As researchers soon discover, the inclusion of noisy (irrelevant) variables in cluster analyses can obscure or distort Atrue@ subgroup structures. This problem, identified and discussed by Milligan (1980), has prompted the search for methods that identify noisy variables and either down-weight or remove them. Several researchers have investigated this problem and have met with limited success (DeSarbo, Carroll, Clark, and Green 1984, De Soete 1986, 1988). Recently, Donoghue (1995) and Carmone, Kara, and Maxwell (forthcoming, 1999) have proposed screening methods to identify and eliminate noisy variables.

Horse IL-7 (HIL-7) cDNA was isolated from adult lymph node tissue by reverse transcription polymerase chain reaction (RT-PCR) using oligonucleotide primers based on horse genomic sequences (The Broad Institute). In addition, to the full-length (FL) 531 bp reading frame encoding 176 amino acids, shorter open-reading frames of 477, 396 and 264 bp were also amplified. Nucleotide sequence analysis of these RT-PCR products demonstrated they were homologous except the shorter species were missing internal sequences consistent with multiple RNA splicing events. Consequently, the shorter open-reading frames were re-named splice variant (SV) 1 (477 bp), 2 (396 bp) and 3 (264 bp). Organization of the horse IL-7 is predicted to be similar to that in humans with exon 5 deleted from SV1, exons 3, 5 deleted from SV2 and exons 3, 4, and 5 missing from SV3. Each of these open-reading frames has the potential to be stably expressed as demonstrated using a polyclonal antiserum against human IL-7 to visualize the protein products produced when the FL HIL-7 and each SV were molecularly cloned into pCI and transfected in brefeldin A treated HEK 293 cells. Furthermore, addition of supernatants to horse PBMC from HEK cells transfected (without brefeldin A treatment) with pCI HIL-7 FL, pCI HIL-7SV1, pCI HIL-7SV2 and pCI IL-7SV3 all induced significant incorporation of 3 H-thymidine in the presence of sub-stimulatory amounts of concanavalin A compared to supernatants from mock-transfected cells. Therefore, all isoforms of horse IL-7 described in this report have the ability to stimulate proliferative responses in ex vivo horse PBMC cultures.

For some time marketers have embraced the idea of market segmentation, i.e., defining homogeneous subsets of the total market. From these market segments one can then select the most profitable; these become the organization's target markets. The assumption is that different marketing mixes can better serve the needs of these homogeneous market segments. The technique most often used for creating data-based, homogeneous market segments is cluster analysis. One problem researchers have discovered with this technique is that the inclusion of irrelevant or noisy variables in a cluster analysis can obscure or distort “true” market structures. This problem has prompted the search for methods that identify noisy variables and either down weight or remove them. Recently, Donoghue (1995) has proposed two univariate screening measures that his research suggests results in more homogeneous, and thus more efficacious, market segments. Our research focuses on the behavior of Donoghue's measures with data from actual marketing research studies. We investigate whether or not the Donoghue procedure leads to a more efficacious market structure. Our results with actual data are less encouraging than those reported by Donoghue, who used simulated data.

This article empirically tests a proposal (due to Hagerty) to use Q-type factor analysis to maximize predictive accuracy in conjoint analysis. Data sets from three different studies are used to compare the accuracy of predictions from optimal weighting with those from individual conjoint parameter estimation. The results do not support the contention that optimal weighting significantly improves cross-validity, as compared to individual conjoint prediction.

The authors consider Greenacre's criticisms of the CGS scaling of two-way correspondence analysis. They suggest that Greenacre's indictment of this scaling approach also implies an indictment of multiple correspondence analysis interpretations—interpretations that have been made by him (and by other contributors to correspondence analysis) prior to the publication of their articles.