I have analyzed zip code data and block data of us census 2010. But I am unable to find which data is more at a granular level. I want to say that household data is more granular. Where can I find household data if it is more granular?
The granularity of the census data is done by design. When the United States Census bureau conducts a sample of a household, they make sure that the granularity is not small enough as to identify individuals at the household level. If there is not enough of a sample at the block group level, it will get aggregated up to the zip code level.
You can read up on data suppression within the American Community Survey here: http://www.census.gov/programs-surveys/acs/technical-documentation/data-suppression.html
both of these data sets contain one-record-per-person
united states decennial public use microdata sample
american community survey