I am unclear as to why people would push Apriori into a binary mode. Perhaps my understanding of Apriori is deficient. My experience is using it to calculate the probability of Y given X. By definition, if it is a measure of probability, it is a real number, not a binary or integer. Given that you are starting with a five-point Likert scale the only way I see to use binary values an Apriori is to attempt to determine something else based on the Likert value. In other words, if Strongly Disagree = 1 then the probability of X = Y%. In that regard each category on the Likert scale is binary, regardless of the scale because the values are all mutually exclusive.

Or, to put it another way, you cannot have a Strongly Agree, and a Strongly Disagree for the same respondent on the same question, thus four of the five possible answers will always be 0, and only one will be 1.

I don't know if that helps. Have fun.

Thanks for your response Thomas Sheppard , but my worry is the kind of rule to generate at the end of the day(as in association rule for decision making). I don't still know how these can be fitting into the algorithm even though only one of the scale points is 1 at a time.