pointOfive
commented
11 months ago HW
-
expanded the header explaining MarkUs autotesting and marking based on updated iterations of that which were made through the first round of content creation
-
Q0: I need to double check that "\%" prints out correctly in MarkUs... probably does :)
- printout generally looked great; although, I revised the prompt and expanded things little bit to better emphasize the important aspects of this prompt and better address the prompt
-
Q1: minor edits
-
Q2: I think adding line breaks is helpful and do so here and will continue to do so
-
Q3: small edits
-
Q4: p-value calculation was accidentally one sided: I may have made this mistake in my recording
- revised what was a fairly good answer where I felt improvement was possible
- My revision on the Q4 hint/explanation got me thinking that this (the general way we use an observed test statistic to critic if a null hypothesis might be wrong) should be emphasized in the week 6 tutorial
- I think it's okay that we didn't do this in week 5 as there we're first emphasizing the process and routine of computing and using p-values
-
Q5: trying to move the language more towards "evidence" statements
- Please change this to a multiple choice question, replacing the jpg image I added with the following:
A. $0.10 < \text{p-value}$ no evidence against the null hypothesis
B. $0.05<\text{p-value}<0.10$ weak evidence against the null hypothesis
C. $0.01<\text{p-value}<0.05$ moderate evidence against the null hypothesis
D. $0.001<\text{p-value}<0.01$ strong evidence against the null hypothesis
E. $\text{p-value}<0.001$ very strong evidence against the null hypothesis
-
Q5: keep the hint the same, but reword the prompt so it's just a multiple choice question (with no restated null hypothesis summary... they've already done that above so it's okay to not do it again)
- I have "hidden" the ABCD answers in the
testvariables for other multiple choice questions: please do that for this one as well
- I have "hidden" the ABCD answers in the
-
Q6 is fun/good
new questions requested
-
new Q7(s)[? it is probably going to be necessary to make this into multiple questions]: have students use a couple non-parametric tests and a parametric p-value calculation
scipy.stats.median_testwhich assumes the medians of the two groups are identicalscipy.stats.mannwhitneyuwhich (more strongly) assumes "no actual difference between groups"scipy.stats.ttest_indwhich assumes the means of the two groups are identical and that the samples come from normally distributed populations
-
have students answer multiple choice questions about which of these are nonparametric/parametric and why: the answers should be in the homework explanation/guidance already, so these should just be a question of making some correct explanations and some mixed up explanations
-
I would also like a question that reminds students of the code based on
scipy.stats.binomfor one-sample p-values for proportions problems; andscipy.stats.ttest_1samp; and helps/asks them to differentiate these from the newly introduced tests up above. This should again just be a question of making some correct explanations and some mixed up explanations. -
new Q8: create a confidence interval of the difference between the two groups. Guide the students to use a random number seed (just as you've done nicely in Q8) and how a two-sample interval gets constructed with bootstrapping and create an autotest to confirm this.
back to original questions
- original Q7: please reformulate the prompt and answer to be analogous to the reformulation of Q0 (but without hints this time). [I.e., use the Q0 form and replace the parts with pieces from Q7]
- I like that the one-sided hypothesis test is introduced here, and we should indeed do this; however, a hint/explanation as to how this works needs to be introduced and included with this question. We need to explain what we're asking them to do and how it's different. This has to do with no longer needing the
abstreatment in the "as or more extreme" calculation. - You'll need to determine how to alter the updated statement of the prompt and/or other problem information to help guide the students as to the difference between what you're asking here and the original two-sided testing style that they're more familiar with
- It's a really good problem to introduce this; but, we need to make this problem an opportunity to help explain what's actually happening here (even if we may have done a little bit of this previously)
- I like that the one-sided hypothesis test is introduced here, and we should indeed do this; however, a hint/explanation as to how this works needs to be introduced and included with this question. We need to explain what we're asking them to do and how it's different. This has to do with no longer needing the
- original Q8: reworked prompt slightly as I think it didn't quite state everything that was asked for
- I've added a hint that I think is necessary/helpful here
- original Q9: keep this but convert it to the style of Q5 (and use the ABCDE options I provide above)
- original Q10: keep this; but, we need to explain what Type I and II errors are. We need an explanation section that reminds students that if we make a formal choice "to reject" or "fail to reject" then we might make an error. Then ask the students the question :)
- original Q11: looks good
- Optional Q12-Q15: these feel like they might better go in the tutorial. Just leave them in the notebook for now and I'll make a final determination on how to use them soon.
MatthewYu06
Homework:
Tutorial: