Absolute Counts 133 216 102 Percentage 29.49 47.89 22.62 Source: Dierks et al., (2005, p. 136).
As depicted in table 2, the first population cluster shows significant trust being expressed towards food safety information provided by alternative and independent sources. Strong distrust, however, is expressed towards food chain actors, and milder distrust towards media and vested interests. This implies that the first population cluster mainly comprehends alternative trusters with little confidence in classic institutions such as industry and media. The second cluster suggests that the respondents assigned to this cluster appear to be directly opposed to the first population cluster since consumers display trust in nearly all sources of information. Since distrust is only expressed towards information provided by independent sources, this cluster appears to comprise consumers characterised as conservative trusters. The third cluster is characterized by trust being expressed towards information provided by media and independent sources whilst strong distrust, in turn, is expressed towards information provided by alternative sources, vested interests, and, even though to a negligible extent, towards information provided by food chain actors. The inconsistency of this pattern allows for characterising it as predominantly comprising sceptic trusters.
3.3 The SPARTA II Model
Following the classification of German respondents into three different population clusters, emphasis is placed on estimating the determinants of consumer behaviour in both a standard situation and after an external shock. The estimation of the SPARTA II model as outlined in figure 3 for both a standard situation and a hypothesised salmonella infestation aims at precisely identifying changes in consumer behaviour directly attributed to a the occurrence of a (hypothetical) food safety incidence.
3.3.1 Consumer Behaviour in a Standard Situation
As illustrated in figure 3, the consumers’ intention to conduct a particular behaviour, I, is determined through the subjective norm, SN, perceived behavioural control, PBC, behavioural attitude AB, and perceived risk, PR. Trust, T, in contrast, is assumed to have an indirect impact on consumer behaviour. The respective estimates for a standard purchasing situation, based on 377 German observations of which 31.8% correspond to alternative, 46.4% to conservative, and 21.8% to sceptic trusters, are depicted in table 4.
Table 4: Determinants of the SPARTA II Model for a standard situation Variable SN Constant Subjective Norm Alternative Trusters -1.2942（0.7499） 0.0691（0.0695） 0.1588（0.0951） 0.3989（0.1061） 7
Cluster Conservative Trusters -0.6704（0.6998） 0.1587（0.0577） 0.1388（0.0802） 0.3814（0.0942） Sceptic Trusters -1.0010（0.8474） 0.0943（0.0866） 0.2281（0.1127） 0.2723（0.1306） PBC Perceived Behavioural Control AB Behavioural Attitude
PR Perceived Risk 0.1057（0.0786） -0.0424（0.0585） -0.0043（0.1049） Standard errors are put in parenthesis. Perceived risk, PR, is expressed as a weighed average of the respondents’ perception of risk factors. The weighs correspond to the level of knowledge of the respective risk factors.
As illustrated above, the German consumers’ intention to purchase chicken in a standard situation is predominantly determined through behavioural attitude, AB. The differences regarding the impact of behavioural attitude, AB, across the clusters indicate that respondents characterised as alternative and conservative trusters are influenced in a clearly stronger manner than respondents characterised as sceptic trusters. Interestingly, the opposite applies to perceived behavioural control, PBC, which has a stronger impact on sceptic trusters than it has on alternative trusters or conservative trusters. Normative beliefs, i.e. subjective norm, SN, also have a positive impact on all population clusters. Perceived risk, PR surprisingly has a positive impact on the intention to purchase chicken of respondents characterised as alternative trusters. Its impact on conservative trusters and sceptic trusters, however, is slightly negative – even though mainly negligible. Trust is effective in this model only via perceived risk – and considering that the latter has no significant impact on intention – it needs to be concluded that trust does not affect the consumers’ intention in a standard situation.
3.3.2 Consumer Behaviour in the Environment of a Food Safety Incidence
Above estimates abruptly change once respondents are confronted with a hypo thetical salmonella outbreak as particularly emphasised through the increasingly negative impact of perceived risk, PR. The respective estimates are illustrated in table 5.
Table 5: Determinants o f the SPARTA II Model after a salmonella outbreak Variable SN PBC Constant Subjective Norm Perceived Behavioural Control AB PR Behavioural Attitude Perceived Risk 0.2698（0.0910） -0.2558（0.0775） 0.3941（0.0914） 0.0029（0.0568） 0.2617（0.1116） -0.1503（0.1009） Alternative Trusters -0.3650（0.7405） -0.0162（0.0689） 0.0009（0.0883） Cluster Conservative Trusters -2.7934（0.7024） 0.0708（0.0556） 0.2377（0.0709） Sceptic Trusters -1.411（0.8750） 0.0118（0.0875） 0.1395（0.1086） Standard errors are put in parenthesis. Perceived risk, PR, is expressed as a weighed average of the respondents’ perception of risk factors. The weighs correspond to the level of knowledge of the respective risk factors.
In contrast to table 4, the above estimates are based upon 424 German observations. Of these, 33.0% correspond to alternative, 43.9% to conservative, and 23.1% to sceptic trusters. As in the standard purchasing situation, behavioural attitude, AB, remains the decisive factor determining the consumers’ intention to purchase chicken, I, in the environment of a hypothetical salmonella outbreak. Again, this also holds for all population clusters. Yet, whilst the impact of behavioural
8
attitude on conservative and sceptic trusters remains nearly unchanged, the influence on alternative trusters deteriorates. Interestingly, this also applies to the impact of subjective norm, SN, on alternative, conservative, and sceptic trusters alike. Except for its influence on conservative trusters which nearly doubles, this furthermore applies to the impact of perceived behavioural control, PBC, on sceptic and alternative trusters. With exception of its negligible influence on conservative trusters, the impact of perceived risk, PR, increases. Following a food safety incidence, perceived risk significantly affects the consumers’ intention to purchase chicken in a negative manner, most notably regarding sceptic and alternative trusters. Generally, the alternative trusters’ intention to purchase chicken, I, appears to be particularly influenced through changes in the impact of perceived risk, PR, attributed to the transition from a standard purchasing situation to the environment of a hypothetical food safety incidence, whilst other population clusters seem to react in a less distinctive manner.
4 Findings and conclusions
The ever increasing number of food safety incidences in recent years has accentuated the need for an improved understanding of the motives behind consumers’ reaction to random external shocks. As literature research suggests, incorporating the element of trust can be interpreted as a plausible strategy to reduce consumers’ uncertainty in the context of decision making, most notably involving the purchase of goods possessing mainly credence qualities. For the purpose of ascertaining the impact of trust on consumer behaviour under uncertainty and discussing the conditions under which trust might be regarded as a market determinant, emphasis is placed on its conceptual and statistical evaluation under divergent scenarios.
Results indicate that in standard situations, trust has a marginal impact on the consumer’s intention to purchase. Generally, attitude appears to be the most relevant determinant. This, however, significantly changed when respondents are confronted with a hypothesized salmonella incidence. Abruptly, trust turns out to be among the most decisive factors influencing the purchasing decision. As results demonstrate, consumers’ reactions appear to be non-linear in situations characterised through random external shocks.
Moreover, findings convincingly indicate that within a static approach, trust as a determinant of consumer behaviour under uncertainty can be adequately introduced into economic analyses by means of an enhancement of Ajzen’s (1991) Theory of Planned Behaviour. In adjacent steps, two-level dynamic approaches as originally proposed by B?cker and Hanf (2000) should be further considered. This, however, remains subject to further research.
9