# ordinal independent variables in ANOVA

#### CullingJ

##### New Member
We have some data from a 2x5 factorial experiment which, when graphed,
looks like an obvious interaction.
The ANOVA, however, says the interaction is non-significant. In fact it even
returns an F ratio just under 1 - not even a sniff of an interaction.

One difference between the way the human eye looks at this data and the way and ANOVA does is that, as I understand it, the ANOVA treats each factor as purely categorical. So, it is, in a sense, blind to the monotonic trend I see when looking across the 5 levels of factor B, which is much stronger for level 1 of factor A than for level 2. Is there, perhaps, any statistical technique that captures the ordinal relationships between levels of an independent variable?

Any other suggestions?

John.