"In a non-sensitive information it is usually taken 5-10%. You may also increase upto 20% if you think the collected information is very sensitive and people will not like to answer."??? Any evidence?
Drop outs in a longitudinal study mean that you will have right censored data, so depending on what exact question you are investigating, it may be useful to apply methods used for survival analysis (Kaplan-Meier, Cox regression etc.) rather than discarding censored data points.
Thanks very much for your answer ., I will accept your feedback with reservation. Because your proposal can not be applied for studies where the out come need long time to be observed or in case of rare diseases ... you are welcome