Penn Computing

Penn Computing

Computing Menu Computing A-Z
Computing Home Information Systems & Computing Penn

Current Load Status
Regular Availability
FAQs & Tips
Password Changer
Support services
About the Data Warehouse
Data Administration
Express Mail
General Ledger
Learning Management
Position Inventory
Research-PennERA Proposals
Salary Management
Travel Expense Management
Tuition Distribution

Verifying Your Query Results

Although query tools make accessing data easier, remember that University data is complex. You can easily compose a "bad query." That is, you may have selected incompatible data elements, the wrong data elements, or structured the conditions in your query such that it returns improper results. Be aware that it is possible to get an incorrect answer or misinterpret the data. For example, when mailing information to students, it is more preferable to select "current address" rather than "temporary address" as the address type. This is because the system looks at the expiration dates for the temporary and local addresses and computes which address is the most current.

To minimize the risk of misinterpreting data or getting the wrong answer:

Compare your results with known reports that are up-to-date and accurate.

Perform "sanity checking." Ask yourself if the information you have is a reasonable answer. For example, few classes at Penn enroll more than 100 students per class so a query returning a class enrollment of 250 students is questionable; grants are usually less than Penn's overall budget thus a grant amount that is listed as twice its Penn budget is unlikely; and Penn's student population is approximately 50% men and 50% women so a query returning a class enrollment of 98% males is probably incorrect.

Consider estimating likely parameters for a reasonable response before executing the query.

Try sequential queries that break large sorts down to components, then add them to be sure the whole is accounted for. For example, if searching for Wharton undergraduate Asian female students, look at: all Asian students, male Asian students, etc. to check the reliability of your query.

Use one or two conditions (query filters) and zero in on what you want. When developing your query to answer a question, start by retrieving detail information, with just one or two record selection conditions. Zero in on what you want, looking at how the results change as you add each condition. If your goal is a summarized report, see how the detail report compares to the summarized one. Make sure the query includes what you want, all of what you want, and nothing but what you want.

Ask someone who knows the data to check your results. Ask someone in your department or school, or contact Data Administration. Also, ask the person requesting the report what results they expect to get. Or ask if the person can suggest another existing report that you can use to check to see if your results are at least in the ballpark. If your query gets results that are nowhere near what the requester expected, your query may be in error.

Continue to become familiar with and knowledgeable about University data by participating in the data collection listservs.

Just look at the report. Sometimes, just looking at the query results can help you spot glaring errors. (Why am I getting so many rows returned? Whoops--I forgot to screen on accounting period!)

Run a similiar query. Try running another query that approaches the question differently, and see if both queries get the same results. For example, if you want figures on telephone charges, in one query, you could set a condition on a list of object codes, and in another query, you could set a condition on the parent object code. If the queries do not get the same results, find out why not.

Check other reliable sources. Check other sources of information, such as BEN Financials or other source system screens or reports, to see how they compare with your query results.


Information Systems and Computing
University of Pennsylvania
Comments & Questions

Penn Computing University of Pennsylvania
Information Systems and Computing, University of Pennsylvania