Risk Benefit Analysis – A key to understanding the decision making process of the Northern Ireland food consumer, with respect to maintaining a healthy.

Risk Benefit Analysis – A key to understanding the decision making process of the Northern Ireland food consumer, with respect to maintaining a healthy balanced diet.

Norma WindrumDepartment of Food Science, Queen’s University Belfast

Introduction

In industrialised countries consumers have come to expect an ever increasing range of foods, largely due to the development of new technologies and the globalisation of the food chain.

They are also more concerned about a growing list of food attributes including quality and freshness, price, taste, healthy eating, convenience and family preferences.

Despite a plentiful food supply, it is recognised that foods not only provide benefits to the consumer, in terms of nutrition, taste and satiation but may also present risks.

Food risks

Acute risks In industrialised countries, an estimated 10-

30% of the population suffer food-borne illness every year.

Most cases resolve themselves within a few days. Occasionally though very serious illness can occur, with young children, elderly people, pregnant women and people already suffering from underlying illness most at risk.

Food risks

Chronic risks:

Concern regarding food consumption has started to focus on the long term health effects of poor nutrition.

Globally, more than 1 billion adults are overweight with at least 300 million classified as obese.

In the UK the level of obesity has more than doubled since 1980, whilst in Northern Ireland currently over half of women and about two thirds of men are either overweight or obese.

Consequences

Health risks Chronic diseases including cardiovascular

disease, diabetes, certain forms of cancer and gallbladder disease.

Psychological consequences Lowered self esteem, anxiety, clinical

depression. Negative attitudes towards overweight

individuals

Consequences

Financial costs

Between 2 and 8% of the total costs of illness in Western countries are attributable to obesity.

UK example The NHS spends £0.5 billion on treatment of

obesity related diseases.

Obesity is responsible for 18 million sick days per year.

The impact on the economy is expected to rise to £3.6 billion annually by 2010.

Risk-benefit trade offs

Consumers are now more interested in an ever increasing range of food attributes, however there may be both risks and benefits associated with each of these.

It’s not always possible to avoid all potential risks as consumers may be interested in a number of attributes but find them to be partly incompatible.

In some situations consumers might choose the option with the highest risk if it had enough compensating benefits.

Risk-benefit trade offs

Consuming fatty foods e.g. chips regularly has been linked to an increased risk of coronary heart disease.

The factors that may balance this risk and invoke purchase could be physiological (hunger, taste), convenience or disbelief in the relationship.

How consumers perceive product attributes is a critical factor in the food choice process.

Decision making process may differ depending on level of risk involved.

Aim of study

To select 3 foods with different levels of nutritional quality and assess if consumer risk-benefit trade offs differ accordingly, when choosing between alternative products.

To develop a predictive food choice model, based on these findings, which can be applied to a range of foods with different perceived levels of nutritional quality.

To determine the effect of demographic and individual characteristics on consumer risk-benefit trade offs.

Aim of today’s presentation

To report the results from a pilot study using conjoint analysis, a statistical method for the analysis of product attribute trade offs, applied to consumer preference of beef burgers.

Conjoint Analysis

This approach takes its theoretical basis from Lancaster’s (1966) theory that products consist of a set of attributes.

It assumes that alternative versions of the same product can be defined as a set of different attribute levels.

Attributes Levels

BrandFlour type Price

National, store brandWhite, partly wholewheat, wholewheat, mixed grain<50p, 50p-99p, 99p-£1.50

E.g. Attributes and levels for a typical bread purchase

Conjoint Analysis

It models how consumers trade off attribute level combinations when making a product choice.

Identifies the attribute combinations which confer the highest utility to the consumer and establishes the relative importance of attributes in terms of their contribution to total utility.

Conjoint Analysis

Main steps in conjoint analysis:

Select product. Determination of relevant attributes. Assigning levels to each of the product

attributes. Determination of method to generate data. Data collection. Analysis of results.

Step 1 – Selecting food products

Questionnaire presented to nutritional experts and food consumers

Asked to rate 20 commonly consumed food items based upon either how risky or beneficial they perceived each of them to be in terms of their contribution to achieving a healthy balanced diet.

Risk-benefit relationship

Steps 2&3 - Establishing attributes and levels

2 focus groups and 10 structured interviews conducted with selected food products to determine relevant product attributes and their associated levels.

The final three foods chosen were those whose characteristics clearly facilitated a risk-benefit analysis.

High risk – beef burgers Medium risk – Sirloin steak Low risk - Porridge

Attributes and levels for beef burger

Attribute Attribute level

Beef Content

Flavour

Cooking time

Saturated fat content

Salt content

77% Beef, 99% Beef

Flavoursome, Bland

8 minutes, 16 minutes, 25 minutes

4g, 8g, 12g

0.3g, 0.9g, 1.5g

Step 4 – Generation of data

The full profile conjoint analysis approach was used.

Study gave rise to 108 possible product scenarios (2 x 2 x 3 x 3 x 3).

Utilised a fractional factorial design, under SPSS conjoint, to reduce the number of profiles to a manageable number, whilst maintaining orthogonality.

Generated 20 product scenarios.

Step 5 – Data collection

Each product scenario was rated by consumers using a 100-point preference scale.

A representative sample of 100 beef consumers completed the questionnaire in July/August 2006.

Contains 99% beefFlavoursome

Takes 16 minutes to cookContains 0.9g salt

Contains 8g saturated fat

PREFERENCE SCORE (1-100)__________

Example - Beef burger product profile

Data analysis

Preference scores for each scenario were determined.

The contribution of each product attribute level is termed its “part worth utility”.

For each respondent, the part worths were estimated using Ordinary Least Square regression analysis.

Conjoint uses the utility ranges to compute importance scores for each attribute.

Conjoint analysis summary (n=100)

Attribute Level Utility Relative importance (%)

Beef Content

Flavour

Cooking time

Saturated fat content

Salt content

99% Beef77% BeefFlavoursomeBland8 mins16 mins25 mins4g8g12g0.3g0.9g1.5g

-20.13-40.26-13.64-27.27-5.35-6.98-4.36-5.14-8.87-11.19-3.26-7.56-8.90

29

20

19

16

16

Notes: *The lowest utility values represent less value from the consumer’s perspective; **The highest utility values represent more value from the consumer’s perspective

Segmentation

The effects of demographic and individual characteristics on the consumer choice process were analysed.

It was found that the decision making process is invariable throughout all segments of the population.

Consequently, it can be concluded that this model can be applied to any population sector.

Validity of method

Pearson’s r and Kendall’s tau association values were used to assess the accuracy of the model.

Value

Pearson’s RKendall’s tau

0.9770.850

Correlations between observed and estimated preferences

Pearson’s r correlation using reported preference scores also tested the validity of the model.

Validity of method

Utility Score

Pre

fere

nce

Sco

re

Regression Analysis

A regression coefficient was calculated in order to predict a Likelihood to Choose Score (LCS) for any product profiles not included in the study.

Variables Coefficient Standard Error of Coefficient

Standardised Coefficient (beta)

t value Sig.

Constant

Overall Utility

9.623

0.965 0.084 0.766 11.442 0.0001

*Dependent Variable: Likelihood to choose score

Predictive model

LCS = 9.623 + 0.965 (Overall Utility)

WhereOverall Utility = Ua + Ub + Uc + Ud + Ue +

constant

Example

The total utility of a beef burger which takes 10 minutes to cook, contains 2g fat, 0.5g salt, 99% beef and is flavoursome is:

(-6.10)+(-5.02)+(-2.75)+(-13.64)+(-20.13)+118.68 =71.04

Therefore the likelihood to choose score can be calculated:

LCS = 9.623 + 0.965 (71.04)LCS = 78

In conclusion

The predictive food choice model, describes how consumers estimate and evaluate the risks and benefits associated with a range of food products in order to make a product choice.

The model will be further tested on a range of foods with different perceived levels of nutritional quality.

Model will be used to support new product development and inform communication and educational strategies for food and nutrition issues in Northern Ireland.

Acknowledgements

Department of Agriculture and Rural Development, Northern Ireland for funding this research.

Supervisors Dr Roy Nelson Dr David McCleery