T-test is used when you do not know the standard deviation of the population (which is almost always). or in other words, if you know the SD of the sample.
z-test is mostly used when standard deviation of the pop. is given, or for proportion testing or estimating the binomial distribution. It should specifically say something like "the standard deviation for the population of students in ohio is .... sample was taken from students in 4 counties". If you're not sure I'd just go with a t-test. and between matched pairs and 2-sample t test, 2 sample requires independent samples.