Evaluating interobserver reliability of interval data.

Hopkins et al. (1977) wrote a math paper for behavior analysts. They built formulas and graphs that show what interobserver agreement should look like if two observers were just guessing.

The team focused on interval recording. They wanted a quick way to know if your 80% IOA is real or just luck.

The paper gives ready-made tables. You plug in your interval length and behavior rate. The table tells you the exact IOA you would get by chance alone.

If your real number beats the table number, your data is probably solid. If not, you need more training or better definitions.

Kangas et al. (2011) turned the 1977 tables into a free Excel file. You now click instead of doing longhand division.

Friedling et al. (1979) added a visual twist. They plotted the chance band as a gray zone on a graph so you can see at a glance if your IOA clears the bar.

Romani et al. (2018) took the idea onto a busy kids' unit. They showed that clickers and simple sheets can push real-world IOA above the chance line for children with ID/DD.

Stop saying “IOA must be 80%.” Check the chance table first. A 70% IOA can be excellent if the behavior is rare; a 90% can be lousy if the behavior is common. Use the formula, the Excel tool, or the graph—whatever is fastest for you—and only trust data that beats chance.