Missing Values

Hi everybody,

I have a problem with analyzing data due to many missing values.

Subjects have to press a button in response to a display. The reaction time is measured. 8 blocks, 96 trials, block 1-6 and 8 are the same, block 8 is different.
One expects an increase in reation time in block 7.
Than I have two conditions easy and difficult.
For each block I compute the mean reaction time by taking the median over groups of twelve values and than computing the mean over the 8 medians.
Than I just compare whether those means differ over the blocks. Thats clear.

Now I have two conditions: Easy and difficult.
In the easy condition the subjects make almost no mistakes. In the difficult conditions subjects produce up to 60 mistakes (mistakes/error = missing values).

In the easy condition I see in the reaction time that they get faster over time and in block 7 they are slower (intended by the design). The error rate stays the same.
Inthe difficult condition, the number of errors fluctuates strongly, Starting with many errors, reducing the rate of errors and in block 7 I see a steep increase in the error rate. Also the reaction time goes up.
E.g. Number of data points over blocks:
86 86 89 91 90 90 80 89

I now wonder how I can compare the reaction times between blocks in the difficult condition despite the big difference in the amount of data point I have.

Can I compare the different conditions(difficult-easy) despite the difference in the number of data points.

What kind of analysis do I need. Where do I have to be careful. I thing unbalanced design and regression analysis are the keywords here, but I still can't figure out what to do.