Dissertation statistics

I am currently writing my dissertation on long term effects of CEO successions, ie. are there any statistically significant effects on measures such as ROA / ROE etc. I have eg. ROA data from the 7 years surrounding the succession of a new CEO, ie. -3 to +3 years and year 0 in the middle. I also have the same data for the relevant sector to benchmark for externalities. By calculating an average I have been able to tell that eg. ROA is usually lower than average for most companies in the years before they change their CEO. There is significant improvement in the year the new CEO starts, before most of the effect wears off and the performance returns to the mean.

My questions is, how can I make more use of this data using regression or time series analysis? How can I determin whether the effects are statistically significant?

Thanks a lot!