Thank you for the prompt reply. The answers of your questions are as follows:
1) We have used the first five random task to generate the utilities.
2) Yes, we are using HB estimation.
I understand, fewer tasks leads to larger individual fit to the data and potentially larger scale too along with that fit which leads over fitting of the model at the individual level. Please let us know whether there is any way to neutralize the impact without increasing the choice task.
With the rapid growth of online survey and the respondents are becoming impatient to a longer survey it is becoming more and more important to keep the number of task within 5-6.
Again thank you very much for the help.