Experimental Design B/C

User avatar
Panda Weasley
Member
Member
Posts: 133
Joined: September 27th, 2014, 6:24 am
Division: C
Has thanked: 0
Been thanked: 2 times

Re: Experimental Design B/C

Post by Panda Weasley »

Sasstiel wrote:Hi! This is my first time doing this event and im a bit nervous...what should i expect that i wouldn't already know?
In terms of the event: Be ready to not have as much room/materials as you expect or are used to. Also time passes faster during events, so keep an eye on the clock. If you are doing practice experiments you should be fine though.

In terms of topics: Normally you only need to know basic physics and chemistry. There have been times when I've had a topic I had no knowledge of, but that doesn't happen often. In those cases my advice is don't freak out. It's highly likely that the other teams also don't know what to do. Just use your best judgement. A lot of the points are just based on completeness, so even if you don't think you experiment is great you get points for just doing the write up.
I wouldn't worry to much about this potential problem because, as I said previously, this doesn't seem to be a common issue.
Llamastwaimzjf wrote:Thanks. What are some tips on the analysis of results? When I went to an invitational not long ago we got a low score on that.
I would check the rubric to make sure you did all of the required elements. If you know that there is a part of it you have trouble doing practice just that part so that you feel more comfortable (for example if you struggle with the 'possible experimental errors' section practice thinking about ways the experiment could be "messed up" or biased).
DFTBA!
Events 2019: Forensics and Fossils
Proud member of Teh Ento Cult.
:ugeek:
Mortem_Haedo
Member
Member
Posts: 20
Joined: March 2nd, 2014, 4:38 pm
Division: B
State: OH
Has thanked: 0
Been thanked: 0

Re: Experimental Design B/C

Post by Mortem_Haedo »

kidcoder923 wrote:
Llamastwaimzjf wrote:Thank you. Also, what my team did is we took the requirements in the write up and split them up. Since my other two partners are new and younger than me I'm doing statistics. For statistics, I include the average,range, mode, median, mean deviation and when we went to invitationals in that part we got 5/6 points what were we missing?
check your work, did you forget to do one? I also do statistics and often find that I sometimes forget one or two
I always do 6+ stats to be safe :) For example- average, range, mode, median, standard deviation, IQR (Interquartile Range), and line regression.
Best of luck!
User avatar
Panda Weasley
Member
Member
Posts: 133
Joined: September 27th, 2014, 6:24 am
Division: C
Has thanked: 0
Been thanked: 2 times

Re: Experimental Design B/C

Post by Panda Weasley »

Mortem_Haedo wrote:
kidcoder923 wrote:
Llamastwaimzjf wrote:Thank you. Also, what my team did is we took the requirements in the write up and split them up. Since my other two partners are new and younger than me I'm doing statistics. For statistics, I include the average,range, mode, median, mean deviation and when we went to invitationals in that part we got 5/6 points what were we missing?
check your work, did you forget to do one? I also do statistics and often find that I sometimes forget one or two
I always do 6+ stats to be safe :) For example- average, range, mode, median, standard deviation, IQR (Interquartile Range), and line regression.
Best of luck!
The rubric also has a list of the required stats.
DFTBA!
Events 2019: Forensics and Fossils
Proud member of Teh Ento Cult.
:ugeek:
nejanimb
Exalted Member
Exalted Member
Posts: 343
Joined: November 14th, 2008, 5:17 am
Division: Grad
State: PA
Has thanked: 0
Been thanked: 0

Re: Experimental Design B/C

Post by nejanimb »

I had the pleasure of supervising ED at the MIT invite this last weekend, so thought I'd give a few pieces of unsolicited advice now that I've read ~70 labs from this year (including from a bunch of top teams):

- The Statement of Problem should, generally, take the form of "How does [IV] affect [DV]?" It makes it so much easier on your scorer though if it is immediately clear from the very beginning what you tested. If the supervisor has to skip ahead to the procedure or the charts to understand what the lab was about, it's a lot harder to give a good score. Expanding on that statement of problem, either by adding additional detail in words or showing an extra diagram (if it's less staightforward what you did) can be a worthwhile investment, even though there isn't a point allocated there on the rubric.

- In your hypothesis, don't just say "We hypothesize that as [IV] increases, [DV] will [increase/decrease]." Also describe the shape of the relationship -- as in, follow that up with "We expect this to follow a [direct/indirect] [linear/exponential/power/quadratic/inverse/etc.] relationship."

- Add more precision, especially in the first sections. Next to zero teams got full credit for operationally defining their variables, and similarly for "enough information to repeat procedure". It needs to be entirely unambiguous what you're doing. For instance, variable definition shouldn't just be "the angle of incidence" nor even "the angle of incidence as measured in degrees with a protractor," it should be "the angle of incidence, operationally defined as the angle between the center of a mirror's face and a theoretical straight line to the emitter of a laser pointer as measured in degrees by a protractor with its measurement origin matched over the center of the mirror."

- For your controlled variables, don't just throw in the kitchen sink. List variables you're intentionally not varying for the purpose of isolating the effect you're hoping to see. Using the same strength laser when you're evaluating what it looks like after refracting through water is clearly important, the air pressure of the room is less so.

- The "all data discussed and interpreted" piece of the analysis section does not, in my mind, mean you should be re-writing the data tables in words. Writing a lot very quickly to just try to snag those points that way makes me think less of the lab overall -- it's better to pick the important points and talk about important findings from throughout the experiment (i.e. maybe list the averages of all of the trials in order, or maybe note that the first trial at each level was the highest, etc.). The commentary on trends should almost always include an interpretation of the trend line or z-tests you show between treatments (i.e. "the slope of M means that for every unit increase in X, we expect an M change in Y") and a discussion of summary statistics (i.e. the R^2 of G means that G% of the variation in Y can be explained by this regression on X).

- For statistics, don't give the overall means/modes -- those are usually meaningless! Give the means across trials for each level.

Of course, continuing to nail the basics and also be creative in your labs is important, but those were areas where I saw even the top teams fall down. Hope that helps!
Harriton '10, UVA '14
Event Supervisor in MA (prev. VA and NorCal)
maxxxxx
Member
Member
Posts: 284
Joined: November 30th, 2015, 8:11 am
Division: Grad
State: PA
Has thanked: 0
Been thanked: 1 time

Re: Experimental Design B/C

Post by maxxxxx »

nejanimb wrote:The commentary on trends should almost always include an interpretation of the trend line or z-tests you show between treatments (i.e. "the slope of M means that for every unit increase in X, we expect an M change in Y") and a discussion of summary statistics (i.e. the R^2 of G means that G% of the variation in Y can be explained by this regression on X).
Just wondering, what do you recommend as the 6th statistic to replace line of best fit or regression line when you don't have a scatterplot or a line graph(i.e. bar graphs, histograms, etc.)?
Lower Merion Class Of 2017
hamingil
Member
Member
Posts: 2
Joined: January 8th, 2017, 10:26 am
Has thanked: 0
Been thanked: 0

Re: Experimental Design B/C

Post by hamingil »

On the scoring explanation document on the science olympiad website, it says that when doing the statistical section, include a graph with a line of best fit. Why is there a need for another graph when you're required to make one before moving onto statistics?
nejanimb
Exalted Member
Exalted Member
Posts: 343
Joined: November 14th, 2008, 5:17 am
Division: Grad
State: PA
Has thanked: 0
Been thanked: 0

Re: Experimental Design B/C

Post by nejanimb »

maxxxxx wrote:
nejanimb wrote:The commentary on trends should almost always include an interpretation of the trend line or z-tests you show between treatments (i.e. "the slope of M means that for every unit increase in X, we expect an M change in Y") and a discussion of summary statistics (i.e. the R^2 of G means that G% of the variation in Y can be explained by this regression on X).
Just wondering, what do you recommend as the 6th statistic to replace line of best fit or regression line when you don't have a scatterplot or a line graph(i.e. bar graphs, histograms, etc.)?
In general, it's easier to do this event if you pick an experiment two quantitative variables so that you can use a scatterplot (and pick that kind of relationship). That said, if you don't, my recommendation would be to run ANOVA (or, if that's too complex, you can build Z tests).

For that and a lot more about experimental design, you can refer to these textbooks online. I've always thought that not enough folks doing this event really dive in on how much awesome depth there is in statistical experimental design.

1) The gold standard: http://www.guilanstat.ir/wp-content/upl ... y-2012.pdf
2) One that's a bit more accessible: http://users.stat.umn.edu/~gary/book/fcdae.pdf
Harriton '10, UVA '14
Event Supervisor in MA (prev. VA and NorCal)
smrt1337
Member
Member
Posts: 16
Joined: January 26th, 2017, 7:31 pm
Division: B
State: WI
Has thanked: 0
Been thanked: 0

Re: Experimental Design B/C

Post by smrt1337 »

Im new to xpd, so could someone tell me if its ok to abbreviate? (eg. DV, IV, SOC)

Thanks!
Sasstiel
Member
Member
Posts: 32
Joined: January 11th, 2017, 12:08 pm
Division: B
State: IL
Has thanked: 0
Been thanked: 0

Re: Experimental Design B/C

Post by Sasstiel »

smrt1337 wrote:Im new to xpd, so could someone tell me if its ok to abbreviate? (eg. DV, IV, SOC)

Thanks!
Just in case, I would write it out, but if you're really stretched for time, it would probably be okay
DFTBA!
Exp Des, Wind, Food, and Bottle
Marlen1356
Member
Member
Posts: 2
Joined: February 5th, 2017, 12:17 pm
Has thanked: 0
Been thanked: 0

Re: Experimental Design B/C

Post by Marlen1356 »

My group is new to this since it was just added to our region. Should we select a leader? And how should you do the graphs? I know the experiments are physics based so is there anything else i need to know? Thanks :P
Locked

Return to “2017 Lab Events”

Who is online

Users browsing this forum: No registered users and 2 guests