Sample size

I have a traffic data collected for every 15 minutes in 24 hours (96 records). I would like to cut the data from from 9 pm to 5 am (33 records) for calculating the measurements on Free Flow Traffic Conditions. How could I show that this 32 records (32 out of 96) are statistically good sample size for free flow condition instead of closely checking all records with free flow condition including those between 5:15am to 8:45pm in 95% confidence level. to make it easier I want to through away 5:15am to 8:45pm records for free flow condition measurement.