Previously, we have covered the challenges of experiments and the need for reproducibility. Recently, there has been an additional occurrence that merits discussion. There has been a study where roughly 100 peer reviewed articles that were published in respectable journals that had sufficient details on the experiments for the results to be validated by other researchers. In that evaluation, 87 of the published results were unable to be verified by researchers. In some cases, even the original researcher was unable to obtain results that could be considered within the experimental error. If this is a problem, what about all the research that is published which does not include enough information to be able to independently validate the initial results. How can these efforts be considered valid? That is not a question that will be answered in this blog.
In previous blogs, I have mentioned the use of equipment that might have the same model number as a published experiment. The fact that the model number indicates the equipment is similar, it may not be identical. In the particular instance that was observed. Two researchers locate hundreds of miles apart ran experiments of a nanomaterial. Each had samples of the material from the same lot produced by a reputable manufacturer. They ran a predetermined process, used similar test procedures and recently calibrated equipment. But, they obtained different results. They exchanged material samples and each duplicate their original results on the new, to them, material, which differed from their colleague’s results. The question was what was happening. It took a lot of work, but they finally found the source of the problem. In the course of a routine maintenance along with the required calibration, some of the worn parts were replaced by factory “originals”. One of the “original” parts had gone through a redesign to make a stirring process more efficient. It worked and was more efficient, but that meant the duration of the process created a different distribution of particle size than the equipment with the original unchanged part. So, checking just model numbers and type may not be sufficient.
Another example involved a supplier and a researcher. There was certification of the properties of material by the supplier, which was retested by the researchers. The material was shipped via standard commercial transport. The research tested the material and found different results. A reshipment of another batch of the certified material was shipped, and the researcher found a difference from the supplier’s certification. After a number of conversations and equipment checks, it was determined there was an external cause. The determination revealed that the material, which had been certified, acquired enough oxygen on the surface of the material to change the characteristics of the nanomaterials. This is an important point to remember with nanomaterials.
While this seemed to cover the possible causes of deflection of results in experiments, a new one was found. While not involving normal experimental equipment, this one involves a standard, commercial over the range microwave that contains two small lights in its bottom surface. One bulb failed. The solution is simple, look at the manual, which identifies the bulb by part number, and order one. Well, simpler said than done. The bulb, in this case a 40W bulb, is no longer available. There is a manufacturer approved 50W specified as a replacement. Fine, order the bulb and installed when it was received. The illumination was considerably better than the original bulb next to it. Over the weeks, noticed that the bottom of the microwave was warm to the touch. When the second bulb failed, searched for a lower wattage bulb. There was a 35W bulb, but it was not available. Decision was made to try to find information on the existing bulb. It turns out there was some very small marking on the second bulb, which itself is not large. It was possible to make out a 20W marking. Two 20W bulbs were ordered and installed upon receipt. The resulting illumination was of the same level as the original bulbs. The instructions and part identification in the manufacturer’s manual were wrong! So someone following the instructions per the manufacturer’s manual would be replacing incorrect parts.
The point of these examples is that in performing research, no assumption should be made without verification of the equipment, materials, and processes being employed. Even manufacturer supplied directions/instructions may be in error.