Carol,

the problem I have is intertwined with the correct level of statistical analysis. From what you have written it is obvious that you regard the individual plant's outcome measure as the input for the statistical analysis. Imagine had had a statistically significant result with this analysis. You could generalise to other plants under exactly the same set of conditions. That is not what you want to have. On the qualitative level of discussion you want to generalise to the action of the independent variable. You want to make sure that your manipulation (and not something else) is responsible for the difference in the dependent variable found.

The way you have designed your experiment any undetected third variable confounded with the experimental variable could as well be responsible for a difference found.

The 'as similar as possible' approach can so easily fail. Drastic examples: Your cat has peed on one plant bed and not on the other (then all plants in one bed suffer from the same interference and therefore the assumption of independence necessary for the analysis you have planned is violated. Or a pregnat worm was in one of the beds and not in the other and so on.

The mistake to overlook such a possibility is one of the most common mistakes even in the published literature.

The best solution is to have a big selection of plant beds and to assign them randomly to the two experimental conditions (much more work). The correct level of analysis for the statistics would be the average weight per plant bed.

To blind the evaluator is one thing. Even more urgent is to blind the caretaker of the plants iof there is any caretaking planned.

I'm sure I don't know all articles about the power of prayer (or any other spiritual influence) on plants, but what I have read (I remember the name Backster) is lacking all precautions against alternative interpretations. The reason for that in my eyes is that many of the researchers with such an agenda are coming from the hard sciences (are biologists, physicists) and lack any methodological knowledge on how to do these experiments correctly. That is understandable for they come from a tradition in which experimenting is easy (I mean the experimental design, and not the apparatus which may be extremely tricky). People coming from the natural sciences into these fields are utterly naive when confronted with the problem of controlling confounding factors.

The parasciences are full of such articles. Thoroughly honest physicists naively extrapolating their experimental knowledge from how to deal with electrons (or whatever other particles) to a much more fuzzy field.

To do what you plan to do correctly is a tremendous amount of work.

Wolfgang