A routine briefing at the Capitol this week by the state’s auditor’s office aroused the curiosity of several lawmakers as they questioned some of the methods employed by the state’s constitutionally required watchdog agency.

The role of the auditor is to conduct periodic, independent, financial performance audits of the state’s agencies, colleges, and universities, with an eye toward reducing costs, and increasing efficiency in government expenditures.  The auditor, in turn, is accountable to the state legislature primarily through the Legislative Audit Committee, comprised of lawmakers from both houses.

Auditing can be approached utilizing statistical sampling, or non-statistical sampling, and in Colorado non-statistical sampling is used most often.  A non-statistical sample is defined by industry standards as any sampling procedure that does not measure risk—such as potential financial losses if the statistics were to play out.

Sally Symanski, the state’s auditor, said non-statistical sampling is used when audits are performed for medicaid spending for two reasons—cost, and federal guidelines.  A statistical sample doubles the number of sample case studies and the federal government only requires the lower threshold of the non-statistical sampling.

Sen. Morgan Carroll, D-Aurora, a member of the audit committee, questioned the wisdom of not calculating risk when the state is footing the bill, also noting that the public may care more about that risk than a statistic.

“It seems that the more likely it is that you’re dealing with financial mismanagement of something, that could add up to real dollars to the state,” said Carroll. “The public may care more about that.”

Committee chairman, Sen. Dave Schultheis, R-Colorado Springs, said that he believes that things might be different if the more rigorous  approach to sampling was employed in audits.  Current methods show financial losses only on randomly sampled cases.

“When we sit down with the heads of departments and they’ve got to justify things– they’re not really motivated when the report said that this (error) turned out to be a loss of $17,000 or $2,300, or whatever,” said Schultheis. “I gotta tell you that if we said, ‘Gee that could translate into $500,000 on the total population,’ that would get their attention fast and they would deal with those issues, and the press would be all over them.”

Schultheis went on to say that the value of the more rigorous statistical audit, which would give a sharper snapshot of errors, is high enough to outweigh any drawbacks.

“In terms of the overall benefit to the state, in these economic conditions, I think it’s extremely valuable,” said Schultheis.

The state auditor’s website offers the following explanation of utilizing smaller samples:

Many state-maintained records, such as human services, correctional, and judicial records, are still manually recorded and filed. Since audit staff cannot review 100 percent of manual files efficiently, a sample is chosen. Samples are selected based on a variety of factors, such as geographic distribution, demographics, volume of expenditures, or rural/urban considerations. When drawing conclusions, auditor judgment, together with the overall sufficiency, competence, and relevance of the evidence, is the basis of all findings even when a limited number of cases is involved.