Text
A Learning-based model for imputing missing levels in partial conjoint profiles
In this research, the authors relax the assumption of the "null effect" for attributes not shown, and they describe an imputation model for partial profiles. This model uses the previously shown values of the given missing attribute, the previously shown values of other attributes, the shown attributes of the current profile, and priors with which people arrive at the experiment (ie, the complete set of information) to impute a missing attribute level. The authors integrate all the information sources using a decay-weighted pattern-matching approach in which attribute patterns that have been previously seen in the experiment affect the values that respondents impute when an attribute is missing, and more recent occurrences weigh more than more distant match occurrences. The authors demonstrate the efficacy of their approach and the existence of attribute imputation in two experiments. The data also reject all simpler extant models as a plausible imputation process. The results imply that respondents construct rather than retrieve product utilities when they evaluate products, which is akin to context effects in survey research. Finally, the authors show that it is possible to influence the way customers impute missing attribute levels by manipulating their priors of coincidence of occurrence of attribute levels..Printed
Ketersediaan
| Call Number | Location | Available |
|---|---|---|
| PSB lt.dasar - Pascasarjana | 1 |
| Penerbit | : American Marketing Association |
|---|---|
| Edisi | - |
| Subjek | In this research the authors relax the assumption of the \"null eff and they describe an imputation model for partial the previously shown values of other attributes the shown attributes of the current profile and priors with which people arrive at the experim the complete set of information) to impute a missi weighted pattern matching approach in which attribute patterns that and more recent occurrences weigh more than more d which is akin to context effects in survey researc the authors show that it is possible to influence |
| ISBN/ISSN | 222437 |
| Klasifikasi | - |
| Deskripsi Fisik | - |
| Info Detail Spesifik | - |
| Other Version/Related | Tidak tersedia versi lain |
| Lampiran Berkas | Tidak Ada Data |