Handling cases with incorrect answers to the manipulation checks

Home Forums Methodspace discussion Handling cases with incorrect answers to the manipulation checks

Viewing 1 post (of 1 total)
  • Author
  • #463
    Amin Nazifi


    I am currently doing an experiment-bases research on termination of customer relationships. I am just wondering if I should exclude any cases from the initial sample based on incorrect answers to the manipulation checks:  For example, I have manipulated the service recovery using three nominal variables: did the firm offer an apology (1.Yes, 2. No), the firm offer any explanation (1. Yes, 2. No) and how much monetary compensation did the firm offer (1. None, 2. $100, 3. $200). So, for those who have answered any of the recovery manipulation checks incorrectly, should they be excluded from the analysis?


    Also, I have manipulated the service termination strategy choice which is a latent variable as follows: “customer oriented” in which the firm offers to help the customers move to another firm vs. “firm-oriented” in which the firm focuses on self-interest and does not provide any help at all. Measured on a scale of 1 (strongly disagree) to 7 (strongly agree), using 4 items, respondents have indicated whether the firm has offered any help to customers at time of service termination. The average mean for customer-oriented approach is 4.07 and the mean for firm-oriented is 2.70 and the difference is statistically significant. But I am just wondering if some of the cases need to be excluded from the analysis; For example, if a respondent has given a score higher than 4 (indicating that they agree that the firm has offered help) in a firm-oriented condition (where no help had been offered), should this case still be kept or excluded for further analysis?


    My impression is that if the firm has not offered any help, but the participant indicates that help has been offered (i.e. a score of above 4), the response is incorrect and the case should be considered as noise and excluded from further analysis. But the same does not seem to be as clear cut and straightforward for the customer-oriented approach. Because whilst in the scenario, it is clearly stated that help is offered to customers to move to a new firm, the respondent may not seem this gesture as helpful enough (compared with the inconvenience caused due to forced service termination) and might still give a low score (i.e.  4 or below). But I can’t seem to find any guidelines on what to do with the results of manipulation checks of latent and observed variables to proceed with data analysis. So, any advice or suggestions would be greatly appreciated.

    Many thanks,


Viewing 1 post (of 1 total)
  • You must be logged in to reply to this topic.