Compare the number of residential customers (or whatever the utility's "denominator" is) to the number of households (at county level). Use ACS microdata to calculate the latter. Calculate the number of natural gas consuming households using the primary heating fuel variable.
Analyze discrepancies. One potential pattern is that customers is lower than households in counties with more multi-unit residences, since apartment buildings might be counted as a single residential customer or classified as commercial sector instead. At this point, we just want to know if/how/why/where discrepancies exist. We can think about correcting the data later, if necessary.
(3b.) If a utility data series goes back to 2009, you can compare statewide residential consumption to the RECS 2009 total. Good agreement is a "necessary but insufficient" condition for accuracy at smaller geographies.
(3c.) If you find discrepancies, email the maintainers of the utility data to ask for possible explanations along the lines suggested above. We want to understand what the utility data measure, exactly, and where the data are "weak".
Compare the number of residential customers (or whatever the utility's "denominator" is) to the number of households (at county level). Use ACS microdata to calculate the latter. Calculate the number of natural gas consuming households using the primary heating fuel variable.
Analyze discrepancies. One potential pattern is that customers is lower than households in counties with more multi-unit residences, since apartment buildings might be counted as a single residential customer or classified as commercial sector instead. At this point, we just want to know if/how/why/where discrepancies exist. We can think about correcting the data later, if necessary.
(3b.) If a utility data series goes back to 2009, you can compare statewide residential consumption to the RECS 2009 total. Good agreement is a "necessary but insufficient" condition for accuracy at smaller geographies.
(3c.) If you find discrepancies, email the maintainers of the utility data to ask for possible explanations along the lines suggested above. We want to understand what the utility data measure, exactly, and where the data are "weak".