Suppose you hear an Solution There are, of course, many possibilities here. One is to look at juvenile delinquency crime rates (JD). You could decide to look at the veracity of the statement by the “old-timer” in, say, your home state of New York (have no idea where you live). A testable null hypothesis would be that the JD crime rate last year in New York was equal to the year 1950. That should be back in the days of the “old-timer”. Other periods could be selected such as a five year period or a decade. So, the null hypothesis, Ho, would be Ho: JD (Year 2010) = JD (Year 1950). The alternative, hypothesis, Ha, is Ha: JD (Year 2010) > JD (Year 1950). This is a one sided hypothesis because the “old-timer” said the latest rate would be higher than the earlier rate – that’s what he actually believes to be the true state of nature. To test this hypothesis you could select a random sample of, say, 20 counties, cities or other governmental units which have crime data for the periods under consideration and then get the JD rate for all units and time periods in the sample. This is considered a paired experiment because a rate for one period is paired to the rate for the same unit but at a different time. A paired t-test could then be used to test Ho. Or, a similar non-parametric test if you judged that the data was not normally distributed..