NIPO ODIN Version 5.17
NIPO Conjoint MeasurementNote: NIPO Conjoint Measurement is a way of doing research, that puts the researcher in a position to follow the weighing-process of respondents, when buying products or services. This process of weighing is being 'staged'. Collecting the data in this manner, the importance of certain features, when choosing a product or service, can be determined much more precise in comparison with 'standard' methods of market research. The result of NIPO Conjoint Measurement will be a clear insight a certain feature has for the respondent when purchasing a product or service. NIPO Conjoint Measurement especially is an important tool when developing new products to determine in advance which specific features of the product are more and which are less essential. It also can play a role at recruitment and selection-investigations, to know which applicants, having a certain set of features, have more probability to get the job. Further analysis of NIPO Conjoint Measurement data can be done in the statistical package of your choice. The theory of conjoint analysis assumes that the value of a product for the respondent, the so-called utility, can be split up in several part-utilities, for every feature of that product a separate part-utility. Traditionally a conjoint measurement survey consists of the showing of a batch of cards to the respondent, each of them containing 2 profiles: profiles are 2 different descriptions of a product. The respondent is then to order these cards; given the ordered cards the (relative) importance of each feature - and even of each level within that feature - can be determined. But having products with for example 5 features, each of them having on average 4 levels, this led to enormous practical problems because of the large number of possible combinations. One solution was to show each respondent only a subset of the total number of cards and to do the conjoint analysis at an aggregate level only, rather than at an individual level. The use of personal computers in the fieldwork has enlarged widely the possible applications of NIPO Conjoint Measurement. Now a computer may be used to do present and accumulate during the interview, so it is no longer necessary to go through the whole set of cards. For now, at any moment during the interview when constructing new profiles, the answers given so far by the respondent can be used to consider only those features/levels that are most relevant to this respondent. In this way it is possible to determine relative importance (part utilities) for each level, and for each respondent. In the NIPO Conjoint Measurement within the NIPO Fieldwork System the respondent is first asked to mention unacceptable levels: the levels the respondent would never agree to, for example chewing gum containing sugar. These levels will no longer appear in the interview - one is allowed to accept all levels. The numbers of these unaccepted levels are not stored in the U-file, but their corresponding utilities will have a specific negative value, so they may be recognized afterwards. Then the respondent has to order the remaining levels according to his preference: now the whole list has to be put in an hierarchical order - the first and the last choice will be stored in the answer file. After this the respondent has to mark the importance of the differences between the first and the last choice, in other words: is the order in the levels all the same to the respondent or do these differences matter. This importance factor (on a scale from 1 to 4) is also stored in the U-file. In the so-called main part the profiles are constructed and shown to the respondent: the respondent has to look at two descriptions of a product that have total utility as equal as possible, and decide to what extend one of the two corresponds to his / her preference (moving a cursor-block to the left or to the right on a scale from 1 to 9). At least 2 and at most 5 features appear on the screen together, and this number of screen-features is increased slowly from 2 to the actual number of features (or to 5, this depends on what is reached first). After an answer has been given utilities of all accepted levels will be re-estimated, and the next profile is being constructed based on these utility-values. The number of profile-questions depends on the number of accepted levels. Based on the answers to these questions final utility-values for all levels are estimated; all these values will be corrected after the final part and will be stored in the U-file. (Storing all the profile-information would cost too much room). In the final part 3 questions follow about the probability of purchasing a product: the product with the smallest sum of utility-values of all levels is shown first, and the respondent is asked to give the probability to purchase that product; in the third question the respondent is shown the product with the highest sum of all utility-values - and also the question is to enter the probability to purchase that product. But in between there was the second question with a product with sum of utility-values somewhere between the first and the third concept. These 3 answers should be increasing, provided the respondent is answering consistently. Then two regression-coefficients (a and b) and a correlation-coefficient are estimated - these 3 coefficients are stored in the U-file. The regression-coefficients are used to correct the utility-values. |
|||
|
