Is the t-test appropriate for testing for changes in bowlers' speeds [Research Proposal Help]? I am currently doing a master's research proposal (experiment hasn't been completed yet) and need to decide what statistical test to use. 
My research question is observing increases in bowling speeds over 18 balls (dependent variable) following a stimulation (independent variable) within trained adolescent fast bowlers.  
All subjects will be carrying out the exercise stimulation just the once.  
Am I right in thinking this is a one sample t-test? 
Would I need to include any other statistical tests? 
 A: At this point, you probably have your Masters, if not your Ph.D., so congratulations!! However, we all must do out part to slowly shrink the yawning void of unanswered questions, so I will take a stab here.
There are many tests you can use, but they depend on what question are you trying to answer. For example, in your case, you are interested solely in an INCREASE in speed. You are equivalently uninterested in the case of "no-change" as you are in the case of "decrease". As such, you can consider one-sided tests, as there is only one interesting direction for you. Were your question about changes in speed, you would more naturally be interested in a two-sided test.
That addresses the sidedness(?). The other consideration is which test to use. There are many assumptions that underlie the most common tests which are important, and sometimes ignored. Specifically, you are considering the "one sample t-test" with the assumption that bowling speeds follow the same distribution regardless of an intervention. Allowing for a bit of simplification, this assumes that the distribution being tested for Null vs. Alternate is Gaussian. This allows us to assume the mean and variance are uncorrelated and that the variance has some nice qualities. Due to the Central Limit Theorem, if your sample is large enough, that's fine. There are other tests which may look similar but have slightly different, or even more relaxed assumptions. For example, you could rephrase the question as a two-sample test. If you have two seperate groups, for example control and affected, you can test a null hypothesis of same mean against an alternate ]one-sided/two-sided] hypothesis of [larger/different] means. In this case you could use a two sided t-test that assumes the same variance for the two populations (even if means differ), a two-sided Welch test if the two variances may be different, or a Mann-Whitney-Wilcoxon which dispenses with the normal assumption completely and deals with the ranks. There are pros and cons to each one, but mentally precisely phrasing your question will help a lot in determining how to more efficiently test your hypothesis.
