Lecture 11
The exam will be a 50-item multiple choice exam online via Canvas. It is weighted at 50% of your overall module mark. You will have 120 minutes (two hours) plus 30 minutes for technical problems; if you have any adjustments, these will be applied automatically. You must attempt the exam in the 24-hour period it is available.
If something comes up unexpectedly that prevents you from completing the exam to the best of your ability, or being able to access/complete it at all, apply for exceptional circumstances.
Questions cover all lectures and tutorials, including reading and interpreting R output. However, you will NOT need to use R yourself to complete the exam - all output will be provided for you in the exam.
All questions will be multiple choice with four alternatives. Questions range from definitions and concepts to reading R output for analyses, figures, and tables and doing some calculations (no more than basic maths).
Below are a list of the topics covered in each lecture and tutorial that you should know for the exam. Note that the tutorial is the one following the lecture.
%>%cor() and cor.test()chisq.test()t.test()lm() function to create a linear modellm() and translating it into the linear model equationlm() function to create a linear modelsummary(), broom::tidy(), and broom::glance()lm() function to create a linear model with multiple predictorssummary(), broom::tidy(), and broom::glance()QuantPsyc::lm.beta()lm() function to create a linear modelanova() for model comparisonTo start, look through your quiz answers to identify where your understanding is already good, and where you need to focus your revision. Once you’ve got the lay of the land, a fantastic way to prep for the exam is to work through the practicals again. Do the tasks, don’t just read the answers - even the hard ones. If you can do this you’ll be well on your way to acing the exam.
You can also test yourself with the Escape Room. The Escape Room draws on many of the skills reading and interpreting output that the exam will also expect (although unfortunately the exam will not be as fun).
If you find yourself still struggling with particular ideas, you can also look for other resources. You can find endless examples of explainers, blog posts, YouTube tutorials, Khan Academy videos, etc on all of the topics we have covered this term. Just keep in mind that you must be able to read and interpret the output that we have focused on in this module.
Finally, use Piazza as you revise. You can ask questions there, or you can look back on previous answers if a similar question has already been asked. If you’re feeling confident about your knowledge, you can have a go answering other people’s questions as well!
The sample exam is available as an unmarked Quiz on Canvas. It’s half the length of the regular exam, but has very similar questions in length, complexity, and style. You can take it as many times as you like to practice.
As part of a research study run by MSc researcher Rich Symon and supervised by Dr Mankin, we are offering optional two-hour revision sessions on Zoom focusing on the linear model. In the session, you will complete a few questionnaires, go through a guided revision of the linear model, and practice with a sample exam question.
You can attend the session even if you don’t want to do the study part, and all materials will be made available to everyone after the last session. There will be a prize draw of five £10 Amazon gift vouchers for those who do participate in the study.
If you would like to attend, sign up on Canvas Calendar for one of the sessions, taking place on May 18th and 19th (just before the exam). See this announcement for more details.
Here are some suggestions for how to prepare for the exam. Naturally, you should also consider your own preferences and strengths in revision.
Look through the topics above and make a list of key concepts (e.g. “p-value”, “standard error”). Then, create a glossary of terms, filling out the entry for each term with a definition (mathematical and/or in words), examples, and notes. You can even team up with friends on the module by having a shared document - for example, via Google Docs or similar - that you compile together.
We recently sent out a list of maths revision resources to help you strengthen your maths muscles. You don’t need to be a mathematician to do well on this exam, but as you know, there are a lot of numbers in this module, so being comfortable with maths concepts will help.
It will also be very beneficial to be able to ballpark numbers. Imagine you see a problem like this:
Assuming the sampling distribution is normally distributed, what is the 95% confidence interval for M = 6.4 and SE = 0.89?
First we need to remember that in this scenario, we can calculate the 95% confidence interval by multiplying the standard error by 1.96. Now I don’t know about you, but I can’t multiply 0.89 * 1.96 in my head. But I don’t need to do this exactly - I can ballpark instead. 1.96 is almost 2, and 0.89 is just a bit less than one. So, I would expect to get a little less than 2 if I multiplied them together. Then, I just need to add 2 and subtract 2 from the mean to get an estimated range - a little more than 4 for the lower bound, and a smidge over 8 for the upper bound, give or take.
This won’t give me exact numbers, but it means that if my calculator tells me the CIs are, for instance 147 and 329, I know I’ve gone quite wrong and made a mistake somewhere! I can also easily see that A and C can’t be the right answers - they’re way too big and too small respectively. If I want to be sure of the exact numbers, I can do the calculation knowing more or less what I’m going to get.
We discuss this idea and lots of other tips and advice for exam preparation in last week’s StatsChats!