Question: A straight line will never fit your data, and using a wiggly line will result in the rejection by referees of the publication of work. Do you get this?

Andrew McKinley answered on 14 Nov 2014:

Ooh – a really good question on data analysis! As spectroscopists, our data sets are everything! (the word ‘data’ is a plural of ‘datum’ – you’ll often see “these data” which is correct!). Ok, when we plot a graph, we almost never ‘join the dots’. Joining the dots means that the data are continuous – if one point increases, by joining the dots we are basically telling the reader that the others either side increase – that they are all linked. Imagine a piece of string strung loosely between two points – lifting one part of the string causes the bits either side to lift as well.
When we make single measurements, resulting in a single data point, we cannot guarantee that there was not a mistake, or “uncertainty” in our measurement. Therefore we can’t ‘join the dots’. What we do instead then is to plot the points, draw a “trend-line” or “line of best fit” and invite the reader to agree with us (or not!) as to whether the data presented can realistically give the trend and if the trend is feasible.

Wiggly lines are the result of connecting data that go up and down a lot – what we would normally call ‘noise’. (exactly the same phenomenon as the ‘hiss’ you hear through your speakers if you turn the volume up when the music is paused). Very few data trends are ‘wiggly’ – in fact, a spectrum of a molecule is about the only place you can get away with ‘wiggliness’ as spectra are showing peaks. These aren’t graphs though – they are spectra.