Saleh, if "sufficiency" means the questions are understandable and written in grammatically correct fashion, personally I do not know of a required size, perhaps you can consider cost as a factor. If it means validity and reliability, statistical sampling computation for the actual survey will be necessary. Available formulae are for estimating population mean (population known, unknown) and for estimating population proportion (population known, unknown). Available selection designs are simple random, proportionate stratified random. disproportionate stratified random, systematic (serial) random, cluster random and multi-stage random sampling.