On 2016 Jun 29, Nicolas Griffon commented:

As implicit scores are ordinal, a model using logits of cumulative probabilities [Agresti A. Categorical Data Analysis. p241] would have been a more proper way to compare implicit scores before vs. after.

On 2016 Jun 29, Nicolas Griffon commented:

As implicit scores are ordinal, a model using logits of cumulative probabilities [Agresti A. Categorical Data Analysis. p241] would have been a more proper way to compare implicit scores before vs. after.