Some open and multiple choice questions allow respondents to give more than one answers. Such questions present a somewhat different tabulation problem because the response percentages need no sum to 100 percent. For example, an open question, such as, what magazines did you read during the past week? will result in multiple answers from some respondents. Such data may be tabulated in four different ways. These different ways are based on which, how many, and what combinations of response categories the respondents chose when answering the questions. The objectives of the study will determine of the four is most appropriate.
What Combinations of Response Categories were more frequently chosen by the members of the sample? The answers to the magazines question could be tabulated to determine what combinations of magazines were read by respondents. Such an analysis would answer the question. How many respondents who read magazine A and also read magazines B, C, D and so on? Such tabulation is often referred to as a duplication analysis. From this kind of tabulation, the overlap or duplication between various magazines is ascertained. The results could help an analysts determine the number of additional answers that were added to an advertiser’s media list.
Example: Researchers wish to do a duplication analysis involving magazines A, B, C and D. To do so, researchers will utilize a ghost questions that asks: What combination of magazines did you read during he past week? There are 16 possible readership combination of these four magazines: (A, B, C, D), (A, B, C), (A, C, D), (B, C, D), (A), (B), (C), (D) and (read no magazines during the past week). These 16 combinations can be coded (1), (2), (3) …… (16). The answer(s) given by each respondents to the question which was actually asked can be used to identify which of the 16 possible coeds will be assigned to the respondent’s answer to the above ghost question. The duplication analysis is then obtained by tabulating all of the respondents’ answers to the ghost question.
Note that, if five magazines were involved instead of only four, a total of 32 readership combinations are possible.
This indicates that, as the number of magazines involved approaches 10 a duplication analysis may become impractical as a result of the larger number of possible readership combination.
Of the total number of responses given by all respondents what percentage was given to a particular response Category? If the researchers were interested in a specific magazine’s share of total readership, such tabulation could be based upon number of different magazines reportedly read during the past week by all respondents. This total readership figure is available from the rightmost columns and it shows that a total of 4,797 magazines were read during the past week. By comparing the number of persons who read a given magazine with the total number of magazines read, it is possible to obtain the given magazine’s share of total readership (i.e. a type of market share figure) The number of persons reading each magazine is available and if these figures are compared with total magazine readership of 4,797, it is possible to develop a able showing ht share of total readership enjoyed by each magazine.
Concluding Comments: The four discussions just above illustrate how it is possible to tabulate multiple response questions in four different ways, each of which organizes the data in a slightly different manner in order to help researchers gain a better understandings of the topic being studied. These discussions also serve to point out that multiple response questions can be reinterpreted as three different kinds of ghost questions and that the original question and the three ghost questions will all have separate and distinct answers, all of which can be coded and tabulated.