methods annex report v2

Upload: jose-manuel

Post on 03-Jun-2018

228 views

Category:

Documents


0 download

TRANSCRIPT

  • 8/12/2019 Methods Annex Report v2

    1/103

    Annex Report (v2)

    Methods used for Household Foodand Drink Waste in the UK 2012

    This report is an annex to Household Food and Drink Waste in theUnited Kingdom 2012 containing a description of the research methodsand calculations used to obtain the estimates in the main report. It alsoincludes information on uncertainty around the estimates.

    Project code: CFP102 ISBN: 978-1-84405-459-6Research date: May 2012 September 2013 Date: November 2013

  • 8/12/2019 Methods Annex Report v2

    2/103

    WRAPs vision is a world without waste,where resources are used sustainably.

    We work with businesses, individual andcommunities to help them reap thebenefits of reducing waste, developingsustainable products and using resourcesin an efficient way.

    Find out more at www.wrap.org.uk

    Written by: Tom Quested, Sophie Easteal (both WRAP), Robert Ingle (support to WRAPfrom SKM Enviros) with contributions from Ipsos MORI, Resource Futures and ExodusResearch

    Front cover photography: Images of food collected during waste composition analysis (2008), WRAP

    While we have tried to make sure this report is accurate, we cannot accept responsibility or be held legally responsible for any loss or damage arising out of or inconnection with this information being inaccurate, incomplete or misleading. This material is copyrighted. You can copy it free of charge as long as the material isaccurate and not used in a misleading context. You must id entify the source of the material and acknowledge our copyright. You must not use material to endorse orsuggest we have endorsed a commercial product or service. For more details please see our terms and conditions on our website a t www.wrap.org.uk

    http://www.wrap.org.uk/http://www.wrap.org.uk/http://www.wrap.org.uk/http://www.wrap.org.uk/http://www.wrap.org.uk/http://www.wrap.org.uk/http://www.wrap.org.uk/http://www.wrap.org.uk/
  • 8/12/2019 Methods Annex Report v2

    3/103

    Methods used for Household Food and Drink Waste in the UK 2012

    Contents1.0 Introduction ................................................................................................. 6 2.0 Rationale for source data used in calculations ............................................. 9

    2.1 Food and drink waste collected by local authorities ....................................... 9 2.2 Food and drink waste home composted and fed to animals ......................... 10 2.3 Food and drink waste going down the drain ............................................... 10

    2.3.1 Seasonal effects ............................................................................. 13 2.3.2 Weather effects ............................................................................. 14 2.3.3 Scope of diary research: disposal routes included ............................. 16 2.3.4 Differences in sample profile ........................................................... 18 2.3.5 Actual reduction in food and drink being disposed of to the sewer...... 19 2.3.6 Conclusions relating to differences in sewer waste estimates ............. 21

    3.0 Overview of calculations ............................................................................ 23 3.1 Calculations for local authority collected waste ........................................... 23 3.2 Calculations for sewer waste ..................................................................... 23 3.3 Calculations for home composting and fed to animals ................................. 24

    3.3.1 Weight of items ............................................................................. 24 3.3.2 Edible portion ................................................................................ 24

    3.4 Calculations relating to reason for disposal ................................................. 25 3.5 Calculations relating to tea, coffee, hot chocolate and formula milk .............. 26 3.6 Calculations relating to carcass meat and bones ......................................... 27

    4.0 Methods for detailed waste compositional analysis ................................... 28 4.1 Overview ................................................................................................. 28 4.2 Sample design ......................................................................................... 28

    4.2.1 Selecting local authorities to participate ........................................... 28 4.2.2 Selecting collection routes and properties ......................................... 29

    4.3 Questionnaire survey fieldwork .................................................................. 31

    4.3.1 Questionnaire ................................................................................ 32 4.3.2 Obtaining informed consent ............................................................ 32 4.3.3 Confidentiality ................................................................................ 33

    4.4 Waste compositional analysis .................................................................... 33 4.4.1 Collection of kerbside sample materials ............................................ 33 4.4.2 Interaction with the public .............................................................. 34 4.4.3 Sorting the waste ........................................................................... 34 4.4.4 Using electronic data entry ............................................................. 35

    5.0 Methods for Kitchen Diary 2012 research .................................................. 37 5.1 Overview ................................................................................................. 37 5.2 Recruitment interviews ............................................................................. 37

    5.3

    Pre-diary questionnaire ............................................................................. 39

    5.4 Kitchen diary ........................................................................................... 39 5.5 Post-diary questionnaire ........................................................................... 40

    6.0 Methods for Local Authority Synthesis ....................................................... 42 7.0 Weighting and household count ................................................................. 43

    7.1 Objective of weighting in this study ........................................................... 43 7.2 Data Sources relating to households .......................................................... 43 7.3 Weighting of detailed waste compositional analysis ..................................... 46 7.4 Weighting of Kitchen Diary 2012 Research ................................................. 48 7.5 Weighting of local authority synthesis ........................................................ 49

    8.0 Calculations of food and drink prices.......................................................... 50 8.1 Source of retail-price data ......................................................................... 50 8.2 Comparison of new and previous methods ................................................. 51 8.3 Application of cost for products where water is added in the home ............... 52

  • 8/12/2019 Methods Annex Report v2

    4/103

    Methods used for Household Food and Drink Waste in the UK 2012 4

    8.4 Adjustments for weight change on cooking ................................................ 52 9.0 Calculations of greenhouse gas emissions and land use requirements ...... 53

    9.1 Overview ................................................................................................. 53 9.2 Methodology ............................................................................................ 53

    9.2.1 Top-down calculation ..................................................................... 53 9.2.2 Bottom-up calculation ..................................................................... 54

    9.3 Results .................................................................................................... 55

    9.4 Calculations of equivalent impacts (cars off the road) .................................. 56 9.5 References for environmental impact calculations ....................................... 56 10.0 Comparison of waste and purchases .......................................................... 60 11.0 The impact of seasonality on the research ................................................. 61

    11.1 Introduction ............................................................................................ 61 11.2 Methodology for investigation of seasonality ............................................... 63 11.3 Adjustment of waste data for seasonality ................................................... 68

    12.0 Investigation into the impact of omission of flats from the detailed wastecompositional analysis ......................................................................................... 72 13.0 Uncertainty ................................................................................................. 77

    13.1 Description of uncertainties associated with estimates in main report ........... 77

    13.2

    Calculation of confidence intervals ............................................................. 84

    13.3 Confidence intervals by food type .............................................................. 86

    Appendix 1: Costing food and drink ..................................................................... 90 Appendix 2: Peer review of methods used ........................................................... 97

    AcknowledgementsThe authors are grateful to the peer reviewer, Andrew Davey (WRc) for useful andconstructive comments on the methods and this report.

    Keith James is thanked for his contribution to the chapter on environmental impacts ofhousehold food and drink waste.

  • 8/12/2019 Methods Annex Report v2

    5/103

    Methods used for Household Food and Drink Waste in the UK 2012 5

    Glossary Avoidable waste a classification used in the report relating to food and drink thrown

    away that was, at some point prior to disposal, edible, e.g. milk, lettuce, fruit juice, meat(excluding bones, skin, etc.); c.f. possibly avoidable and unavoidable waste.

    Defra Department of the Environment, Food and Rural Affairs

    Down the Drain A study classifying and quantifying household food and drink wastebeing disposed of down the sewer. Published in November 2009, based on 2008 fieldwork(http://www.wrap.org.uk/sites/files/wrap/Down%20the%20drain%20-%20report.pdf )

    Family Food statistics a survey commissioned by Defra quantifying food and drinkpurchases in the UK (https://www.gov.uk/government/organisations/department-for-environment-food-rural-affairs/series/family-food-statistics )

    Household Food and Drink Waste in the UK published in November 2009, this reportrestated the information in The Food We Waste (i.e. it presented revised 2007 estimates).It included diary-based estimates for waste going down the sewer, home composted or

    fed to animals(http://www.wrap.org.uk/sites/files/wrap/Household%20food%20and%20drink%20waste%20in%20the%20UK%20-%20report.pdf )

    LA Local authority McCance and Widdowson's The Composition of Foods (2002) publication by the Foods

    Standards Agency and Royal Society of Chemistry detailing information about thenutritional composition of food and drinks and other useful information.

    NRS - National Readership Survey ONS Office of National Statistics

    Possibly avoidable waste a classification used in the report relating to food and drinkthat some people eat and others do not, e.g. bread crusts and potato skins; c.f. avoidableand unavoidable waste.

    Synthesis of Food Waste Compositional Data 2010 previous research collatinginformation from existing local authority waste audits and WasteDataFlow(http://www.wrap.org.uk/sites/files/wrap/Synthesis%20of%20Food%20Waste%20Compositional%20Data%202010%20FINAL.pdf )

    Synthesis of Food Waste Compositional Data 2012 current research published alongsidethis report collating information from existing local authority waste audits andWasteDataFlow (available from: www.wrap.org.uk/household-food-waste )

    The Food We Waste (TFWW) original report by WRAP on household food and drinkwaste, published in 2008 and based on 2007 fieldwork (available on request from WRAP)

    Unavoidable waste a classification used in the report relating to waste arising from foodand drink preparation that is not, and has not been, edible under normal circumstances,e.g. meat bones, egg shells, pineapple skin, tea bags; c.f. avoidable and possiblyavoidable waste.

    WDF WasteDataFlow, a reporting system for waste collected by local authorities in theUK(http://www.wastedataflow.org )

    WRAPs regular household food waste questionnaire WRAP conducts a consumer surveyon household food and drink waste twice a year. This is currently conducted using on-linepanels with a sample that is representative of Great Britain.

    http://www.wrap.org.uk/sites/files/wrap/Down%20the%20drain%20-%20report.pdfhttp://www.wrap.org.uk/sites/files/wrap/Down%20the%20drain%20-%20report.pdfhttp://www.wrap.org.uk/sites/files/wrap/Down%20the%20drain%20-%20report.pdfhttps://www.gov.uk/government/organisations/department-for-environment-food-rural-affairs/series/family-food-statisticshttps://www.gov.uk/government/organisations/department-for-environment-food-rural-affairs/series/family-food-statisticshttps://www.gov.uk/government/organisations/department-for-environment-food-rural-affairs/series/family-food-statisticshttps://www.gov.uk/government/organisations/department-for-environment-food-rural-affairs/series/family-food-statisticshttp://www.wrap.org.uk/sites/files/wrap/Household%20food%20and%20drink%20waste%20in%20the%20UK%20-%20report.pdfhttp://www.wrap.org.uk/sites/files/wrap/Household%20food%20and%20drink%20waste%20in%20the%20UK%20-%20report.pdfhttp://www.wrap.org.uk/sites/files/wrap/Household%20food%20and%20drink%20waste%20in%20the%20UK%20-%20report.pdfhttp://www.wrap.org.uk/sites/files/wrap/Household%20food%20and%20drink%20waste%20in%20the%20UK%20-%20report.pdfhttp://www.wrap.org.uk/sites/files/wrap/Synthesis%20of%20Food%20Waste%20Compositional%20Data%202010%20FINAL.pdfhttp://www.wrap.org.uk/sites/files/wrap/Synthesis%20of%20Food%20Waste%20Compositional%20Data%202010%20FINAL.pdfhttp://www.wrap.org.uk/sites/files/wrap/Synthesis%20of%20Food%20Waste%20Compositional%20Data%202010%20FINAL.pdfhttp://www.wrap.org.uk/sites/files/wrap/Synthesis%20of%20Food%20Waste%20Compositional%20Data%202010%20FINAL.pdfhttp://www.wrap.org.uk/household-food-wastehttp://www.wrap.org.uk/household-food-wastehttp://www.wrap.org.uk/household-food-wastehttp://www.wastedataflow.org/http://www.wastedataflow.org/http://www.wastedataflow.org/http://www.wastedataflow.org/http://www.wrap.org.uk/household-food-wastehttp://www.wrap.org.uk/sites/files/wrap/Synthesis%20of%20Food%20Waste%20Compositional%20Data%202010%20FINAL.pdfhttp://www.wrap.org.uk/sites/files/wrap/Synthesis%20of%20Food%20Waste%20Compositional%20Data%202010%20FINAL.pdfhttp://www.wrap.org.uk/sites/files/wrap/Household%20food%20and%20drink%20waste%20in%20the%20UK%20-%20report.pdfhttp://www.wrap.org.uk/sites/files/wrap/Household%20food%20and%20drink%20waste%20in%20the%20UK%20-%20report.pdfhttps://www.gov.uk/government/organisations/department-for-environment-food-rural-affairs/series/family-food-statisticshttps://www.gov.uk/government/organisations/department-for-environment-food-rural-affairs/series/family-food-statisticshttp://www.wrap.org.uk/sites/files/wrap/Down%20the%20drain%20-%20report.pdf
  • 8/12/2019 Methods Annex Report v2

    6/103

    Methods used for Household Food and Drink Waste in the UK 2012 6

    1.0 Introduction

    This document contains details of the research methods and analysis used to obtain theresults in Household Food and Drink Waste in the United Kingdom 2012 . The research iscomprised of many elements, which are detailed below. The rationale for the use of thesemethods is given in Chapter 2.0. In addition, the analysis required to draw these elements ofresearch together to obtain national estimates of household food waste is described in

    Chapter 3.0. Elements of primary research:

    Synthesis of Food Waste Compositional Data 2012: This research collates recentwaste audits (waste compositional analyses) commissioned by local authorities. Theinformation from these audits is used alongside waste and recycling tonnage data fromWasteDataFlow a reporting system for waste collected by local authorities in the UK.The analysis gives an estimate of the total amount of household food and drink waste inlocal authority collected waste streams. Th e householders details were not linked to thewaste collected and it is not possible to identify the household within the published data.See separate report, available from: www.wrap.org.uk/household-food-waste , for more

    details of the methodology and report. Detailed waste compositional analysis: This study classifies and quantifies food and

    drink waste from approximately 1,800 households who gave their consent to be included.This gives detailed information on the types, avoidability and state of food wasted. At thepoint at which informed consent was requested, households completed a face-to-facesurvey similar to that undertaken as part of the kitchen-diary research. Participants wereasked to sign a consent form, giving permission to the project team to collect and analysetheir waste. Any information collected about the householder was treated as strictlyconfidential. The householders details were not linked to the wa ste collected and it is notpossible to identify the household within any published data. The detailed methodologyfor this part of the research is given in Chapter 4.0.

    Kitchen-diary 2012 research : Information on the types of food and drink wasted wasrecorded in diaries by 948 households. For each instance of waste, the amount of waste,the reason for disposal and which waste stream was used was recorded. The researchincluded waste streams that are hard to measure from compositional analysis (materialpoured down the kitchen sink or other inlets to the sewer, home composted or fed toanimals). Each household was also surveyed before and after the diary research via aquestionnaire, giving information on attitudes, stated behaviours and householdcharacteristics. The detailed methodology for this part of the research is given in Chapter5.0.

    Details of weighting factors and information sources for the number of households in the UKare provided in Chapter 7.0. How the retail price of food and drink waste was calculated isdescribed in Chapter 8.0, and calculation of the greenhouse gas emissions and land-usechange associated with this waste are given in Chapter 9.0. Chapter 10.0 describes howlevels of waste are compared to purchases.

    Chapters 11.0 and 12.0 investigate two potentials areas of uncertainty emanating from themethodology. Chapter 11.0 explores how seasonality can affect the results and develops amethod for adjusting for it. For practical reasons, flats were largely omitted from the samplefor the detailed waste compositional analysis and the impact of this omission on the results isinvestigated in Chapter 12.0. Chapter 13.0 details other contributions to uncertainty in the

    estimates presented in the main report.

    http://www.wrap.org.uk/household-food-wastehttp://www.wrap.org.uk/household-food-wastehttp://www.wrap.org.uk/household-food-wastehttp://www.wrap.org.uk/household-food-waste
  • 8/12/2019 Methods Annex Report v2

    7/103

    Methods used for Household Food and Drink Waste in the UK 2012 7

    The report and its methodology have been peer reviewed. The reviewer s comments can befound in Appendix 2 of this document.

    The definition of household food and drink waste, as given in the main report, covers anumber of disposal routes and waste streams. How these pieces of research, describedabove, were combined to obtain estimates of household food and drink waste in the UK issummarised in Table 1; an overview of the calculations is given in Figure 1, with more details

    of these calculations provided in Chapter 3.0. For each element of the results, the mostappropriate information was used, as discussed in Chapter 2.0.

    Table 1: Summary of information used to obtain estimates of food and drink waste in thecurrent report

    Estimate for food waste Amount of foodwastedType of food

    wastedReason for

    waste

    Collected by local authorities

    Synthesis of Food

    WasteCompositionalData 2012

    Detailed wastecompositionalanalysis

    Kitchen Diary2012 research

    Household sewer Derived fromDown the Drain Derived fromDown the Drain

    Kitchen Diary2012 research

    Home composted Kitchen Diary2012 research

    Kitchen Diary2012 research

    Kitchen Diary2012 researchFed to animals

    For waste going down the household sewer, it was also decided to use the information fromthe Down the Drain 1 research, rather than information from the Kitchen Diaries 2012 research as originally intended. This decision is described in 2.3.

    1 http://www.wrap.org.uk/sites/files/wrap/Down%20the%20drain%20-%20report.pdf

    http://www.wrap.org.uk/sites/files/wrap/Down%20the%20drain%20-%20report.pdfhttp://www.wrap.org.uk/sites/files/wrap/Down%20the%20drain%20-%20report.pdfhttp://www.wrap.org.uk/sites/files/wrap/Down%20the%20drain%20-%20report.pdfhttp://www.wrap.org.uk/sites/files/wrap/Down%20the%20drain%20-%20report.pdf
  • 8/12/2019 Methods Annex Report v2

    8/103

    Methods used for Household Food and Drink Waste in the UK 2012 8

    Figure 1: Overview of calculations to obtain estimates of household food and drink waste

    Local authoritycollected waste 3.1

    Sewer waste 3.2 Home composted orfed to animals 3.3

    Coded dataset from

    Kitchen Diary 2012 research

    Weighted estimates

    Amount of food anddrink by type

    Weighting

    Seasonaladjustment

    Coded dataset from

    detailed wastecompositional analysis

    Weighted estimates

    Seasonally adjustedestimates

    Weighting

    Seasonaladjustment

    Scaled to total inSynthesis research

    Amount of food anddrink by type

    Weighted resultspreviously calculatedfrom Down the Drain

    Seasonallyadjusted estimates

    Seasonaladjustment

    Scaled tochange in LA

    collected waste

    Amount of foodand drink by type

    Total amount of foodand drink by type

    Cost of food and drinkby type

    Greenhouse gasemissions of food and

    drink Reason for disposal by

    food and drink type

    Application of

    cost factors Application of

    carbon factors

    Data fromkitchen-diaryresearch 3.4

    Seasonally adjustedestimates

  • 8/12/2019 Methods Annex Report v2

    9/103

    Methods used for Household Food and Drink Waste in the UK 2012 9

    2.0 Rationale for source data used in calculations

    This chapter explains why certain methods were used for the different element of theresearch, as laid out in Table 1.

    2.1 Food and drink waste collected by local authorities

    There are a number of sources of data from which an estimate of the total amount of foodand drink waste collected by local authorities could be derived. These include:

    Combining information from WasteDataFlow and local authority waste compositionalaudits as undertaken in the Synthesis of Food Waste Compositional Data 2012 report.

    Scaling the information from the detailed waste compositional analysis: the averageamount of food and drink waste per household in these waste streams can be calculatedand scaled up to the total number of households in the UK.

    Scaling the average amount of food and drink waste per household in the kitchen diariesassociated with these waste streams to the whole UK population.

    The synthesis method has a number of advantages for calculating the total amount of foodand drink waste. Firstly, it uses a large number of compositional studies already performedby local authorities to derive waste estimates. This means that the effective sample size ofthe study is much higher (over 12,000 households) than either of the other two methods,and consequently sampling uncertainty in the results is much lower.

    Secondly, it uses information in WasteDataFlow in combination with waste audit data, thusmeaning that the estimates are partially based on a census of the total amount of material inrelevant waste streams. For streams where all the waste is likely to be food and drink (i.e.separate food waste collections), the information is not subject to sampling errors againthe uncertainty in the results is much lower.

    Another strength of the method is that waste audits undertaken by local authorities usuallyhave less contact with the households that they are sampling from (compared to either thedetailed waste compositional analysis or the kitchen diaries). This is possible because thelocal authority is unlikely to link the waste data to individual households and thereforehouseholds are rarely interviewed as part of the research. This means that the degree towhich households change their behaviour as a result of any contact is likely to be smaller.

    Finally, because the data does not need to be linked to households, it is easier for the wasteaudits to include flats and other shared properties, allowing the estimates to be morerepresentative of the UK population of all households.

    In contrast, diary methods suffer from under-reporting of the amount of food and drinkwaste. For the food element of the local authority collected waste streams (i.e. omittingdrink) a comparison between diary research and detailed composition analysis wasperformed on data from WRAP for 2007 2. This indicated that estimates from the diary werearound 40% lower than compositional analysis. The discrepancy appeared to be greater inhouseholds with more occupants and for waste classified as avoidable. The exact reasons forthis discrepancy are not known, but are likely to include both the diary changing behaviourand some wasted items not being reported (either consciously or inadvertently). Nocomparisons were possible for other waste streams as only diary data were available, withno alternative methods to compare against.

    2 SB Hj: 2011, 'Metrics and measurement methods for the monitoring and evaluation of household food waste preventioninterventions', M.Bus thesis, University of South Australia, Adelaide.

  • 8/12/2019 Methods Annex Report v2

    10/103

    Methods used for Household Food and Drink Waste in the UK 2012 10

    For the above reasons, information from Synthesis of Food Waste Compositional Data 2012 was chosen to provide the total amount of food and drink waste collected by local authoritiesfrom UK households. However, this synthesis research does not contain details of theamount of different types of food and drink or why it was thrown away.

    Although the detailed waste compositional analysis suffers from some drawbacks in

    obtaining the total amount of food and drink waste, these drawbacks are less importantwhen being used to determine the proportion of different types of food and drink thrownaway. For instance, the omission of flats from this research is explored in Chapter 12.0 andfound to have minimal impact on the results. Furthermore, it does not suffer from the under-reporting issues associated with diary research given that diaries appear to under-reportavoidable food and drink waste more than unavoidable waste, use of the diaries could skewthe results accordingly.

    Despite the shortcomings of diary-based research, the kitchen diaries are the only reliablesource of information on the reasons why food is thrown away. In research from 2007, thesample size of the Kitchen Diary 2007 research was smaller (just under 300 participants)

    compared to that undertaken in 2012 for this report (948 participants). This meant thatthere were insufficient instances of all commonly-wasted food types to be able to report thereasons for disposal accurately in the 2007 research, and information was inferred from thestate of the waste in the local authority collections. However, the increased sample size inthe 2012 research means that the diary information can be used for the reasons for disposal.The kitchen diaries from 2012 have therefore been used to estimate reasons for disposal forall waste streams in the research 3.

    2.2 Food and drink waste home composted and fed to animals

    Diary-based research remains the only feasible method for obtaining information on theamount of food and drink waste that is home composted or fed to animals. Other methodshave been considered (e.g. compositional analysis of a compost heap), but major problemsexist with such methods (e.g. in the example above, knowing when food was placed on thecompost heap).

    For this reason diary-based estimates are used in this research for these two disposal routes. Although local authority collected waste suffers from a relatively high degree of under-reporting in diary research, the level of under-reporting for home composting and food fedto animals is not known, as there are few (if any) comparable estimates from which tocalculate it. The degree of under-reporting may be substantially less than for local authoritycollected waste: if diaries influence behaviour, then there is the possibility that material isdiverted towards disposal routes that are seen as socially desirable (such as homecomposting), which could counteract any incompleteness of diary entries.

    2.3 Food and drink waste going down the drain

    At the outset of the current research, the intention was to obtain estimates for the amountof food and drink waste going down the sewer from the kitchen-diary research in May 2012.Scaling the waste going to the sewer reported in the Kitchen Diary 2012 research gave anestimate of 490,000 tonnes (to two significant figures) using the methodology described in

    3

    It is possible that the under-reporting of avoidable has an impact on the reasons for disposal, if there are differences in theunder-reporting for different reasons for disposal. There is no direct evidence to quantify this (due to no comparable studiesanywhere in the world), and it has not been explored in this report.

  • 8/12/2019 Methods Annex Report v2

    11/103

    Methods used for Household Food and Drink Waste in the UK 2012 11

    Chapter 5.0 and 7.4. This methodology includes weighting for the number of occupants inthe household and for the presence and type of food waste collection. It also includesadjusting for seasonal trends in purchasing.

    This level of waste is approximately four times smaller than the estimate of 1.9 milliontonnes obtained from the Down the Drain research (fieldwork March 2008) 4. This differenceis substantial and warranted further investigation, which is described in this section.

    In particular:

    How does the difference manifest itself? Are there differences in the type of waste beingthrown away? Are there fewer instances of waste in 2012, or are there a similar numberwith each weighing less?

    Is there any other corroborating information that suggested a marked reduction in theamount of food and drink waste going down the sewer? For instance, had there been anychanges to purchasing patterns for products commonly poured down the sink such asmilk, drinks and sauces? Have there been changes to behaviours in or around the homethat could have caused a change on this scale?

    Is there any suggestion from the data that this difference is due to differences in researchmethods employed?

    The main elements of Down the Drain and the Kitchen Diary 2012 research methods aredescribed in Table 2. Both the Kitchen Diary 2012 and the Down the Drain research usedsimilar methodologies with diary keeping starting mid-week and lasting 7 days. In bothstudies participants were contacted on at least 3 occasions during the 7 days to ensure theywere happy with what they were doing and ensure interest levels were maintained.

    Two major differences that may have affected the total amount of waste are the time ofyear the research was conducted ( 2.3.1) and which disposal routes were covered by theresearch ( 2.3.3) . These are explored in the following sections alongside potential effectsfrom the sample profiles ( 2.3.4) and whether actual changes in sewer waste were likely(2.3.5) .

    4

    The original estimate published was 1.8 million tonnes ( http://www.wrap.org.uk/sites/files/wrap/Down%20the%20drain%20- %20report.pdf ) . When adjusted for seasonality to make it consistent with the 2012 estimate, there was a slight increase in theestimate to 1.9 million tonnes (to 2 significant figures).

    http://www.wrap.org.uk/sites/files/wrap/Down%20the%20drain%20-%20report.pdfhttp://www.wrap.org.uk/sites/files/wrap/Down%20the%20drain%20-%20report.pdfhttp://www.wrap.org.uk/sites/files/wrap/Down%20the%20drain%20-%20report.pdfhttp://www.wrap.org.uk/sites/files/wrap/Down%20the%20drain%20-%20report.pdfhttp://www.wrap.org.uk/sites/files/wrap/Down%20the%20drain%20-%20report.pdfhttp://www.wrap.org.uk/sites/files/wrap/Down%20the%20drain%20-%20report.pdf
  • 8/12/2019 Methods Annex Report v2

    12/103

    Methods used for Household Food and Drink Waste in the UK 2012 12

    Table 2 : Summary of Down the Drain and Kitchen Diary 2012 research, includingdifferences in research methods

    Down the Drain Kitchen Diary 2012 Research

    MethodEntries for individual items ofwaste in structured diary

    Entries for individual items ofwaste in structured diary

    Fieldwork duration 7 days 7 days

    Fieldwork start date Wednesday, 12 th March 2008 Thursday, 24 th May 2012

    Geographical locationof fieldwork Great Britain England and Wales

    Sample size 319 948

    Sample type

    Quota sample designed to berepresentative of the UKpopulation on the followingvariables:

    Household composition; Age; Housing type and tenure; Ethnicity; Employment status; and

    Rurality.

    Quota sample designed to berepresentative of the UKpopulation on the followingvariables:

    Household size andcomposition;

    Gender and age; Employment status; Ethnicity; Housing type; Household activity for

    composting food;

    Presence and type ofcouncil food wastecollections; and

    Rurality.

    Coverage (types ofitem) Food and drink Food and drink

    Coverage (disposalroutes) Sewer-related

    Sewer-related, residualwaste, council collectionstargeting food waste, homecomposting, fed to animals

    Method of estimationUse of measuring jugs,measuring spoons andestimation by respondent

    Use of measuring jugs,measuring spoons andestimation by respondent

    Table 3 illustrates that a major difference between the two studies is the number ofinstances of food and drink waste disposed of down the sewer, this is around six timeshigher in the Down the Drain research compared to the Kitchen Diary 2012 research. Theaverage weight of each instance of waste is slightly higher in the Kitchen Diary 2012 research.

    These differences between the two studies are consistent with Down the Drain respondentsrecording a greater proportion of instances of waste in their home, possibly due to them onlyhaving to record food and drink waste relating to one disposal route. In contrast, the Kitchen

  • 8/12/2019 Methods Annex Report v2

    13/103

    Methods used for Household Food and Drink Waste in the UK 2012 13

    Diary 2012 respondents may have a higher level of under-reporting due to the requirementto report all disposal routes.

    Table 3 : Descriptive statistics relating to number and average weight of instances of waste:Down the Drain and Kitchen Diary 2012 research

    Down the Drain (2008)

    Kitchen Diary 2012 Research

    UK estimate of food and drink waste tosewer for year of research 1,900,000 tonnes 490,000 tonnes

    Number of instance of waste perhousehold per week 24 3.9

    Average weight of waste per instance 63 grammes 94 grammes

    The following sections contain discussion the potential differences between seasons and theimpact of the different methodologies used.

    2.3.1 Seasonal effects

    Due to external constraints on the research, it was not possible for the Kitchen Diary 2012 fieldwork to occur at the same time of year as for Down the Drain . Given that there are veryfew pieces of research quantifying the amount of food and drink disposed of to the sewer,the seasonal variation in waste is not well understood. Therefore, the effect of this differencein time of year for the fieldwork is not known definitively.

    However, analysis has been performed as part of the current study to investigate theseasonal variation in household purchases of different types of food and drink. Chapter 8.0provides further details on the approach taken to investigating seasonality of food purchases.

    Although there may be differences between the seasonal patterns of waste and purchases, itis likely that there will be more waste for a given type of food when more of that food isentering the home. By using seasonal patterns in purchases as a guide for seasonality ofwaste, it is assumed that the proportion of purchases that are wasted is approximatelyconstant over the course of a year.

    Seasonal variations in purchases are shown in Figure 2 for food and drink types important to

    sewer waste. Soft drinks show increased purchasing in summer months, while increasedpurchases of tea, coffee and soup are associated with winter months. Alcoholic drinks have apeak in purchases around Christmas, whilst breakfast cereals show the converse (albeit at aweaker level). Milk shows little variability over the year.

    Given that the estimates of food and drink waste to sewer have been adjusted forseasonality in purchases (Chapter 11.0) and that the difference occurs for both unadjustedand adjusted estimates, it is highly unlikely that seasonality is contributing substantially tothe discrepancy under investigation. There may be some seasonal effects not picked up bythe method used (e.g. where the proportion of purchases that are wasted for a certain typeof food vary greatly within a year), but these are unlikely to explain a four-fold change in theestimate of food and drink waste going down the sewer.

  • 8/12/2019 Methods Annex Report v2

    14/103

    Methods used for Household Food and Drink Waste in the UK 2012 14

    Figure 2 : Seasonal trends in purchasing of food and drink types important to sewer waste.Note: y- axis has scales of different magnitude.

    Error bars show 95% confidence intervals of the mean amount of food purchased

    Source: Family Food Survey, Defra, 2001-2011

    2.3.2 Weather effects

    The seasonality data described in Figure 2 looks at the variation throughout the yearaveraged over 2001-2011. However, there may be impacts from seasonal factors (such asweather and festivals) that are different each year. For instance, the weather during theweek of the Kitchen Diary 2012 research (24 th 31 st May 2012) was unseasonably warm inthe UK. This is illustrated in Figure 3, which shows the average temperature in centralEngland for May 2012 in comparison to the long-term average for May (1981-2010). Thefirst-half of the week of the research was especially warm, with temperatures 5-6C warmerthan the average.

  • 8/12/2019 Methods Annex Report v2

    15/103

    Methods used for Household Food and Drink Waste in the UK 2012 15

    Figure 3 : Comparison of temperatures: May 2012 and long-term average (1981-2010) forthat month; Central England dataset 5

    The magnitude of temperature anomaly seen towards the end of May 2012 did not occurduring the Down the Drain research (March 2008), with temperatures within 3C of theseasonal average (Figure 4) .

    Figure 4 : Comparison of temperatures: March 2008 and long-term average (1981-2010) forthat month; Central England dataset

    The temperature anomaly in May 2012 could have had an impact on food waste arisings. Forinstance, hotter than average temperatures may mean that items purchased with coolerweather in mind (e.g. joints of meat and accompanying root vegetables for roasting; Figure5) are less likely to be eaten and additional purchases of foods consumed during hotterweather (e.g. salads, meats suitable for barbeques such as burgers and sausages) are made.

    5 Data from: Parker, D.E., T.P. Legg, and C.K. Folland. 1992. A new daily Central England Temperature Series, 1772-1991. Int.J. Clim., Vol 12, pp 317-342 (PDF) (downloaded from http://www.metoffice.gov.uk/hadobs/hadcet/cetdl1772on.dat )

    http://www.metoffice.gov.uk/hadobs/hadcet/cetdl1772on.dathttp://www.metoffice.gov.uk/hadobs/hadcet/cetdl1772on.dathttp://www.metoffice.gov.uk/hadobs/hadcet/cetdl1772on.dathttp://www.metoffice.gov.uk/hadobs/hadcet/cetdl1772on.dat
  • 8/12/2019 Methods Annex Report v2

    16/103

    Methods used for Household Food and Drink Waste in the UK 2012 16

    This could lead to higher levels of food and drink waste, especially for the cooler -weatherfoods.

    Figure 5 : Further examples of seasonal foods; top: winter foods; bottom: summer foods .Note: y- axis has scales of different magnitude.

    Error bars show 95% confidence intervals of the mean amount of food purchasedSource: Family Food Survey, Defra, 2001-2011

    But would the temperature anomaly in May 2012 have led to a lowering of the amount offood and drink being thrown down the sewer of the magnitude seen? If foods werepurchased in May 2012 in anticipation of cooler conditions that were subsequently not eatendue to the arrival of warmer weather, this would have led to higher levels of waste (ratherthan the lower levels seen). Furthermore, many of the highly seasonal foods are notroutinely disposed of down the sewer (e.g. those in Figure 5) . It therefore seems unlikelythat the weather during the research weeks explains any substantial proportion of the lowersewer waste estimates for May 2012.

    2.3.3 Scope of diary research: disposal routes included

    The other major difference between the methods of the two pieces of research is thecoverage of disposal routes: the Down the Drain research covered only sewer waste; theKitchen Diary 2012 research also covered four other routes (residual collections, councilcollections targeting food waste, home composting, and feeding to animals).

    It seems plausible that this difference in disposal routes included could have influenced thelevel of completeness of the diaries being kept. In essence, the amount of effort to recordthe waste to the other four disposal routes could have led to an under-reporting in the

    sewer-waste element of the Kitchen Diary 2012 research. This is despite people being asked

  • 8/12/2019 Methods Annex Report v2

    17/103

    Methods used for Household Food and Drink Waste in the UK 2012 17

    to record all instances of food and drink waste going down the sewer in both pieces ofresearch.

    Analysis of kitchen diary research and waste compositional analysis from 2007 showed that,where comparisons were possible (for local authority collected waste streams and food),under-reporting was around 40% 6. One of the explanations put forward for this under-reporting was related to the amount of effort required to fill in the diary completely 7. This is

    consistent with the observation that larger households had higher levels of under-reporting:the level of effort required to record all data from larger households is generally higher thanthat for smaller households.

    If this under-reporting were to occur, one might expect it to be more pronounced for: Smaller instances of waste that the diary respondent may not feel are important enough

    to report (given the effort involved). Items that are less obviously perceived as waste: drink waste going down the sewer may

    fall into this category.

    Table 3 indicates that the average size of an instance of waste was lower in Down the Draincompared to the Kitchen Diary 2012 research. This is found for some of the individual foodtypes shown in Figure 6: for milk and soft drinks the distribution of waste instances is shiftedto larger amounts for the Kitchen Diary 2012 research . For fruit juice, there is little differencebetween the two distributions, whilst for tea the Kitchen Diary 2012 research appears tohave more instances of waste in the central range. (The absence of large amounts of teawaste could be connected with fewer pots of tea being made in hot weather).

    6 SB Hj: 2011, 'Metrics and measurement methods for the monitoring and evaluation of household food waste preventioninterventions', M.Bus thesis, University of South Australia, Adelaide.7 Other reasons also cited in H js study related to behavioural reactivity and social desirability bias.

  • 8/12/2019 Methods Annex Report v2

    18/103

    Methods used for Household Food and Drink Waste in the UK 2012 18

    Figure 6 : Histograms showing the number of instances of waste for amounts of wasteDown the Drain and Kitchen Diary 2012 research for common drinks-related food types.(Note the non-linear categories on the horizontal axes).

    This under-reporting has the potential to occur for all waste streams reported in the KitchenDiary 2012 research. However, under-reporting could be more pronounced for sewer wasteas many people are less aware of drink-related waste and that poured down the sewer 8:therefore, a lower proportion of waste might be recorded in the diary for sewer relatedwaste compared to other disposal routes. Further discussion of the use of diary data for foodand drink waste estimates is discussed in 2.1 and 2.2.

    In the post-diary questionnaire, 6% of respondents stated that they didnt record everythingthat they threw away in the diary if the amounts were quite small or a fiddle to record. Given

    a social desirability bias, the actual number who knowingly didnt record all their waste mayhave been higher, and there could also have been other instances of non-reporting of wastethat was not consciously noted by respondents. However, we dont have this information forthe Down the Drain research to compare these results against.

    2.3.4 Differences in sample profile

    It is possible that, were there to be substantial disparities in sample profile between the twopieces of research, these could account for the differences in sewer waste estimates.

    8

    For example, in WRAPs regular household food waste questionnaire , only 4% of people claim to throw away food down thedrain: however, the majority of people in the Kitchen Diary 2012 research recorded at least one instance of material beingdisposed of via this route.

  • 8/12/2019 Methods Annex Report v2

    19/103

    Methods used for Household Food and Drink Waste in the UK 2012 19

    The variable that correlates most strongly with food and drink waste levels in the UK isnumber of people in the household (or household size). Both studies weighted the data toensure that the estimate is representative of the distribution of household sizes within theUK, therefore eliminating it as a cause of a discrepancy.

    Once household size has been accounted for, few socio-demographic variables correlate

    strongly with the amount of waste generated in a household. However, age correlates withwaste levels, with over 65s wasting less than the rest of the population. However, the twopieces of research had similar proportions of over 65s as the main diary keeper 9, so this willnot have had any substantial effects on the results.

    Other variables that could correlate strongly with food waste levels (e.g. strength ofrespondents resolution to reduce food waste) were not measured consistently between thetwo pieces of research, and hence it is difficult to make comparisons.

    Given the above, it seems unlikely that the profile of respondents has had more than anegligible impact on the sewer waste estimates, and certainly nothing that could contribute

    strongly to a four-fold difference in results.

    2.3.5 Actual reduction in food and drink being disposed of to the sewer

    It is, of course, possible that the difference between the Down the Drain and the KitchenDiary 2012 research estimates represents an actual reduction in waste going down the sewerbetween 2008 and 2012. There is no other data source known to the authors that measuresfood and drink waste to the sewer which could be used to verify this proposition. However,this section explores various information sources to investigate the likelihood that a four-foldreduction in food and drink waste to sewer has occurred.

    Many of the behaviours that are associated with food and drink waste lead to different itemsof waste that are commonly disposed of via different disposal routes. For instance, poorstock control of food and drink in the fridge may lead to some waste that is commonlyplaced in the bin (mouldy cheese) and other waste that is poured down the kitchen sink(sour milk). The disposal route is often determined by what the food is (with liquidcommonly going down the drain).

    Given this, if there has been a drop of the magnitude seen in waste going down the sewerand this was largely due to waste prevention behaviours, we would expect to see similardrop in waste going to other disposal routes. However, for waste collected by localauthorities (where we have independent data sources the local authorities studies) thereduction over a similar time period (cross reference 2007 to 2012) is much smaller (15%c.f. around 75% for sewer waste). This evidence alone makes it unlikely that the differencein sewer waste figures is entirely related to a change in behaviour.

    Information from WRAP s regular household food waste questionnaire 10 (Figure 7) gives noevidence of a decrease in the proportion of people who state that they dispose of food anddrink waste down the drain. This is consistent with similar trends in the amount of food anddrink waste going down the sewer compared to other disposal routes.

    9 27% for Down the Drain; 21% for Kitchen Diary 2012 research values once weighted has been applied. Note discussion inChapter 7.0 relating to difficulties in using age in sampling.10 WRAP conducts a consumer survey on household food and drink waste twice a year. This is currently conducted using on-linepanels with a sample that is representative of Great Britain (effective sample size = 2,600-2,900, dependent on wave).

  • 8/12/2019 Methods Annex Report v2

    20/103

    Methods used for Household Food and Drink Waste in the UK 2012 20

    Figure 7: Proportion of respondents stating that they throw away food waste to each of thedisposal options suggested.

    Source: WRAPs regular household food waste questionnaire . Question asked: Thinking about when you havethrown food away, how i s it disposed of in your home? Multiple responses are allowed, hence the results do not

    add to 100%. Dotted lines indicate a change in mode from face-to-face to on-line panel.

    It is also possible to look at food and drink purchasing data for evidence that supports orrefutes this idea of similar trends between food and drink waste going down the sewercompared to other disposal routes. Table 4 contains spending data from the Family FoodSurvey for 2008 and 2011 11. Similar reductions in purchases are seen between 2008 and2011 for food ( 1.5%) and drink (including milk; 1.6%). Again, this is broadly consistentwith similar trends in waste for the sewer and other disposal routes. For food and drinkcategories commonly thrown away down the sewer, there are few food types for whichstatistically significant trends can be seen: only milk and tea show significant changes, andthese show 3.8% and 7.8% reductions respectively: relatively small compared to the four-fold difference in sewer waste.

    11 The ideal comparison would have been between 2008 and 2012 data. However, at the time of writing, the most recent datapublished were for 2011.

  • 8/12/2019 Methods Annex Report v2

    21/103

    Methods used for Household Food and Drink Waste in the UK 2012 21

    Table 4: Trends in food and drink purchases between 2008 and 2011 (figures quoted aregrammes or millilitres per person per week)

    Food type Units 2008 2011 Change Statisticallysignificant?

    Drink and Milk g 4,763 4,687 1.6% Data notavailableFood g 6,859 6,757 1.5%

    Milk (whole, semi-skimmed,skimmed)

    ml 1,565 1,506 3.8% Yes

    Tea (as purchased) g 30 27 7.8% Yes

    Coffee (beans, ground andinstant)

    g 17 18 +4.2% No

    Soft drinks (excluding fruit juice)

    ml 1,682 1,630 3.1% No

    Alcoholic drinks ml 706 728 +3.2% NoFruit juices ml 325 307 5.7% No

    Breakfast cereals g 130 132 +2.1% No

    Soup g 81 80 1.1% NoSource: Family Food Survey, Defra

    Given all of the above, it is unlikely that these modest changes in purchasing have led to afour-fold reduction in waste going down the sewer. Furthermore, no evidence can be foundthat suggests that people have shifted away from using the sewer as a disposal route forfood and drink waste.

    2.3.6 Conclusions relating to differences in sewer waste estimates

    It is not possible to draw firm conclusions as to why the Kitchen Diary 2012 researchestimate for sewer waste is approximately four-times smaller than that from the Down theDrain research. However, it seems more likely that it is connected to methodologicaldifferences rather than reflecting an actual reduction of waste of this magnitude. Inparticular, the difference could largely stem from the difference in number of disposal routescovered by the two pieces of research.

    If this is the case, it seems likely that the Down the Drain estimate is more accurate than theKitchen Diary 2012 research the lower estimate in the latter is likely to be attributable to ahigher level of under-reporting than in the former.

    The general lessons that can be learned from this section are: Comparisons over time require sufficient data collected with a consistent method. This is

    particularly important where there are no other data sources against which estimates canbe checked (as is the case for food and drink waste going down the sewer).

    This is especially important for diary research, where under-reporting is a known issueand the level of under-reporting is affected by the methodology. To minimise under-reporting in similar diary research, constraining the scope of the diary (to a limited

    number of disposal routes) may increase the accuracy of the estimates.

  • 8/12/2019 Methods Annex Report v2

    22/103

    Methods used for Household Food and Drink Waste in the UK 2012 22

    It should be noted that constraints on the current research made it difficult to commission aseparate diary exercise focusing only on sewer waste (alongside the other elements ofresearch). However, this has made comparison over time problematic and the followingoptions are available to WRAP for estimating sewer waste in 2012:

    1. Use the data from the Kitchen Diary 2012 research as it is. This has consistency ofmethodology within the current project, but means that a comparison with 2008 sewerdata is likely to be misleading due to the methodological reasons described above.

    2. Use the Down the Drain research data for a 2012 estimate without adjustment. This hasthe implicit assumption that changes in the amount and type of sewer waste arenegligible between 2008 and 2012.

    3. Assume that sewer waste has fallen proportionately to local authority collected wastebetween 2008 and 2012. This assumption is based on an argument of logic thatchanges in food waste prevention behaviour will have a similar effect irrespective of thedisposal route and that other external influences will have had negligible impact on theamount of waste going to the sewer. This is consistent with the similar trends inpurchases between food and drinks (including milk) between 2008 and 2011. However,this empirical evidence is indirect and there is little other empirical data to substantiate

    (or contradict) this assumption.

    None of these options is ideal. On balance, it was decided to apply option 3, as this appearedto be most consistent with other (indirect) evidence relating to the amount of food and drinkwaste going down the sewer most notably trends in purchasing (Table 4) .

    Given this decision, it was also decided to use the information from the Down the Drain research to obtain an estimate of the amount and types of food being thrown away.However, the reasons for disposal still come from the Kitchen Diary 2012 research to allowconsistency across the waste streams.

    Finally, it should also be remembered that sewer waste accounts for a minority (around22%) of food and drink waste coming from UK homes. Although it is important to obtaindata that accurately reflects the amount of waste going to the sewer, the results from otherwaste streams especially that collected by local authorities, for which better quality dataexist form a greater part of the food and drink waste estimates.

  • 8/12/2019 Methods Annex Report v2

    23/103

    Methods used for Household Food and Drink Waste in the UK 2012 23

    3.0 Overview of calculations

    This chapter details how the calculations were performed to obtain estimates for HouseholdFood and Drink Waste in the United Kingdom 2012 . The calculations are different forindividual waste streams and an overview is given in Chapter 1.0 (Table 1 ). Details for eachwaste stream are given in the sections below.

    3.1 Calculations for local authority collected wasteFor the local authority collected waste, the detailed information by food type comes from thedetailed waste compositional analysis. Once the dataset was coded, the results wereweighted by number of occupants in the household as described in 7.3. This led to anaverage amount of food waste per household and this was scaled to the total number ofhouseholds in the UK as detailed in 7.2.

    These figures were then seasonally adjusted using the method outlined in 11.3. Seasonalfactors were calculated to convert information collected in one period of the year to anannual figure.

    The total amount of food and drink waste was then scaled so the total amount is equal tothe estimate from the Synthesis of Food Waste Compositional Data 2012 . This scaling wasnecessary as the latter research provides the more robust estimate of the total amount offood and drink waste in local authority collected waste streams (see 2.1 for more details).The scaling factor used was 1.23, i.e. estimates in the detailed compositional analysisneeded increasing by 23%. This is similar to the scaling factor used for 2007 estimates inHousehold Food and Drink Waste in the UK (published in 2009). The reasons for thisdiscrepancy are discussed in Chapter 13.0 alongside other uncertainties in the results.

    Households that were interviewed and consented to have their waste collected but did nothave any waste collected and analysed were excluded from the sample. This decision wasmade because the reasons for the waste not being collected were varied. For instance, somehouseholds did not present waste at kerbside for reasons that are unknown. For others, itwas not possible to link the waste to a household; this happened when bins and bags wereplaced or piled together making identification impossible. Had the reasons all beenconnected with non-presentation of waste (inferring a waste prevention effect), then therewould have been advantages to leaving the households in the sample, but recording theirfood and drink waste as zero. However, as this wasnt always the case, they were excluded .

    3.2 Calculations for sewer waste

    Given the discussion in 2.3, the calculations for food and drink waste going down the sewerwere not as originally envisaged. The method used for calculating 2012 estimates involvedthe following stages:

    Obtain seasonally-adjusted results for 2007 by taking the original calculations for sewerwaste (as published in 2009 in Household Food and Drink Waste in the UK ) and applyingseasonal-adjustment factors as described in 11.3.

    Obtain 2012 estimates by scaling the 2007 estimates according to the change in food anddrink waste collected by local authorities between 2007 and 2012 12, as detailed in 2.3.6. This assumes that the change in sewer waste has been proportional to the change in localauthority waste over the same period.

    12

    The fieldwork for the sewer waste estimate was undertaken in early 2008. However, in the past it has been treated as a 2007estimate despite being undertaken just after the end of 2007. For the purposes of adjustment, it has also been treated as a2007 estimate, hence the use of adjustment factors relating to local authority collected waste between 2007 and 2012.

  • 8/12/2019 Methods Annex Report v2

    24/103

    Methods used for Household Food and Drink Waste in the UK 2012 24

    These estimates exclude water that was added to food and drink products in the home, suchas water to dilute squash, or make tea or coffee. This is consistent with previous researchand an estimate of this excluded water is given in the main report.

    3.3 Calculations for home composting and fed to animals

    For food and drink waste that is home composted or fed to animals, data from the KitchenDiary 2012 research was used. As described in 7.4, weighting factors relating to thenumber of occupants in the household and the presence and type of food waste collectionswere applied and the results scaled to the total number of households in the UK. Theseresults were then seasonally adjusted using the method in 11.3.

    The following section describes details of the processing of the kitchen-diary data prior toweighting and seasonal adjustment.

    3.3.1 Weight of items

    Where items were recorded in the diary as a volume or the number of items, these requiredconverting to weight. Factors used in the Down the Drain and Kitchen Diary 2007 researchwere used. For diary entries recorded as the number of items, examples of these items werepurchased from supermarkets and weighed to inform the estimate. Small, medium and largeitems were purchased and applied as indicated in the item description; where no descriptionwas given, the weight of a medium-sized item was used.

    In both cases, this conversion will provide reasonable estimate of the weight of items, butthe process will introduce uncertainty. This is discussed alongside other uncertainty inChapter 13.0. However, only 15% of items were recorded as number of items, so the impactof the factors used in the calculations will be relatively small.

    3.3.2 Edible portion

    To calculate the weight of partial items such as apple cores or peel from a single orange,data from McCance and Widdowson's The Composition of Foods (2002) has been used. Thisreference includes factors on the edible portion of food types; for example, it states that66% of a banana is edible, with the remaining 34% the peel (Table 5) . These factors areapplied to the total weight of an item to arrive at estimates for apple cores, peel from asingle orange, etc.

    The McCance and Widdowson inedible fractions differ from those used in The Food WeWaste . For example in The Food We Waste it was assumed that a banana was 25% skin incomparison with 34% from McCance and Widdowson . The difference can be seen in thefollowing table. Information from McCance and Widdowson has been used as it providesauthoritative and comprehensive data for over 1,200 of the most commonly consumed foodsin the UK.

  • 8/12/2019 Methods Annex Report v2

    25/103

    Methods used for Household Food and Drink Waste in the UK 2012 25

    Table 5: Inedible portion of food types

    Food typePercentage of total weight

    McCance and Widdowson The Food We Waste

    Apple (core) 11% 20%

    Apple (peel) 13% 17%

    Banana (skin) 34% 25%

    Grapefruit (rind) 32% 20%

    Melon (rind) 34% 20%

    Nectarine (stone) 11% 20%

    Orange (peel) 30% 20%

    Peach (stone) 10% 20%

    Pear (core) 9% 20%Pear (peel) 21% 17%

    Pineapple (skin) 47% 17%

    Plum (stone) 6% 20%

    Satsuma (peel) 29% 20%

    Avocado (skin and stone) 29% 25%

    Broccoli (stalks) 39% 20%

    Cabbage (core and outer) 23% 14%

    Cauliflower (outer leaves) 55% 10%

    Carrot (peeling) 22% 17%

    Courgette (tops and tails) 12% 10%

    Cucumber (peel) 3% 17%

    Leek (peeling) 43% 20%

    Lettuce (leaves) 26% 2%

    Onion (peeling) 9% 20%

    Parsnip (peeling) 28% 17%Pepper (seeds) 16% 20%

    Potato (peeling) 20% 17%

    Spring onion (ends) 31% 10%

    3.4 Calculations relating to reason for disposal

    To calculate the amount of avoidable food and drink waste associated with each reason fordisposal, information was taken from the Kitchen Diaries 2012 . The information given in thediary as to why each item was thrown away was coded according to the five categoriesgiven below:

  • 8/12/2019 Methods Annex Report v2

    26/103

    Methods used for Household Food and Drink Waste in the UK 2012 26

    Not used in time; Cooked, prepared or served too much; Accidents (contaminated, burnt or spoilt); Personal preference; and Other.

    Estimates were obtained for each food type by calculating the amount of avoidable food and

    drink waste associated with each reason. These amounts were calculated by weighting thewaste records in the kitchen diaries by the factors calculated for each household (based onnumber of occupants and presence and type of food waste collections; 7.4) . From theseamounts, a percentage for each reason for disposal was obtained for each food type (e.g.73% of the standard bread waste is not used in time). These percentages were then appliedto the total amount of avoidable waste for that food type.

    If there were fewer than 30 instances of avoidable waste recorded in the diary for a givenfood type, information on reason for disposal is not given in the main report due to a highlevel of uncertainty in the result.

    3.5 Calculations relating to tea, coffee, hot chocolate and formula milk

    Although the calculations for the vast majority of food and drink types were undertaken inthe same way, there were a small number of exceptions relating to tea, coffee and hotchocolate.

    Unused tea bags and liquid tea that has not been drunk are both classified as avoidable,while used teabags are classified as unavoidable waste. For tea that was not drunk, the

    avoidable tag relates to all material used to create the tea: milk and sugar where applicable,and the tea bag or tea leaves.

    Instances of undrunk tea were only related to sewer disposal and an estimate is made of theweight of teabag, milk and sugar relating to these. Furthermore, the weight of the milk andsugar is estimated from the diaries, but not the tea bag or tea leaves, as these are disposedof via the residual bin or home composted (>80% of tea bags and tea leaves are disposed ofvia the residual bin). However, all of the used tea bags and tea leaves disposed of via theseroutes would have been classified as unavoidable, given that it would have been impossibleto tell if the cup of tea that they were used to make was drunk or not. Hence, a small butsubstantial proportion of the tea bags disposed of via the residual bin was reclassified asavoidable.

    The amount of tea disposed of via the sewer in 2012 was estimated at 0.99 million tonnes

    (including added water). At 325 ml per cup, this equates to the equivalent of 3.1 billion cupsof tea. If each cup of tea were made using one tea bag and the average tea bag weighs 5grammes when disposed of, then approximately 15,000 tonnes of the 405,000 tonnes ofused tea bags and tea leaves found in the compositional analysis are associated with theavoidable tea disposed of via the sewer. This 15,000 tonnes has been reclassified fromunavoidable to avoidable waste and the results in the main part of this report reflect this.

    There are similar issues for materials such as coffee, formula milk powder and hot chocolatepowder. For milk powder and hot chocolate, there is no material to reclassify fromunavoidable to avoidable associated with drinks that have not been drunk. For fresh coffee(i.e. made from largely insoluble grounds), there is a small amount of material thrown away

    that is associated with undrunk coffee which should be reclassified from unavoidable toavoidable; however, the amount in question is very small (and difficult to calculate as it is

  • 8/12/2019 Methods Annex Report v2

    27/103

    Methods used for Household Food and Drink Waste in the UK 2012 27

    not always known whether coffee is made from insoluble grounds as opposed to solublegranules or powder); consequently an adjustment has not been made.

    For the cost and greenhouse gas calculations for both tea and coffee, an estimate of theamount of milk, sugar, teabags / leaves, and coffee granules and cocoa has been madeagainst which cost and carbon factors have been applied.

    3.6 Calculations relating to carcass meat and bones

    The information available from the composition waste analysis was slightly more detailedthan previous research for bones and carcass meat. For instance, where a chicken carcasshas been thrown away, the proportion of meat left on the carcass was assessed visually andrecorded in the following categories:

    Majority meat (60%+); Around half meat (60%-40%); Some meat (40%-20%); A little meat (20%-10%); Very little meat (10%-1%); and None (

  • 8/12/2019 Methods Annex Report v2

    28/103

    Methods used for Household Food and Drink Waste in the UK 2012 28

    4.0 Methods for detailed waste compositional analysis

    4.1 Overview

    The overall aim of the waste compositional analysis was to create a set of cases, comprisinga sample of households, representative of the UK, including information from a questionnairerelated to food waste and, for the same households, waste compositional data relating to the

    types and quantities of food waste.The fieldwork was undertaken by Ipsos MORI (recruitment and questionnaire) and ResourceFutures (waste compositional analysis) during 2013. The two components were linked usinga unique identification number for each household. All households taking part in the wastecompositional analysis provided informed consent for their waste to be collected andanalysed.

    4.2 Sample design

    The survey sampled households along five collection routes in each of 12 local authoritiesselected for the research. The following sections describe how the local authorities, collectionroutes and households were selected to yield a sample that was representative ofhouseholds across the UK.

    4.2.1 Selecting local authorities to participate

    The fieldwork was undertaken in England and Wales, reflecting the budget for the project.However, the sample was selected to be representative of the whole UK (for both the localauthorities selected and the households within them). The effect of using data from Englandand Wales to obtain a UK estimate is discussed in Chapter 13.0.

    For the local authorities, a number of criteria were used to guide the selection: Include authorities with a mix of waste collection schemes especially residual waste and

    any collections targeting food waste that was representative of those in operationacross the UK;

    Include a range of deprivation levels and population density that is broadly representativeof the UK; and

    Include authorities in England and Wales.

    There was sufficient budget to sample from 12 local authorities, of which three were inWales and nine in England. The types of waste collection scheme for these authorities are

    shown in Table 6. As mentioned previously, while interviewing only took place in Englandand Wales, the survey was designed to represent the UK and selection of local authoritieswas, therefore, based on UK figures. For example, the proportion of local authorities in theUK operating a weekly food, fortnightly refuse collection service is 22%; the samplecontained 3 local authorities with this collection system, which equates to 25%.

    It is also important to note that because of major differences among the UK nations in theprevalence of food waste collections, it is not be possible to derive reliable estimates of foodwaste for individual nations. However, a robust estimate for the UK the purpose of thestudy is feasible.

  • 8/12/2019 Methods Annex Report v2

    29/103

    Methods used for Household Food and Drink Waste in the UK 2012 29

    Table 6: Number of local authorities split by waste collection scheme for the sample and theUK population

    Food wastecollection: Weekly Fortnightly No food

    collect-ion

    AllResidual

    collection: Fortnightly Weekly Fortnightly Weekly

    Sample (no.) 3 1 2 0 6 12

    Sample (%) 25% 8% 17% 0% 50% 100%

    UK population* 22% 8% 18% 3% 48% 100%*Data for the UK population from WRAPs audit of waste collections for 2012

    Information on population density and deprivation was appended to each of the 348 localauthorities in England and Wales. Population density was described by the number of peopleper hectare for each local authority. Local authorities were then grouped into three quantiles.Similarly, deprivation was calculated by appending the average score for the Index ofMultiple Deprivation (IMD) for each local authority and grouped into three quantiles.

    Any local authorities that were trialling new collection schemes or had more than onecollection scheme in place were removed from the list of candidate authorities. Theremainder were then stratified by UK nation, waste collection scheme (i.e. frequency of foodwaste and residual collections), density and deprivation and 37 local authorities wereselected at random to approach 13. The density and deprivation stratification was used to helpensure the survey sampled a representative selection of local authorities. This helped withthe next stage of the process: the selection of waste routes from within the participatinglocal authorities (see 4.2.2) .

    WRAP initially contacted as many of these 37 local authorities as possible, with Ipsos MORIfollowing up to obtain their permission to conduct the survey and waste composition analysisin their area. When approaching a local authority, an introductory email to a contact wassent first, before following up with a telephone call enabling Ipsos MORI to answer anyquestions or concerns they might have about the research. The telephone call was also usedto get a steer as to whether they would be willing to participate, and what furtherinformation or permissions they would like to see. The approach resulted in quotas for localauthorities being met.

    4.2.2 Selecting collection routes and properties

    Once the local authorities had agreed to participate in the survey, the project team gatheredinformation about the waste collection routes: number of houses, time and days of collectionand actual route of the waste collection round. From this information, Ipsos MORIdetermined which five routes, within each of the 12 local authorities, would be best toinclude in the study to ensure that the households included would be representative of theUK. The ultimate intention was to make the waste routes selected within a waste collectiontype as representative as possible (within the confines of the study) of the UK by rurality and

    13 LAs were selected based on a quota system (region and waste collection system) and an appropriate number were randomly

    selected for each quota, based on our expected inclusion rate. Further LAs were contacted in Wales and in the FortnightlyFood, Fortnightly Refuse waste collection type at a later date to ensure that some authorities with this waste type wereincluded in the sample.

  • 8/12/2019 Methods Annex Report v2

    30/103

    Methods used for Household Food and Drink Waste in the UK 2012 30

    social grade, and thereby negate the need for use of post-weighting to the dataset on thesevariables.

    In previous studies just two routes per local authority were selected to participate in theresearch. However, by increasing the number of routes from two to five it was possible togather the opinions of a wider range of areas at no additional cost. This helped ensure amore representative sample, reducing the need for weighting, and reducing uncertainty in

    the final results.The selection of waste routes was balanced by rurality and social grade (the AB profile 14), toensure they were representative within waste collection type. This was achieved usingpostcode information about each of the waste routes and using census data to appendinformation on rurality and social grade to each collection round. Where detailed postcodeinformation was not provided, a bespoke approach was used, including using informationfrom the local authority in question about the socio-economic profile of different areas withintheir authority.

    No flats or properties with shared waste collection were included in the research. For this

    research, it is necessary to link information about food waste to individual (yet anonymous)households to help establish the link between demographic information and waste. Thismeans that households for which waste cannot be identified (in a shared bin or wastereceptacle) were omitted from the research. This is consistent with previous research.Omission of flats was partially counteracted by weighting of the calculations (see 7.3) ,which takes into account that flats generally have fewer occupants than houses. However,this weighting does not take into account any other differences between flats and housesthat may impact food and drink waste generated. The potential impact of this aspect of themethodology has been assessed from other data sources including the current diary research(Chapter 12.0) and found to have minimal impact on the results.

    As noted above, social grade and rurality was controlled for through the selection of thewaste routes. In addition, quota targets were set for interviewers on a number of other keyfactors to ensure the sample of households was as representative as possible, as previousresearch has shown that they affect the amount of food and types of waste generated byhouseholds. The quota cells are shown in Table 7. It should be noted these are notinterlocked 15 and were based on population information for all households in the interviewerassignments selected. For example rather than setting a quota of two women aged 18 to 34who live on their own and dont have children , separate quotas were set requiring 6-9women, 2-5 people aged 18 to 34, 2-4 single person households and 4-6 households withchildren. It was felt that more complex quotas cells would have made the survey pointsunworkable.

    14 A well-established system, as well as the most widely known and used, is that of social grading, derived from the BritishNational Readership Survey (NRS). This splits social grades into six groups A, B, C1, C2, D and E.15 In thi s context, interlocked means that the variables used for quotas and weighting are considered together. Taking ahypothetical example, if interlocking quotas for age and gender were used, the sample would need to include a representativenumber of, say, 18-34 year old men. If the quotas were not interlocked, it would be sufficient to have, in the previous example,

    a representative number of men and a representative number of 18-34 year olds of either gender. Interlocking is importantwhere there is a correlation between factors (e.g. if a higher proportion of older people are female compared to other agegroups).

  • 8/12/2019 Methods Annex Report v2

    31/103

    Methods used for Household Food and Drink Waste in the UK 2012 31

    Table 7: Quota controls and cells for questionnaire survey

    Quota control Quota cells

    Age 18-34; 35-54; 55+

    Household size 1 person, 2-3 people, 4+ people

    Presence of children Yes (children aged 0-18), No (no children)

    Evidence suggests that well designed, properly executed quota sample surveys will deliverperfectly good and relatively unbiased survey estimates on most of the occasions they areused. Random probability methods should be used if it is very important to minimise biasacross all variables and / or to ensure that there cannot be substantial bias for any individualvariables. Quota sampling methods should be used where resources are constrained andwhere the risk of some (generally low level) bias is considered acceptable 16. This study useda quota sampling approach as it was possible to substantially increase the number of

    interviews within the pre-existing budget.

    4.3 Questionnaire survey fieldwork

    An in-home, face-to-face approach was adopted for the fieldwork. This was the sameapproach as for previous research ( The Food We Waste ), ensuring consistency in themethod used and helping to ensure high rates of response and consent. In particular, a face-to-face approach encourages consent to the waste composition analysis phase and allowsthe recruitment to this stage of research to be formalised.

    The questionnaire took no longer than 27 minutes to conduct. Survey fieldwork wasconducted during the day and in the evening, along with weekend and weekday shifts toensure that a cross-section of respondents participate in the research. Only those who areeither solely or mainly responsible for food shopping or food preparation were included.

    During the research it was identified that the consent rate and more importantly waste set-out rates were lower than had been expected and this was leading to lower numbers ofhouseholds participating in the waste compositional analysis than required. Therefore, anadditional 500 interviews were conducted so consent could be gained to ensureapproximately 2,000 households could be sampled for the compositional analysis. It was notpossible to evenly distribute these additional interviews across all 12 authorities taking partdue to the constraints of time, cost and the route sizes within some authorities (e.g. all theaddresses within the routes available had been contacted). Therefore, the additionalinterviews took place in one Welsh authority (70 interviews were conducted in Neath PortTalbot) and three English authorities (250 in Birmingham, 115 in Poole and 65 inScarborough).

    In total, 3,786 interviews were carried out across twelve local authority areas. Of these,2,660 consented to have their waste analysed. The fieldwork was undertaken between 16 th January 2013 and 24th June 2013. This avoided responses to the questions being affected

    16 Source: Is random probability sampling really better than quota sampling?, Patten Smith, Director of Research Methods,

    Ipsos MORI, 2008.( http://surveyresearch.weebly.com/uploads/2/9/9/8/2998485/is_random_probability_sampling_really_much_better_than_quota _sampling.doc )

    http://surveyresearch.weebly.com/uploads/2/9/9/8/2998485/is_random_probability_sampling_really_much_better_than_quota_sampling.dochttp://surveyresearch.weebly.com/uploads/2/9/9/8/2998485/is_random_probability_sampling_really_much_better_than_quota_sampling.dochttp://surveyresearch.weebly.com/uploads/2/9/9/8/2998485/is_random_probability_sampling_really_much_better_than_quota_sampling.dochttp://surveyresearch.weebly.com/uploads/2/9/9/8/2998485/is_random_probability_sampling_really_much_better_than_quota_sampling.dochttp://surveyresearch.weebly.com/uploads/2/9/9/8/2998485/is_random_probability_sampling_really_much_better_than_quota_sampling.dochttp://surveyresearch.weebly.com/uploads/2/9/9/8/2998485/is_random_probability_sampling_really_much_better_than_quota_sampling.doc
  • 8/12/2019 Methods Annex Report v2

    32/103

    Methods used for Household Food and Drink Waste in the UK 2012 32

    by the Christmas period. The effects of seasonality on food waste levels are discussed andquantified in Chapter 8.0.

    4.3.1 Questionnaire

    The questionnaire design drew in the most part from questions asked on WRAPs regular

    household food waste questionnaire. The questionnaire was administered as a paperquestionnaire, due to the flexibility it provides in covering the interviewer assignments. Clearinstructions were provided to the interviewer to ensure the data quality was as high aspossible.

    The questionnaire was also translated into Welsh and respondents in Wales were offered theopportunity to participate in Welsh. In other ethnic communities where the targetrespondents did not speak English, the interview was conducted through an interpreter(either another member of the household who speaks English or using Language LineSolutions: a translation service that offers telephone interpreting).

    The booking in and data processing of questionnaire returns was done through scanning andmanual verification. The scanning programme defines both hard checks (where values aredefinitely wrong) and soft checks (where potentially incorrect values are brought to theattention of the research team to correct).

    4.3.2 Obtaining informed consent

    Once householders had completed the survey, the option of the waste analysis work wasintroduced to them. If they agreed to participate, they were asked to sign a consent form,giving permission to the project team to collect and analyse their waste. Informed consentmeans that respondents a) know what participation in research entails, its purpose, how theresults will be used, and by whom, and b) agree to participating in the research. This wasexplained at the end of the interview by the interviewer, outlined in the consent form, aswell as in a supporting leaflet. The consent form and information leaflet was also translatedinto Welsh.

    Ipsos MORI drew upon its experience of designing supporting materials for other studies tohelp maximise the consent rate. This included using a letter with the loc al authoritys logowhich highlighted their support for the study. The performance of each interviewer wasmonitored, and best practice shared between interviewers, and additional support providedto any interviewers who were achieving a lower than expected consent rate.

    When the respondents agreed to take part in the study, their name, address and signaturewere collected on the consent form as evidence that they were willing to participate. Theirdata were then recorded on a separate sample list containing the addresses of allhouseholds that had agreed to have their waste collected. A copy of the consent letter wasleft with the householder along with the information leaflet, which gave information aboutthe project generally and how the waste collection worked. Care was taken in the phrasingof the leaflet to minimise any research effect on the respondents waste behaviour.

    It is important to recognise that a survey of waste disposal behaviour could affect actualbehaviour to some extent. In order to mitigate any research effect on the waste analysiswork, a period of at least two weeks was left between the end of the survey period and the

    collection of the waste analysis data, and for the vast majority of respondents this gap waslonger.

  • 8/12/2019 Methods Annex Report v2

    33/103

    Methods used for Household Food and Drink Waste in the UK 2012 33

    Importantly, the leaflet also contained a contact name, a direct free-phone telephonenumber and an e-mail address which recipients could use if they wished to contact someoneabout the survey or if they wished to withdraw from the study at any stage. The approachused resulted in only a very small number of those interviewed using the free-phone numberor email address for further information or to withdraw their consent for the compositionalanalysis. Any