# Determining statistical significance of data changing over time

I have collected motion data from three people over two days. Each of the subjects collected motion data from themselves for around 5 hours a day, for the two days, so I have about 10 hours of data in total per person. So far I've done two things:

1. I've averaged the two days, per person, together to get one data set per person which represents their motion over time.

2. I've applied a line of best fit to this single data set per person, and then calculated the percentage change from start to finish using this line (see below). From the image I've calculated a 138% change over the averaged trial of one subject. However, I've been asked to determine if this is statistically significant and I'm not sure how to go about this.

Ideally I want to calculate the significance of change per person, as opposed to against each other, but if I could also calculate the significance between them that would be good too. Would I also want to determine the significance between their two day collection?

I have access to an array of statistical tools, MATLAB, Excel, and SPSS. If someone could help me work this out it would be most appreciated.

• You will need to use time series methods. That should be doable in MATLAB, although 'how to... in MATLAB' is off-topic here. Someone who is an expert in TS will be able to explain the ideas for you, & then you may need to search for a MATLAB TS tutorial to implement them. – gung Apr 24 '15 at 19:28