There has been growing awareness and debate across the research community that standalone stated preference experiments may tend to provide biased willingness to pay estimates as a result of the so called "focus effect". As a result, researchers have come up with alternative ways of mitigating this which include "masking" and "packaging" approaches. In masking, the subject of interest is made part of a broader subject and respondents are slowly made aware of and drawn to the subject before the regular SP experiment(s). The idea here is that a "truer" value is obtained through having the issue under review explored in its broader context. However, this can then lead to another issue which concerns the appropriate evaluation of a range of issues when there are two or more stated preference exercises involved. The concern here is that if price is included in both (or more) exercises then the monetary valuations obtained from each exercise will potentially be derived as if those were the only issues being explored and valued. If the respondent had known, it is argued, that there were more issues to evaluate then s/he might have responded differently and "held back" some monetary resource. A key methodology to help counteract this has become known as the packaging approach whereby an additional SP experiment is carried out to scale across the different stated preference exercises. The approach works by showing packages of improvements from the so called "lower level" exercises where all the attributes are either , for instance, "on" or "off". The estimates of the packs are then used to factor the individual estimates from the regular lower level experiments. There is, however, no clear indication on how this package is to be dealt with. One approach is to use at least one variable as a common variable (generally the cost variable) across all the stated preference experiments (regular and package) and use it as linking variable in the analysis. The disadvantage of this approach is that the exercises are overloaded unnecessarily with repetitions of the common variable. This has significance when dealing with a large number of attributes that are divided into multiple SP experiments. On the other hand, some practitioners feel uneasy if price only gets included at the overall package level. This paper presents the results from experiments designed to test the effectiveness and potential differences of the package experiment being conducted with and without the linking variable.
