be a novel and effective method of obtaining drinking water consumption data from respondent individuals of all ages. The individual drink coupons have been developed and used effectively in the field and were completed fully by almost all respondents. The main merits of adopting this method in preference to the face to face interview approach alone, has been to allow much more detailed consumption data to be compiled.

E35 The potential usefulness of the survey data obtained from the diary coupons has not been fully exploited in the analysis here. For example, it would be possible to examine the time of day and day of week when drinks of different types were consumed. This could sharpen up data on exposure by highlighting the beverage ingestion routes most likely to involve 'first draw' or unflushed tap water - important from the viewpoint of exposure to heavy metals in the supply. It would also differentiate better between hot water (treated) and cold water intakes with consequent microbiological exposure considerations.

E36 The introduction of additional self-completion questionnaires on seasonal and long-term trends in drinks consumption has added to knowledge by highlighting periods of the year when particular exposure routes may be most significant. In particular it has highlighted the potentially much higher level of tap water consumption in the summer period.

E37 Overall the data from the survey on total liquid consumption show no significant change from the 1978 survey. Data on tap water consumption show a significant increase but this may be an artefact of the new methodology as we are not, strictly, comparing like with like. Overall, it is noteworthy however that if the new method provides a more accurate account of liquid consumption, it suggests that tap water comprises a somewhat larger fraction of total liquid consumption than had previously been thought.

E38 The other strategically significant finding is that the increase in bottled water consumption may not have substituted to any great extent for tap water. Instead, bottled water may simply have displaced other purchased beverages. Little is known about this however and we did not address it in the survey. Further research on this issue, and indeed the interchangeability and long term trends in consumption of the various alternative liquid sources, would be useful in highlighting the degree to which tap water is likely to remain the principal source of liquid intake of the population in the long term.

E39 In general therefore our conclusions and recommendations are that:

    1. the survey has generated useful data to update statistics on population consumption of liquid and tap water sources;
    2. further analysis could be carried out on the existing data to characterise the distribution and patterns of behaviour in more detail;
    3. more information is needed on particular seasonal or temperature-related consumption patterns to identify periods of higher exposure than revealed in the annual average;
    4. a panel-based application of the methods developed here would be an appropriate means of gathering primary data for such an investigation; and
    5. a more sophisticated statistical approach should be developed to the investigation of exposure to epidemiological risks from tap water by examining the population sub-groups most exposed and most vulnerable to the potential environmental health effects.
Copies of this report may be available as an Acrobat pdf download under the 'Pre 2000 Reports' heading on the DWI website.