|
Objectives -- After completing this module, you will be able to:
Class Preparation and Participation: See Topic 4 below -- Everyone has to post to the Week 8 Discussion Board. The instructions are at the bottom of the page because you need to cover the key readings about statistical data analysis before you try to make your posting. Everyone must make a posting this week. Topic 1: Why do we have to be SOPHISTICATED about statistics? Let's start with a very educational and very short video. Statistically Funny Topic 2: Covering the basics Other Not So Funny Basic Stuff on using statistics and the logic of statistical testing. You may want to read my "logic of data analysis" piece before you tackle the open dialogue between Gorard and others that is central to our discussions this week (see below). If you feel comfortable with your basic understanding of statistics, feel free to skip these materials BUT do use these readings in Assignments 3, 4 and 5 please. Once again please refer to Comparative Characteristics of Design Groups. Look this time at the relationships between the nature of the research question the type of design that is appropriate to answer the question, and the kind of statistical analyses (if any) that are most appropriate for typical uses of the design. My cheat sheet -- it's not great, but it may be useful to you. Statistics Cheat Sheet The Statistics Tutor's Quick Guide to Commonly Used Statistical Tests I think this is much better than my cheat sheet. I think it is easier to use and easeir to read. Section 1 which defines key terms explains what these terms mean and why they are important and how to interpret the 'statistic' used to express the term in very simple language with lots of good, graphic examples. If it were me, I would use this instead of the Swisher statistics cheat sheet. Topic 3: What is the relationship between design decisions, key concepts of internal validity, external validity and explanatory power and the use of statistics? The Logic of Data Analysis Using Statistical Techniques (Swisher -- this is an effort to make it easy for you to statistical testing to the key concepts we have discussed in this class -- e.g., internal validity, external validity, and explanatory power. This will help you on Assignment 3 -- I think. At least that's why I created it.) Gott, R. & Duggan, S. (2003) Understanding & Using Scientific Evidence, Sage, Thousand Oaks, CA. Ch. 8, Samples and Populations, pp. 139-159. e-reserve Highly recommended if you have little experience understanding the outputs of statistical analysis and how they can be interpreted. This focuses on the interpretation of statistical results. For you, right now in this class, knowing how to interpret the statistics is probably more important than anything else. Of course, once you get to the point where I ask you to create your own design (Assignment 4), knowing how to pick the right statistical test is more important. I highly recommend that you read this now. Topic 4: Why is this topic so complicated? It seems like there are fairly simple "rules to the game" of using statistics. The most important reading this week is the Gorard reading cited below (not in your textbook). A copy of the article will be available on the Week 8 discussion board in Canvas. This reading of Gorard's discusses ways in which researchers misrepresent or misuse very basic statistical measures. I see these errors all the time. And some people who agree that there is widespread misuse of statistics argue that it doesn't matter, that it has no detrimental effect on our science. Gorard obviously does think it is detrimental, and I tend to agree with him. In either case, I don't want to teach you to misuse statistics and you probably don't want to misuse statistics or make "wrong" statements when you discuss your results. The reading by Gorard is the basis for the required posting this week. Each of you must post this week and everyone needs to read pp. 1-11 in the Gorard et al article. This is where he presents his key concerns about the "misuse" of statistics. There are six commentaries by other authors and a final set of rebuttals by Gorard. I will assign one of you to summarize the key points in pages 1-11. A will assing one of you to each of the responses by Glass, Howe, Putwain, Styles, Van Daal & Ader. These are short readings. One of the commentaries (White) is lengthy and I will assign two people to this author's comments. You will need to determine how to split or share the task between the two of you. The seventh document is Gorard's rebuttal to the commentaries. I will assign two people to Gorard's rebuttals. You will need to determine how to split or share the task between the two of you. Gorard, S., Glass, G.V., Howe, C., Putwain, D. et al. (2014) Open dialogue: The widespread abuse of statistics by researchers: What is the problem and what is the ethical way forward? Psychology of Education Review 38(1), 3-32. You may also want to read this article by Marozzi. He argues for the use of non-parametric statistics as one solution to the "misuse" problem (if you agree that there is a misuse of statistics). I am a conservative on this. I think statistics are "misused" to some degree and I think much of it has to do with a sort of "insistence" on using parametric statistics. My own position is "If you can't meet the assumptions for parametric tests, run a non-parametric test." Yes, the tests are less "powerful" (less like to detect small differences between comparison groups for example), but missing an occasional small difference does not seem to be me to problematic in most of my work. At any rate, read Marozzi. Everything he talks about in medicine are things that are just as prevalent in the social science research. Marozzi, M. (2015) Does bad inference drive out good? Clinical and Experimental Pharmacology & Physiology 42:727-733. doi: 10.1111/1440-1681.12422 |
| Research
Reports None |