Here is my suggestion ( I am not including pre-processing or post processing methods of images, I assume you know this)
1- You set a time interval to study the NDVI based on the availability of data and your study area (the change of NDVI in desert cities is different than that in other cities).
2- Of course you gonna use Landsat, so for the missing date, you gonna use SPOT images with better spatial resolution.
3- You usually do not enhance the spatial resolution of Landsat images because these images are the majority and the errors produced of each fusion will be huge for all images. what you do is to decrease the SPOT images spatial resolution to 30m, if that makes sense?. I even do not recommend fusing the images for NDVI analysis because their radiometric values will not be retained and you gonna lose a lot of information as Lurker said.
4- Do the statistical analysis.