To appraise the quality of the translations, a panel of language
Contents
1. ty b NEW BOOKS IN REVIEW To appraise the quality of the translations a panel of language professors and bilingual marketing experts were consulted The panel selectively analyzed translations in their language of expertise In general they agreed that the overall translation quality is good Exceptions were noted for translations that were too literal translit erations over decentered conceptual equivalents non concurrence with gender choice for some nouns incon sistencies with the endings for adjectival nouns and a few misspellings An interesting feature of the glossary is that in some cases the English term is given as the translation be cause it is used frequently in the foreign language For example the term marketing is used in the five other languages along with other choices but the English term has a more extensive meaning than the translations Syn onyms are given when applicable but without clarifi cation of usage This book appears to be unique in terms of the number of languages included and the specialization on market ing research Other marketing glossaries are only bilin gual Lleu 1983 English French Fisher Rossi 1979 Spanish English or specialize only in advertising terms Koschnick 1987 English German The glossary is a solid contribution that should help marketing research practitioners and users around the world understand each other better It should be consid2. ered a useful reference for language translations and a should have addition to international business librar ies HUMBERTO VALENCIA American Graduate School of International Management Thunderbird PATRIYA TANSUHAJ MICHELLE MCCANN Washington State University REFERENCES Fisher Rossi Konrad 1979 Glosario de Mercadeo Glos sary of Marketing Mexico Editorial Limusa Koschnick Wolfgang J 1987 English German Standard Dictionary of Advertising Mass Media and Marketing New York French and European Publications Inc Lleu J 1983 English French Glossary of Marketing Terms The World Almanac and Book of Facts 1989 New York Pharos Books CONJOINT LINMAP New York Bretton Clark 1989 CONJOINT LINMAP is Bretton Clark s latest addi tion to its suite of programs for use in conjoint analysis projects joining CONJOINT DESIGNER Carmone 1986 CONJOINT ANALYZER Green 1987 SIM GRAF and BRIDGER Albaum 1989 It is a menu driven personal computer version of the mainframe pro gram LINMAP IV a nonmetric optimization technique for analysis of conjoint and other types of data developed by Srinivasan and Shocker 1973a b Unlike Bretton Clark s CONJOINT ANALYZER and Sawtooth Soft 117 ware s ACA System Carmone 1987 the other major conjoint programs CONJOINT LINMAP can be used to analyze data from either full profile or tradeoff stud ies The package is shipped on one disk the application
3. Models Journal of Marketing Research 16 August 313 22 Srinivasan V and A Shocker 1973a Linear Programming Techniques for Multidimensional Analysis of Preferences Psychometrika September 337 69 and tiple Attributes in a Composite Criterion Using Pairwise Judgments Psychometrika December 473 93 Wittink Dick R and Philippe Cattin 1981 Alternative Es timation Methods for Conjoint Analysis A Mont Carlo Study Journal of Marketing Research 18 February 101 6 34 RAGTIME 3 0 Narbeth PA Bruning amp Everth Soft warepartner Gmbh 1973b Estimating Weights for Mul 119 ALDUS PAGEMAKER 3 02 Seattle WA Aldus QUARK EXPRESS 2 11 Denver CO Quark READY SET GO Paramus NJ Letraset QUESTIONNAIRE DESIGN WITH DESKTOP PUBLISHING SOFTWARE FOR THE APPLE MACINTOSH PC The four software programs reviewed are representa tive of desktop publishing software packages for the Ap ple Macintosh PC The review compares the software programs in general and also in the particular context of survey questionnaire design with desktop publishing software The reader must be forewarned that even though the ease of use of Macintosh software has been a hallmark for Apple desktop publishing software requires a higher order of diligence to master The reason is that desktop publishing is an unfamiliar discipline for most acade micians Besides mastering a new vocabulary the
4. art worth model features must be in a specific order If they are not a utility program is included to rearrange the levels If an ideal point model is being used addi tional constraints on shape of curve concave convex unknown can be specified Finally for vector models the only constraint possible is whether it is increasing or decreasing After specifying the form of the input data rating or rankings card numbers included etc one calculates the utilities for each individual for each feature level These utilities are stored in a file that can be used for further analysis by CONJOINT LINMAP or other soft ware e g SIMGRAF and BRIDGER Data Cleaning Once the utilities have been calculated one can see how well the predicted values fit the actual data This is done by looking at pairs of stimuli and matching the pre dicted preferences with actual preferences A table of violations is presented that shows the distribution over the group of respondents of the percentage of pairs vi olated This is the goodness of fit measure used in CON JOURNAL OF MARKETING RESEARCH FEBRUARY 1991 JOINT LINMAP and is included for each individual in the utility file for further use as a screening variable One can view individual respondent data and on the basis of a user specified threshold value delete re spondents with poor fit i e at or above threshold vio lations Group Statistics Holdout Data Calculations and Simulatio
5. could have been set as default values for the naive user Perhaps subse quent versions will have this flexibility Finally users of CONJOINT LINMAP will find the manual easy to use and full of helpful hints e g the discussion of price elasticity and segmentation Even the experienced conjoint analysis researcher will find the material helpful In summary researchers who find the nonmetric op timization techniques appealing probably would be will ing to accept increased computing time for a better algorithm CONJOINT LINMAP would certainly be the software of choice GERALD ALBAUM University of Oregon FRANK J CARMONE Drexel University REFERENCES Albaum Gerald 1989 BRIDGER and SIMGRAF Jour nal of Marketing Research 26 November 486 8 Carmone Frank J 1986 CONJOINT DESIGNER Jour nal of Marketing Research 23 August 311 12 1987 ACA System for Adaptive Conjoint Analy sis Journal of Marketing Research 24 August 325 7 Paul E Green and Arun K Jain 1978 The Ro bustness of Conjoint Analysis Some Mont Carlo Results Journal of Marketing Research 15 May 300 5 Green Paul E 1987 CONJOINT ANALYZER Journal of Marketing Research 24 August 327 9 Jain Arun K Franklin Acito Naresh K Malhotra and Vijay Mahajan 1979 A Comparison of Internal Validity of Al ternative Parameter Estimation Methods in Decompositional Multi Attribute Preference
6. le in several foreign languages The newest version 3 0 has many improvements ranging from a cleaned up user interface to expanded color capabilities This version will do color separations mail merge macros and font sizes up to 999 pts as well as having 90 degree rotation capability and improved file importation capabilities Additional printing options are Copyright of Journal of Marketing Research JMR is the property of American Marketing Association and its content may not be copied or emailed to multiple sites or posted to a listserv without the copyright holder s express written permission However users may print download or email articles for individual use
7. ns Once the data are cleaned the user can calculate average utilities for the sample evaluate holdout cards or tables and run simulations These steps are done by simply requesting the appropriate option from the main menu specifying the names of the files defining the products to be simulated and so on These steps are straightforward in this package For simulations up to 30 products can be analyzed at a time by using a first choice or maximum utility model SAMPLE RUNS Included with CONJOINT LINMAP are two sample design files and datafiles one each for full profile and tradeoff table tasks The manual is well written and easy to follow so we were able to run the sample data with no problems We used a Compaq Plus with a 386 chip and math coprocessor and a Toshiba 3100 with only a 286 chip Norton Computing Index of 16 5 and 8 3 re spectively to run the sample with full profile data The computing time per respondent was 45 minutes for the Compaq and 12 minutes for the Toshiba With a large number of respondents and only a 286 based machine one could run the analysis overnight The tradeoff sam ple data were run on a UNISYS with a 386 chip but no coprocessor and a Zenith Supersport with an 8088 chip and a math coprocessor The compute time per respon dent was 50 minutes for the UNISYS and 5 minutes for the Zenith We conclude from the sample runs that it is best to have a 386 chip or at least a math coprocessor wi
8. program sample datafiles and two utility programs with a detailed user s manual This version of LINMAP in cludes enhancements for simulation validation data cleaning and other features However some of the flex ibility of the original mainframe version is limited nonconjoint applications are not possible and a few op tions that are user specified in LINMAP IV are preset in the PC version Bretton Clark believes this ts an advan tage as the options chosen are near optimal for most applications whereas user specification could lead to in appropriate choice of options we say more on this de sign decision subsequently Installation Installation of CONJOINT LINMAP from the floppy onto a hard disk is straightforward The program works also from the floppy drive alone However the user must insert a programmable hardware Key that is provided with the software into the parallel printer port in order to run the program This key is used to ensure that the program can be run on only one machine at a time The program itself is not copy protected and backup copies can be made The printer is always available for use as it is plugged into the key To some users such a key will be an inconvenience and perhaps even an annoyance but it appears to be a necessary feature if unauthorized copying of software is to be minimized CONJOINT LINMAP needs no special equipment other than the key provided with the program disk The soft wa
9. re runs on IBM and compatible PCs with at least 256K of memory though 640K is highly recommended The program supports an XXX87 math coprocessor but will run without one As we show the math coprocessor is almost a necessity unless the researcher is in no hurry The linear programming technique used in calculating utilities takes a great amount of computation time The program requires DOS 2 0 or higher and either a color or monochrome monitor can be used An unlimited num ber of product features and cards can be handled Estimating Utility Functions The initial menu lets the user select from among the following options New analysis Retrieve previous analysis Run simulations Screen clean data Analyze holdout data Calculate group utility function Exit CONJOINT LINMAP requires a design file and a da tafile as input The datafile contains either full profile or tradeoff table data and the design file contains either card j 118 or table specifications A full profile design file created by CONJOINT DESIGNER can be read without modi fication The program is flexible enough to be used with any full profile design file written as an ASCII file if it follows the structure shown in the Appendix to the user s manual This structure may be necessary if interaction terms are to be estimated and a non CONJOINT DE SIGNER design file is to be used Tradeoff design files also must be in ASCII format and structured a
10. s specified in the manual CONJOINT LINMAP is fairly flexible in how input datafiles can be structured ASCII files must be used and structured as shown in the manual One acceptable ver sion for full profile data is the basic form supported by Version 2 of CONJOINT ANALYZER Assuming a new analysis the program requests the various file names and type of data full profile or trade off tables The user then is asked whether the data are weighted This is a useful option especially for appli cations in forecasting from nonrepresentative samples For each feature the user indicates whether it is quali tative part worth model or quantitative ideal point or vector model For quantitative features values must be entered for each level of the feature Any errors can be corrected after all features have been specified as there is an option to review what has just been entered The next step is to specify the constraints for selected features This is one of the capabilities of CONJOINT LINMAP that differentiates it from CONJOINT ANA LYZER and the ACA System That is one sets con straints prior to calculating individual utilities instead of first calculating the utilities and then forcing monotonici ty as is done in other conjoint software The user can automatically force the individual s utilities to be at least weakly monotone on features that a priori are assumed to be monotone To use this capability the levels for the p
11. th any machine Ideally one should have both to run CON JOINT LINMAP in a reasonable amount of time Af ter we calculated the individual utilities it was very easy to calculate group statistics and run a few simulations Even the naive user should have little difficulty in using CONJOINT LINMAP with his or her own data as long as the input files are in proper format Summary A general question is whether the significantly long time needed to run a nonmetric algorithm with con straints is worth the effort Carmone Green and Jain 1978 Jain et al 1979 and Wittink and Cattin 1981 have found ordinary least squares OLS procedures to be about as good as the nonmetric procedures under many conditions The researcher must ask whether or not his or her data can be analyzed best by using a nonmetric procedure NEW BOOKS IN REVIEW many researchers feel a nonmetric algorithm is more in tuitively satisfying than a simple OLS model Any re searcher who is so persuaded would have no difficulty in using CONJOINT LINMAP Its major weakness is derived from the algorithm not the software to run a sample of reasonable size would require a powerful com puter One other item of concern is the fixing of parameter values by the program We believe it would have been better to leave as much to user choice as technically and practically feasible rather than making decisions for the user The optimal parameter values
12. user must master commands that are not as intuitive as those in many Mac software applications Nonetheless we be lieve the time invested in mastering the software will be well rewarded with time savings in the designing of questionnaires alone Desktop Publishing Before reviewing the software we must define desk top publishing The simplest definition is that desktop publishing is the ability of software to create documents in virtually any format desired by the user The user in essence is able to create a multitude of formats de pending on his or her needs by controlling the text graphics and creation of the visual layout of the infor mation before publishing at a micro level This ability minimizes one s dependence on outside resources who are experts in such matters One thing the software cannot do for the user is supply creativity in undertaking the layout and related activities The software allows for creativity to take place readily but it must be supplied by the user RagTime 3 0 RagTime recommends that the user have two disk drives one of which must be a hard disk The software will operate on most Macintosh configurations provides basic color support is HFS compatible and is not copy protected The software comes packaged in a vinyl con tainer with about a 350 page manual that includes a five lesson tutorial five glossaries and two 800K formated disks containing the program The program is availab
Download Pdf Manuals
Related Search