Login for PhD students/staff at UCPH      Login for others
International PhD course: Food and health policy analysis using home-scan data
Provider: Faculty of Science

Activity no.: 5256-23-03-01 
Enrollment deadline: 13/11/2023
Date and time04.12.2023, at: 09:00 - 08.12.2023, at: 16:00
Regular seats40
Course fee135.00 €
LecturersSinne Smed
ECTS credits5.00
Contact personCharlotte Bukdahl Jacobsen    E-mail address: cja@ifro.ku.dk
Enrolment Handling/Course OrganiserSinne Smed    E-mail address: ss@ifro.ku.dk
Teaching languageEnglish
Exam formCourse participation ; Writting assignment
Grading scaleCompleted/ Not completed
Course workload
Course workload categoryHours
Reading60.00
Course35.00
Report writing45.00
Sum140.00
Aim and content
With the availability of scanner data collected from household panels, researchers are able to analyse research questions that rely on having repeated records of detailed purchases on barcode level for a large sample of households over time. Combined with socio-demographic information about the household this gives the possibility to perform detailed analysis of the heterogeneity in consumer behaviour in relation to relevant societal issues. Because these data have existed for over twenty years and often consists of long panels, it is possible to conduct retrospective analysis of the impact of past food and health policies on consumers, as well as predicting the effect of future policies.
This type of data has some clear advantages as they include very detailed product information for packaged goods, allowing for mergers with nutrition data from various sources and environmental indicators. The data often includes substantial information about participating households. The possibility to issue additional questionnaires to the participants allows for combination of stated and revealed preference data and to analyses the effects of attitudes on behaviour. The long panel structure also facilitates examinations of long-run effects and consumption dynamics. However, this type of data also has some limitations. This includes for example the problems associated with handling very large datasets, as the richness and granularity of the data can also work against the researcher if the data are incorrectly prepared for analysis. It could be simple coding error or trickier aggregation decisions. The difference between consumption and purchases (e.g., stockpiling) can also lead to inadequate policy recommendations especially if the analysis is conducted at daily or weekly frequencies to inform the long-run effect of a potential policy. Aggregation over time can also lead to spurious evidence of habit formation and endogeneity can also be an issue if the potential sources are not thought over carefully. Finally, representativeness can be an issue if the focus is on low-income or ethnic minority populations.

This implies that there is a need for a clear understanding of the nature of the data and how this affects the interpretation of results to fully exploit their potentials. The availability of home-scan data therefore implies the use of specific research methods and techniques to handle extremely large datasets and to ensure representativeness as well as internal and external validity.

Some of the methodological issues that must be addressed when preparing and using home-scan data for analysis include;
1) How to combined purchase data with other datasets such as geographical indicators, nutrient content, health claims or environmental indicators to construct relevant variables for analysis,
2) How to adjust reported quantities to common units appropriate for the analysis at hand,
3) How to adjust quantities for underreporting due to either issues with coverage or failures to report shopping occurrences,
4) How to correctly aggregate the data to reduce the number of parameters, the prevalence of zero purchases or level out discrepancy between purchase and consumption,
5) How to address issues such as attrition or lack of external validity of the panel members
6) Which methods could be used to account for zero purchases?
7) What sources of endogeneity may exist and how can they be addressed?

The aim of the course is to work the course participants through the main issues related to preparing home-scan data or other larger datasets for analysis and address the methodologies applied to perform food and health policy analysis using home-scan data. The central themes of the course are the theoretical foundation of and the methodologies and techniques applied within four areas; 1) Preparing data in appropriate ways for analysis. 2) Handling problems related to self-selection and attrition in relation to self-reported data 3) Handling econometric problems in relation to analysis food policy and health policy issues using micro data, and 4) Validation and interpretation of results from analysis using home-scan data.

Course form and activities:
The first part of the course consists of reading the relevant literature recommended in the reading list. The first 3 days of the course will consist of theory lectures with associated examples on how to handle home-scan data. The last 2 days of the course will consists of selected methodological approaches to do micro-econometric analysis using consumer data. The main lecturers will teach the classes, but will probably invite guest speakers that will give shorter (30 – 45 min) presentations of how to use home-scan data in practice. The applications will also address the issues discussed through the theory lectures in the first 3 days of the course. In the last part of the course the participants work on a short report/paper (suggestion for a research paper) that addresses some of the issues that has been dealt with in the course.

Learning outcome
After having completed the course, the course participants will know the most common methods for using home-scan data for food policy research. The participants will know the limitations and advantages of the data and will know the typical pitfalls and how to handle them. They will also be aware of the vast amount of possibilities related to using large-scale consumer data. Furthermore, they will be familiar with how to prepare the data for specific analytical purposes. The course participant should be well equipped to start doing own research using large consumer panels after having followed the course.

Teaching and learning methods
The theory will be taught mostly through (interactive) lectures, active learning, dialogue teaching, self-study and exercises. The practice will be taught mostly through self-study and the exam (i.e., writing a report/paper) based on the suggested literature and the lectures.

Exam form and criteria for assessment
The exam consists of a brief report/paper/research idea (maximum ten pages double-spaced) which each participant has to write and send to the course organizer no later than two months after the end of the course. The participants need to find a suitable research question and if needed data for their analyses themselves. The research question needs to be within the social sciences and needs to address a food policy issue using home-scan data or other types of consumption data. It is encouraged that the research question is part of the participants’ PhD projects. In their reports/paper, the participants need to state their research question. The reports/papers should follow the structure for scientific papers: Abstract, Introduction, Methodology, Data, Results and discussion, Conclusion. A participant passes the exam if the report/paper indicates that the student has obtained the intended learning outcome (see section “learning outcome”). The course is assessed as "pass" or "fail" based on the quality of the work handed in.

Preparation and self-study
It is expected that the participants prepare for the course by reading the course material (mostly journal articles and selected book chapters) that will be sent to the students approximately 6 weeks before the course starts. The participants are expected to participate actively during lectures and it is also expected, that the participants recapitulate the contents of the course by reviewing the lecture notes and other course material in order to appropriately conduct the analyses and write their report/paper for the exam.

Tentative Program:
Monday 4th (9 am – 4 pm):
Advantages and possibilities
Introduction to the course (Chen Zhen (CZ) and Sinne Smed (SS))
Lecture 1: What is scanner data and why is it useful for food policy and health research? (CZ)
Lecture 2: Enriching the dataset with price, label and nutrition data and environmental indicators (CZ)
Lecture 3: Adding socio-demographic information from questionnaires or other data sources (SS)
Guest lecturer by GfK Consumerscan Scandinavia (yet to be confirmed)
Social evening event – Course dinner

Tuesday 5th (10 am – 5 pm):
Limitations and challenges
Lecture 4: External validity of home-scan data as compared to other data sources (CZ)
Lecture 5: Attrition and self-selection issues related to self-reported panel data (SS)
Guest lecturer by Statistics Denmark (yet to be confirmed)
Social evening event: City walk and coffee

Wednesday 6th (9 am – 4 pm):
Preparing data for estimation
Lecture 6: Methods for adjusting for under-reporting and misreporting (CZ)
Lecture 7: Methods for handling true and false zeros; missing data and infrequency of reporting (SS)
Lecture 8: Aggregation over time and over individuals; methods and pros and cons (CZ)
Discussions about research project with lecturing experts

Thursday 7th (9 am – 4 pm):
Applications and methods: Estimating food demand systems
Lecture 9: Estimating a truncated Exact Affine Stone Index Demand system with endogenous prices (CZ)
Lecture 10: Estimating a multi-stage censored LA- food demand system (SS)

Friday 8th (9 am – 4 pm):
Applications and methods: Evaluation of food and health policies and segmentation
Lecture 10: Evaluating food policies events using home-scan data (CZ and SS)
Lecture 11: Consumer segmentations based on questionnaire data, comparing revealed and stated preferences. (SS)
Guest lecturer using machine learning methods to segment consumers (yet to be confirmed)
Social evening event: Snacks and drinks

Lecturers
Professor Chen Zhen, University of Georgia
Associate Professor Sinne Smed, Department of Food and Resource Economics, University of Copenhagen

Workload

- Reading: 60 hours
- Course: 35 hours
- Report: 45 hours
- Total: 140 hours

Prerequisites
In order to get full benefit of the course the participants should have knowledge of micro-econometrics at intermediate level as well as econometric knowledge at Msc- level. It is an advantage to have some experience with data work and coding in either R, Stata or SAS for statistical analysis.

Remarks

Registration, course fee and accommodation
For participating in the PhD course you must register and pay a fee on 135 EUR/1000 DKK. The course fee covers water/coffee/tea during the course, a course dinner and some refreshments at social events. The course fee must be paid at registration.

Travel, accommodation and other must be organized and paid by yourself. We have pre-booked some single rooms at CABINN Scandinavia (522 DKK/ 70 Euro per night). The hotel is within walking distance to the conference venue. Book directly here:CABINN and get a 10% discount and use the code: BLKKBHINS

Search
Click the search button to search Courses.


Course calendar
See which courses you can attend and when
JanFebMarApr
MayJunJulAug
SepOctNovDec



Publication of new courses
All planned PhD courses at the PhD School are visible in the course catalogue. Courses are published regularly.