Measurement Error

 

Image Electroshocking

A WORD ABOUT… Measurement error as a threat to the validity of the conclusions one may want to draw. This type of error can be attributed to the methodologies we use to collect data and, more specifically, to the inherent limitations of the measuring device and/or the manner in which the device is employed. Pictured here is a team from the EPA electroshocking a major tributary of the Pawcatuck River about fifteen years ago. The team moved, four abreast, through a stream reach about 400 feet long and 20-30 feet wide. The business end of the shocking wand is about 6 feet long and the current ranges about 3-4 feet out from there. The sample collected consisted of quite a number of eels and large crayfish but very few wild brookies; and those trout were small. Shall we conclude that this river supports very few trout and only small ones? That might be valid (the truth) … or not, given the limitations of the methodology employed here. Is it likely that four guys marching around in a stream together can get within 9-10 feet of these wild fish? It might well be that a few small ones dove for the nearest cover and got zapped and counted while the other trout scooted out of range of the equipment. All methods of data collection have limitations which then impose limitations on what can be done with the data collected. The people who do the science know this and take it into consideration when interpreting their results and drawing conclusions. The general consumer of data, gleaned from secondary sources such as the print and electronic media, may not have these things in mind and those media typically do not discuss issues regarding validity. PRIBT encourages the use of primary sources i.e. scientific journals. There you can find not just the raw data but a careful consideration of that data and its interpretation. You may not understand everything that you find in a research report published in scientific journal but consider a paraphrase of what Muhammad Ali once said: “50% of something is more than 50% of nothing.”

1 comment to Measurement Error

Leave a Reply