[babystrangeloop]

I have been tracking both Gasbuddy and the US DOE EIA.

The US DOE EIA publishes their national gasoline prices here:
http://www.eia.gov/petroleum/gasdiesel/

Gasbuddy has
http://www.gasbuddy.com/gb_retail_price_chart.aspx
You can adjust that chart to show a range of one month which gives you a clear view of individual days.

The major difference is that gasbuddy uses moderated submissions from individual agents that are awarded points for contributing, basically "crowdsourcing" the gasoline prices they consolidate into their average price.
http://en.wikipedia.org/wiki/Crowdsourcing

The US DOE EIA uses some other mechanism for generating prices, probably with fewer people involved since crowdsourcing notorious for attracting crowds.

Historically I see a pattern: The Monday price published by the US DOE EIA effectively predicts what gasbuddy's price arrives at by the next week.

US DOE EIA national average price posted February 27, 2012 is $3.721
Gasbuddy on March 4, 2012 is about $3.71
Gasbuddy on March 5, 2012 is about $3.72

US DOE EIA national average price posted March 5, 2012 is $3.793
So by next Monday the gasbuddy price will be just about that.

How do they do it? Does the US DOE EIA know the price that gasoline will rise to by the next week? Is there a formula or do they have connections? Is this situation manipulated and to what extent?

[dolanbaker]

I would imagine it's a simple fact that if the cost of the stuff that goes into a refinery has increased by 1% and it takes a week to go through the system, then petrol pump prices will be 1% higher the following week!

[pup55]

Is there a formula or do they have connections?

Gasoline Price Data Collection Procedures

Every Monday, retail prices for all three grades of gasoline are collected by telephone from a sample of approximately 900 retail gasoline outlets. The prices are published by 5:00 P.M. Monday, except on government holidays, when the data are released on Tuesday (but still represent Monday's price). The reported price includes all taxes and is the pump price paid by a consumer as of 8:00 A.M. Monday. This price represents the self-serve price except in areas having only full-serve. The price data are used to calculate weighted average price estimates at the city, state, regional and national levels using sales and delivery volume data from other EIA surveys and population estimates from the Bureau of Census.

http://www.eia.gov/petroleum/gasdiesel/ ... cedure.cfm

So the answer is: both a survey and a calculation.

To estimate average prices, sample weights were constructed based on the sampled outlet's number of pumps, a proxy for sales volume. These weights are applied each week to the reported outlet gasoline prices to obtain averages for the specific formulations, grades and geographic areas. Weights used in aggregating grades, formulations and geographic areas were derived using volume data from the EIA “Monthly Report of Prime Suppliers Sales of Petroleum Products Sold for Local Consumption”, and demographic data from the Bureau of the Census and Department of Transportation on population, number of gasoline stations and number of vehicles.

Definition of the Coefficient of Variation

The Coefficient of Variation (CV) of the mean, a measurement of variability of the mean price, is calculated as the standard error divided by the mean. As a measure of precision reflecting the error due to sampling, it can also be used to construct a confidence interval for the population mean. For example, a mean price of $1.67 and a CV of 0.004 (.4%) indicates at the 95% confidence level that the population mean price has an upper limit of $1.68 ($1.67 + 1.96*0.004*$1.67) and a lower limit of $1.66 ($1.67 - 1.96*0.004*$1.67).

and this explains why they're off a penny from time to time....

[babystrangeloop]

and this explains why they're off a penny from time to time....

But it doesn't explain why they are faster than gasbuddy.
Why does it take gasbuddy a week to catch up to the US DOE EIA?
If gasbuddy simply censored high prices they would not catch up.
Is there something about crowdsourcing that makes it slow?