Peter Kennedy was an asso­ciate edi­tor of the Inter­na­tional Jour­nal of Fore­cast­ing and a superb applied econo­me­tri­cian. He died unex­pect­edly in August 2010. He was best known for his excel­lent book A Guide to Econo­met­rics as well as his “Ten Com­mand­ments of Applied Econo­met­rics”. He pro­vided a vari­a­tion on his ten com­mand­ments in advice to his stu­dents in the form of the fol­low­ing ten rules:

1. Use com­mon sense (and eco­nomic theory)
2. Avoid Type III errors (pro­vid­ing the right answer to the wrong question)
3. Know the context
4. Inspect the data
5. KISS (Keep It Sen­si­bly Simple)
6. Make sure your results make sense
7. Under­stand the costs and ben­e­fits of data mining
8. Be pre­pared to compromise
9. Do not con­fuse sta­tis­ti­cal sig­nif­i­cance with mean­ing­ful magnitude
10. Always report a sen­si­tiv­ity analysis

Although these were framed in the con­text of applied econo­met­rics, they apply to any data analy­sis (except sub­sti­tute “eco­nomic the­ory” for the rel­e­vant field in the first rule).

The rest of his advice is also well-​​worth reading.

Related Posts:

• Update on a Stack­Ex­change site for sta­tis­ti­cal analysis
• Sta­tis­ti­cal Analy­sis Stack­Ex­change site now available
• Cross­Val­i­dated launched!
• How to avoid annoy­ing a referee
• A Stack­Ex­change site for sta­tis­ti­cal analysis?