Examples Sheet 3 1-6. Most knew what to do, if not why to do it. They all looked up the appropriate test from the notes and carried it out. Three simply didn't understand what was going on. 4c and 5b caused some problems: several simply didn't know what to do in 4c, perhaps through confusing themselves with lots of numbers, and there was a fair bit of confusion about what "Type II Error" meant in 5c: what does it mean to condition on H0 being false? 7. Many didn't realise that they could work out the MLE very easily: they didn't connect this part of the course with what had come previously. They just looked up MLE in the linear regression part of the notes, and decided that MLE=LSE only when the X coordinates are balanced about 0. The rest of the question was fine. 8. There were a three silly models. The rest were mostly fine: mostly general linear regressions, but there were a few who treated the initial temperature as exact in setting up the model, but as random in doing the regression. 9. The first part was fine. Everyone was led astray in the second part, because they looked up the confidence interval for an observation rather than for the true value. 10. Only two really knew what they were doing. Around half did the right test. 11. No one attempted it. Typically, half way through my explanation, they would say something like "Oh, is that boot.. err bootlace estimation?" I remember that last year at least some of my students tried it. But they seemed to think it wasn't a real question. 12. Supervised on the day of the lecture. There was lots of confusion about what the question meant (though mostly because people had skipped the lecture!)