We have argued that failure to publish SCD

results with small effect sizes is a bias, which

leads to evidence-based practice reviews that

systematically overestimate the effectiveness of

treatments in SCD research. However, at least

some SCD researchers believe that demonstrat

ing a visually compelling functional relation

(and thus a large effect size) is not a bias but

rather is good SCD research practice and

should be an important consideration in publi

cation decisions. These researchers assert that

studies that do not demonstrate a visually large

functional relation are uninterpretable; for

example, a negative result may not mean that

the treatment failed but rather that the

researcher failed to implement the treatment

adequately or failed to measure the outcome

with enough reliability or validity. These state

ments may be true, although it would be better

to base publication decisions on direct evidence

about poor treatment implementation or poor

measurement reliability than on indirect evi

dence of small effect sizes. Even so, a negative

result may sometimes mean the treatment does

not work well. SCD researchers need to better

defifine professional standards for publishing

negative effects and the process for document

ing intervention ineffectiveness. Knowledge of

what does not work should have just as great a

place in evidence-based practice reviews as

knowledge of what does work. Also, studies

with negative results may differ from studies

with positive results in having different kinds of

cases, settings, treatment variations, or out

comes. Omitting results that are negative for

this reason deprives the fifield of knowledge

about what moderates the size of an effect.