# More on rounding

Sudden thought at 6am prompted by a student who is 0.01% off an A…. Of course there are situations where rounding has to happen. It is the second decimal place.  The grading scale defined in my syllabus p.6 is

###### A 95-100% A- 90-94.9% B+ 87-89.9% B 83-86.9% B- 80-82.9% C+ 75-79.9% C 70-74.9% D 60-69.9% Fail <60

The gradebook works to two decimal places; my syllabus defines the cut-offs to one decimal place, meaning that I have to round second decimal places. Thus, 94.90-94.94% = 94.9%=A-; 94.95%-94.99%=95% = A.  So I was NOT wrong when I said in class that rounding happens, I was just wrong to say it happened at first decimal place.

Incredibly, this second decimal place issue did affect the grade of FOUR students, so we now have one more A, one more A- and two more B+’s than we did yesterday. Merry Christmas.

# Rounding

I made a ghastly mis-speak in class last week. I was asked if grades get rounded, and I said yes. WRONG Andrew. You can’t round grades. The grade distribution has to be exactly as in the syllabus (p. 6). By University decree, I am not allowed to change that even if I want to. So there can be no rounding. Period.

What was I thinking? Now I am going to get a ton of emails demanding a bit here and bit there, and I will have to point them all to this post, and the highly related one of yesterday.

Sorry folks. This is why I hate speaking off the cuff about grading issues in class. We all need to read the syllabus.

# The bottom line for 2015

The class average: 87% (B+). Of the 352 students we started with, 326 made it to the end. At least half the losses were from people trying to maintain their GPAs. Among the finishers, 45% of the finishers got some kind of an A, 67% got a B+ or better, and 79% got a B or better. One student scored over 100% through extra credit. Compared to last year, more A’s but also more students on a B- or less, the net effect being an almost identical average, as with 2013.

I say it every year, but there is simply no way to know what to make of this grade distribution. Am I setting the bar too high, too low, or just about right? It is one of the mysteries of higher education. I have heard it said in some places that an ideal grade distribution is about 20% getting some sort of A. Where does that figure come from? And why should students be competing against each other, rather than an absolute level of achievement?

# Final exam 2015

The final exam went without a hiccup (cf last year’s fiasco!).  The only two questions which tripped a lot of students were the two covering material from the last week of semester, when attendance was shockingly poor. Obviously I take that as evidence that coming to class matters… The class average was 87% (B+) among the 319 students who took the exam. Seven students did not take it, three of whom look like they have dropped the course (\$10 says I am about to get three emails with a variety of explanations for why I should exempt those students the final exam. If you are one of those, my response is here.).

The average performance was about in line with Class Test 4. Altogether, 72 students got 100% on my ask-28-questions-grade-out-of-25 algorithm. Again, no one got everything correct, but 14 students got 26/28. There were eleven fails. Setting the final exam into context, the picture looks like this (total number of students on the y-axis).
So all in all, lots of improvement over the year; not much in the week between Class Test 4 and the Final. I wonder if I should do away with Class Test 4? Probably not. Psychologically important I think. Plus it plays to my take the-two-best-of-four-tests algorithm, and my philosophy that students should be able to fall over once or even twice and, if they improve, not be penalized for that.

It is that time of year when I get inundated by emails for the sort:  Dear Professor. The grade I earned on SC200 is not what I hoped it would be. I need a higher grade to (a) stay in the country, (b) keep my funding up, (c) get into Schreyer’s Honors College, (d) get into/stay in Smeal Business College…. Please can you increase my grade?

Sometimes that message is prefaced with ‘Sorry to bother you‘, and sometimes it comes with comments like ‘I know I really mucked up  but…‘, and sometimes with ‘I worked really hard and came to class….‘, and sometimes it includes a statement like ‘I don’t think I should be penalized for my poor performance on a Gen Ed course when my classes in my major went so well‘, and other times with ‘Yours was my favorite class and I was relying on it to get me…‘.

I find it all really tough. Some of the stories are heart rendering. They all make me want to scream one of: Why the hell didn’t you work harder/engage more/get better organized/treat me and my class with respect/listen to my repeated warnings/look at your grade on Angel when you still had time to save it/take the extra credit options when they were available?

A group of professors are meeting to explore better teaching practice in Gen Ed. At today’s session, I raised the issue. The 15 faculty rose up as one. Bottom line: You have to stick to the syllabus grade algorithm. It is what the university mandates. And it is only fair to the other students who worked hard and performed well. Making concessions to individuals is unfair — and that way lies madness.

Well, that’s always been my philosophy too. Standards are standards. And Freshmen especially, as most of my class are, have to learn early in life that you have to earn it. Get organized, engage with it, do it, take control of your own education – and earn it.

# bums on seats

Coming to class is important for SC200. It’s not just me that thinks that: the students always tell me that, especially those that realize too late that we have no text book and you can’t wing this class from the powerpoint slides. But no matter how or how much I tell them that, it transpires that you have to incentivize them ahead of time. So after many years of ruminating (and complaints from students that I did not tell them strongly enough how much it matters (!)), I decided to go brutal. I randomly take attendance, and for 10% of the final grade, they have to be at 9 of at least 12 of those attendance takings. If they are not, they get none (zero, zip, nuttin) of that 10%. That 10% percent can lead to a very serious grade drop (A to B, B to C etc).

And with that algorithm, this was the attendance pattern for 2015:
So over 80% attendance until the week after Thanksgiving, by which time most students had hit the required nine, and by the end, c.95% (308/326 students) had hit the attendance requirement (remarkably similar to 2014 when the requirement was 7 of 10). [Note that the denominator for all these data are the number of students who finished the course; there are likely bad attenders who dropped the course for whom the algorithm failed to work.]

Of course, after I had taken attendance for the 12th time (second last class of the year), there were a  number of students sitting on 8, desperate for one last change to hit the magic 9. Here they are at the end of the last class of the year, lined up to sign on for their 10%.
Naturally, I have since had emails from some of the 18 who never hit the mark offering a variety of explanations about why they deserve the 10% for attendance even though they could not be bothered attending regularly. This email begging is definitely a downside of the system. But upsides:

• I don’t have to deal with paperwork for absences: under this algorithm, you can have quite a few of life’s little traumas (and even the odd hangover) and you are not in trouble. You just can’t persistently miss class.
• I get data on individual attendance. So when someone is moaning about their test performance, I can see its because they seldom came to class.
• Attendance itself is high until after Thanksgiving; without this arrangement, it sags a month earlier.
• I can incentivize attendance early on, when I need them to focus on the pop quizzes which are the practice for the class tests, and then I can do nothing for a while to incentivize them to hang on until the end.
• Next year, I think I should take attendence for the 9th time after Thanksgiving….

Finally, in the last class of the year I held a discussion about whether this attendance scheme was good. No one — and I mean no one — thought it was a bad idea. I wonder what the 50% students who were absent made of it.

# Blog Period 3 — the end

I am never sure what to make of the third blog period. I take the best of the three periods as the final grade. This means there are a bunch of people who leave it to the last minute and then post something half-assed because they never got their act together all semester. There are others who try very hard but haven’t done anything before so they have no feedback to work off and so struggle to do well. And then there are some who use the feedback on previous blog periods and blow it out of the water.

So, for what it is worth, the average score was 68.4% excluding the 153 students who did nothing because they had done very well earlier in semester or could not be bothered trying to improve their previously average attempts. The distribution broke down as A, 4; A-, 8: B+, 8; B, 30, B-, 18; C+, 37, C, 18; D, 17; Fail 36. If I get time, I’ll return to this post and give links to some of the excellent work we got this blog period.

# Class Test 4

Finally, over a week after Class Test 4 actually happened, I get some time to post the results. Most notable, the class average (excluding the 45 no shows) was 87.4% (B+), up from the previous tests. But again, its the distribution that’s really impressive. Altogether, 121 students (over a third of the class) got an A, 94 of whom got 100% on my ask-28-questions-grade-out-of-25 algorithm (as always, no one got everything correct). You can clearly see from the graphic the steady student improvement across the semester (the y-axis is number of students). The distribution moves beautifully from a right skew to a left skew as time goes by. Dare I hope this is a sign of learning?
Still, 18 students failed and 12 got a D. Where were those students at the final review session?
This is the view at the final review session of the year, my last visual contact with students from the class of 2015. Only a few of these few were there because they did disastrously on the test. Most had done well and wanted to do better… Trying to get those who fail or get D’s to do something about it seems to be one of the insoluble problems of Gen Ed teaching.

# The \$100 challenge

Scientists are all over each other looking for flaws in ideas and data.  This is what makes science so powerful: relentless peer review. It’s why we can all be pretty confident about scientific consensus in fields we do not have technical expertise to assess, like climate change. Yet powerful forces in the (mostly US) public arena believe deeply that scientists can go horribly wrong because scientists go in for mass delusion or worse, mass conspiracy to get grant money (scientists could put together a conspiracy if they tried).  For a classic and truly frightening example of this, brace yourself and check this out.

I try various ways to get all this across in class. I talk about Lysenko and what he did to Soviet plant genetics (and hunger levels). Abetted by politicians, Lysenko was successful by negating the scientific process (not least by having scientific critics killed or imprisoned). I also talk about the trials and tribulations of Penn State’s very own Mike Mann, who has been harassed by politicians who think they know his science better than he or his colleagues do.

But to get the point across, I hit on a fabulous scheme. I said in class, and followed up in an email:

” I will give \$100 to the first person who finds an example where bad, fraudulent, mistaken or incorrect science was first demonstrated by someone other than a professional scientist (e.g. a politician, lawyer, lobbyist, concerned parent). To play safe, the example should come for the 20th or 21st Century. I contend that the process of science (formal and informal peer review) does more to keep science honest than anything else. Be the first to prove me wrong and you are \$100 richer!”

I thought that would be the end of it, as it had been when I made the same challenge before. But not this time!! Here is me losing my \$100 in front of 300 students. After a spirited and very thoughtful Thanksgiving e-discussion with Isaac Will, I decided he had found a case. Dr Andrew Wakefield did terrible damage to public health by linking the MMR vaccine to autism in a paper in The Lancet. Many, many scientists subsequently showed with different data there was no link, and I believe that their scientific work was most important in debunking Wakefield’s mischief. But that still left Wakefield’s original data – all rather poor but nonetheless unarguably still there. Issac argued that investigative journalist Brian Deer showed just what crap it was.  He figured out that Wakefield had terrible financial conflicts of interest AND that his patients were not random cases but carefully chosen to try to show an MMR-autism link. Deer’s work demonstrated that Wakefield’s poor (shockingly poor) science was actually fraudulent. This led to the retraction of the paper by The Lancet. The story is well summarized here. In essence, Deer showed the only ‘data’ supporting Wakefield’s case was garbage, thereby taking the final study off the table. I felt Issac had a point: Deer indeed revealed important scientific weaknesses not discovered by scientists. So I became \$100 poorer.

But oh, what a teachable moment. One case in over 100 years. Next year, I’ll post the \$100 challenge again. Can anyone think of another case?