I do not agree with your insistence on a mechanism.
In 1964 the U.S. Surgeon General published a report where he concluded that smoking causes cancer. At that time the mechanisms behind this was not understood. However there was statistical evidence in the form of a study with over 1 million subjects that showed that smoking and cancer were highly correlated.
That meant that either
- A. smoking causes cancer
- B. cancer causes smoking
- C. another third factor causes both smoking and cancer
B. was easily be ruled out by only including those who reported to have smoked for some time prior to getting cancer. From what I understand C. was ruled out by adjusting for plausible third factors.
However I agree with your sentiment that much of the research within this field is low quality (underfunded but performed according to scientific methodology).
Looking at the metastudy linked by JustSnilloc I see some problems with many of the studies included:
1. The number of subjects is too low
3 of the studies have n = 7, 9 and 11. I doubt much trust can be placed in the results of a study with so few subjects. There is appearantly a statistical method called MBI that lets you conclude with high certanity from extremely small sample groups. This method has however been disproved and the only field where it is being used is sports science.
2. The timespan of the study is too short
For untrained subjects in particular the "newbie effect" causes hypertrophy and strength increases almost regardless of protocol to begin with.
3. It test the wrong hypothesis
It does not test the hypothesis coming from experience in the bodybuilding and strength
communities: 3-5 repetitions are ideal for strength whereas 8-12 is
ideal for hypertrophy. Instead it typically compares 8 repetitions
with 20 repetitions. There seems to be a general problem with papers being produced by people with little or no practical weight training experience: Rippetoe: The Problem with “Exercise Science”
4. The methodology in the included studies are too different.
I also question the value of pooling many studies with slightly varying experiments in a metastudy. As an example say that study A says that 15 reps are better than 10 reps for hypertrophy in the leg press for young men. However study B says that 5 reps are better than 12 reps for hypertrophy in the bench press for women over 40. What are we to conclude from this? The statistical method used to combine all the studies seems highly complex. It also seems to involve a lot of weights and assumptions. I remember the quote "With four parameters I can fit an elephant". Meta-analysis seems to have been used to generate some strange conclusions. This paper questions the use of meta-analyses to evaluate resistance training: "In conclusion, considering the large number of variables
involved in resistance training and the methodological inconsistencies in the current literature, it seems impossible to make
comparisons of different studies or include different studies in
the same analysis".
To me it seems that within this field a lot of reasearch is produced, but much of it is low quality or not really relevant.
There may be a problem with funding.
There seems to be enough funding for running journals and doing peer reviews etc. but maybe not enough funding to actually do long running experiments with a large number of subjects. Instead many resort to trying to extract more info from existing studies by doing meta-studies of questionable quality.
It may also be that such experiments have been conducted by large olympic teams in the past (USSR in particular) but never published. The result may now be common knowledge among top level coaches and they see no need to rerun the same costly experiments.