The 'method' phenomenon
Recently, I have been collecting and classifying different pedagogical and instructional methods. What I have come across is the "there is no best method" phenomenon and that different methods achieved indeed the same learning outcomes (although it is hard to believe at times and would need a detailed look into the actual studies). What was suggested by the text (Koskenniemi 1968!)1 was that the effectiveness of a method is always determined by the goals and factors of a learning situation: there is no such thing as a global recommendation for one method.
The discussion on the effectiveness of e-learning reminds me of this very phenonmenon. A new technology arises and people start forecasting that it is the solution for our educational needs. This already took place when movies, television and computer-assisted instruction entered the arena: The revolution was always predicted but never took place. Same with e-learning: people predict that it will change the way education is done and expect e-learning to be more effective, more efficient etc. Accordingly, the CHIRON-project is currently conducting a survey to [quote] "analyse individual experiences on evaluation models of e-learning effectiveness". Another example is the Computer Assisted Language Learning Research (CALL), which has been trying to prove for years that computer training/instruction is "better" than traditional instruction. Perhaps they are asking the wrong question to begin with.
In this regard, you may also be interested in the study collection of Thomas Russell, who found no significant difference in student achievement across hundreds of studies comparing "traditional" methods and technology use for teaching.
1 Original publication in finnish under the title: opetuksen teorian perusaineksia.