Jump to content
Ryn

(2016 Edition) - Predict your chances for Ontario schools. Generate your OLSAS cGPA, L2, & B3

Recommended Posts

(This is a follow up to this old thread. If you want some background on how this whole thing started, look there.)

 

Check it out if you're impatient or if you only want to know your OLSAS GPA stats.

 

Hey everyone! It's your favourite amateur statistician who, with two weeks to go before 2L starts, has nothing better to do than work on things like this ;) Many of you probably remember the old chances predictor. Well, I've updated it for 2016 with fresh data and a new model. I had this whole long explanation typed up with extended details, but Chrome is pretty useless and closes with no confirmation if you accidentally press Ctrl+Q (I tried to press Ctrl+Tab to switch tabs. Oops), so all of it is lost. I guess it's an opportunity to exercise brevity. So off we go:

 

Data

 

I gathered about 1,000 data points from 2011-2016 from the accepted and rejected threads on the forums. I included, in addition to cGPA and LSAT like last time, L2 scores (more on that in a bit). I used these data points to generate models for five Ontario schools: Ottawa, U of T, Queen's, Osgoode, and Western. I didn't include Windsor or Lakehead, because like last time there either was no significant correlation between stats and acceptances (Windsor) or there wasn't enough data (Lakehead). I thought about expanding to other provinces, but the OLSAS scale that Ontario uses makes things really simple; once you get out of province, standardizing the data becomes much more difficult.

 

For the purposes of the binary logistic regressions, I used significance of p <= 0.2 (hey, this isn't hard science; get out of here with your p <= 0.1 or p <= 0.05). Anything above that I just discarded. Cutoffs for the accepted/rejected points varied between 0.5 and 0.78, depending on the school. I was trying to shoot for at least 80% accuracy in both accepted/rejected, with a bit of bias towards false negatives (I'm pessimistic by default) than the opposite.

 

Model

 

I took a different approach than last time, where I only regressed CGPA and LSAT. This time, I also took in to consideration L2 as well as the effect of moderators. What's a moderator, you ask? A moderator is a variable that has an effect on another variable. For example, if I say LSAT moderates cGPA, it means that a high LSAT can potentially boost the effect of a low cGPA in order to improve your chances (this is a fairly simplified explanation but it will suffice for our purposes; if you're super curious, here's a more detailed explanation). 

 

The result was that three schools (Osgoode, Queen's, and Western) had significantly stronger models when I included a variable describing LSAT's moderation of cGPA. Unfortunately, the effects of L2 on any of the schools was inconclusive. This likely stemmed from the fact that not everyone included their L2 scores (especially in rejected threads) and thus I had to throw out the variable completely. It was the same case with U of T and B3, though it's very clear that B3 would have been a better choice than cGPA (ever so slightly, anyway). This is a limitation of the state of the data and not reflective of the schools admissions practices per se.

 

Results

 

  • U of T: I wasn't able to use B3 like I wanted, but cGPA is only slightly less accurate (at least given the data set). The model accounts for about 54% of the effect on admission.
  • Queen's: LSAT moderates cGPA. L2 wasn't significant enough to include. The model accounts for about 60% of the effect on admission.
  • Western: LSAT moderates cGPA. Again, L2 wasn't significant enough to include. Accounts for about 47% of the effect on admission.
  • Osgoode: LSAT moderates cGPA. L2 isn't a factor at all at Oz (unlike with Western and Queen's, where it was easy to see there was an effect but it wasn't usable for the model). Accounts for about 61% of the effect on admission.
  • Ottawa: Straight up LSAT and cGPA. In fact, LSAT is nearly irrelevant, as was evidenced last time. L2 has no effect either. Ottawa must also have some sort of X factor, however, as only 25% of one's admission is accounted for by this data.

The table on this page gives some more stats, including means and medians of accepted applicants for each school.

 

 

Conclusion

 

I'm pleased with how the updated model turned out, especially with the moderation interactions between cGPA and LSAT. You can check it out yourself here. The results page is a bit different; in lieu of raw percentages, I opted for displaying a range of chances along with a description of what that range means. I think this method is a bit better in presenting the information as raw numbers tend to be somewhat meaningless for many people (including me).

 

As before, the actual predictions are useless if you're a splitter (very high LSAT and very low cGPA or vice-versa). And finally, don't use the predictions to make life-altering decisions; it's not a substitute for critical thought and the data could very well be extremely wrong.

 

Check it out here.

  • Like 11

Share this post


Link to post
Share on other sites

Hey all -

 

Just a quick note that this thread should not turn into a generic "chances" thread like the last one did on occasion.

 

So if you just want to know your chances, post a new topic WITH YOUR GPA AND LSAT INCLUDED IN THE TITLE and make sure to mention where you are applying, applicable last two / best two calculations, and whatever soft factors you deem relevant. Helpful to note if it isn't the regular category too - mature and access categories have different criteria.

 

Any chances posts that ignore this request will be spliced into their own topic and I may even use an amusing title.

 

Thanks :)

  • Like 2

Share this post


Link to post
Share on other sites

After punching in the CGPA and LSAT values, I noticed that as my CGPA decreases; I have a better chance at getting into Queens. 

 

LSAT 157

L2 3.78

(CGPA 3.78)

 

When I punch in LSAT 157, CGPA 3.5; I seem to have better chances at getting into Queens. Something wrong with the formula?

 

(strong LORs and ECs -- I founded a social enterprise that focuses on food literacy and skills training)

Edited by TielliFan
  • Like 1

Share this post


Link to post
Share on other sites

According to program, I have better chances at getting into Queens with a CGPA of 3.5 than I do with a CGPA of a 3.78 (LSAT the same). Something wrong with the formula maybe?

 

EDIT: It appears that there *is* something wrong with the Queens predictor as it is suggesting that I would have an excellent chance of getting in with a 2.5 CGPA... lol.

Edited by TielliFan

Share this post


Link to post
Share on other sites

Both Queen's and Western act weird because of the moderation effect, since the result isn't linear. It's more accurate but because I don't think my sample size for "rejected" data points is large enough, the equation does weird stuff.

 

I'm probably going to swap it out for a linear equation, because while it's less accurate it doesn't give stupid results with odd GPA/LSAT combos.

  • Like 2

Share this post


Link to post
Share on other sites

Okay I had some time and fixed Western and Queen's.

 

I also changed Osgoode's to a linear formula. I was debating keeping the moderation element in it, as it only acted weird if you put in a cGPA of like 2.0 and an LSAT below 140, but I decided against it. I feel like it's better if people see consistently predictable (and linear) results than some weird combination because it can be confusing, despite the fact that it's (marginally) less accurate (though when dealing with odds and chances, what is "accurate" anyway?). If I had more samples of rejected stats, the formulas could be better but unfortunately there aren't many posted here that I can use.

  • Like 1

Share this post


Link to post
Share on other sites

This is great thanks. Quick question, does the "year" section refer to the year that you took the course in or the level of the course? 

Share this post


Link to post
Share on other sites

This is great thanks. Quick question, does the "year" section refer to the year that you took the course in or the level of the course?

The year you took it in.

Share this post


Link to post
Share on other sites

Would someone explain that "cutoff" means in the chances calculator? For Ottawa I got a 78% cutoff... what would that mean?

Share this post


Link to post
Share on other sites

Would someone explain that "cutoff" means in the chances calculator? For Ottawa I got a 78% cutoff... what would that mean?

 

The cutoff is the chance percentage at which your odds of being accepted drops off dramatically. This percent is derived from testing the algorithm against existing data and seeing how accurate it is. Below this "cutoff" percentage, the model became very inaccurate. 

 

I can be more specific if you're actually interested, but the tl;dr is that it's not specific to you but to each school generally, and won't change based on what you enter. The actual prediction (likely, unlikely, etc) takes in to account this cutoff when displaying your chances.

Share this post


Link to post
Share on other sites

Any idea how accurate the GPA calculator is? Like in terms of determining how OLSAS will calculate it?

Share this post


Link to post
Share on other sites

Any idea how accurate the GPA calculator is? Like in terms of determining how OLSAS will calculate it?

Obviously I can't make any guarantees but it's as accurate as I believe it can be given the information we have about the conversion scale. The cumulative GPA, anyway. This assumes that you attended only one undergrad institution, though. Once you insert a factor like having transferred universities, it becomes unclear how OLSAS takes it in to consideration.

 

The last/best scores are estimates for sure, because each school will be using a different formula on what it considers a year, etc. And anyway, OLSAS doesn't generate those scores.

 

So yeah I think it's pretty good but your real OLSAS GPA might differ from what you get on the calculator. The app should only be used to ballpark your competitiveness and not as definitive.

  • Like 1

Share this post


Link to post
Share on other sites

How would you suggest that I fill in my grades for the cGPA, given that I completed my first year of school at one university, and then transferred to another for the rest of my undergrad?

Share this post


Link to post
Share on other sites

How would you suggest that I fill in my grades for the cGPA, given that I completed my first year of school at one university, and then transferred to another for the rest of my undergrad?

I don't know how OLSAS handles this situation. My suggestion would be to call them and ask! Then let me know here so I can update how the app works :)

Share this post


Link to post
Share on other sites

How would you suggest that I fill in my grades for the cGPA, given that I completed my first year of school at one university, and then transferred to another for the rest of my undergrad?

 

 You should probably convert your first year grades to the scale your second school uses, then enter them all into the calculator as if they came from the second school - I believe this will give you your cGPA on OLSAS's scale.

Share this post


Link to post
Share on other sites

How would you suggest that I fill in my grades for the cGPA, given that I completed my first year of school at one university, and then transferred to another for the rest of my undergrad?

 

 

Also, but this only applies if both of your schools use the same number according to this table: https://www.ouac.on.ca/guide/olsas-conversion-table/, then I think you can just input all of your grades together into the same calculation when using the calculator (since there would be no difference from the perspective of the calculation). 

Share this post


Link to post
Share on other sites

Also, but this only applies if both of your schools use the same number according to this table: https://www.ouac.on.ca/guide/olsas-conversion-table/, then I think you can just input all of your grades together into the same calculation when using the calculator (since there would be no difference from the perspective of the calculation).

 

Oh good idea, they do use the same scale. Do you know if OLSAS averages all your grades and then converts to their GPA or vice versa?

Share this post


Link to post
Share on other sites

I don't know how OLSAS handles this situation. My suggestion would be to call them and ask! Then let me know here so I can update how the app works :)

Haha, I'd feel a bit awkward calling them, but perhaps once I need more tools of procrastination during exam season I will.

Share this post


Link to post
Share on other sites

Create an account or sign in to comment

You need to be a member in order to leave a comment

Create an account

Sign up for a new account in our community. It's easy!

Register a new account

Sign in

Already have an account? Sign in here.

Sign In Now

  • Recently Browsing   0 members

    No registered users viewing this page.

×