Experimental Design B/C

[letsallbefriends]

hi! i've got a quick question:
does our independent variable necessarily have to be connected, or show a relationship or trend? for example, could we have different brands of napkins as the independent variable? cuz the different brands of napkins have no correlation, they are simply different brands. so our graph would be a bar graph, and is this an okay hypothesis for that: if we drop water onto different brands of napkins, then the brand browny will be the most absorbent because browny is the thickest, therefore it will have the strength to hold the water.
is that still considered to show a relationship or trend? cuz typically, our hypotheses look like this (the format, not the content): if we increase the size of the napkin, then the absorpency of the napkins willl decrease because larger napkins are more dense and do not have enough space within them to retain water.
thx!

[aaplrox]

What does it mean on the rubric where it says "At least three levels of IV given?"
Also, I read the wiki, but I still don't really understand the analysis part in general. How exactly are you supposed to discuss your data?

[kittybug65]

What does it mean on the rubric where it says "At least three levels of IV given?"
Also, I read the wiki, but I still don't really understand the analysis part in general. How exactly are you supposed to discuss your data?

This means that You have to test with three diffferent strenghts 0f the IV. say you were testing Height of ball dropped vs. how high it bounced back. You would have to test at say, one foot, two feet and three feet. if you were doing a different experiment were you tested ball material vs. bounce height, you would have to use at least three differnt types of balls. hope this helps!

hi! i've got a quick question:
does our independent variable necessarily have to be connected, or show a relationship or trend? for example, could we have different brands of napkins as the independent variable? cuz the different brands of napkins have no correlation, they are simply different brands. so our graph would be a bar graph, and is this an okay hypothesis for that: if we drop water onto different brands of napkins, then the brand browny will be the most absorbent because browny is the thickest, therefore it will have the strength to hold the water.
is that still considered to show a relationship or trend? cuz typically, our hypotheses look like this (the format, not the content): if we increase the size of the napkin, then the absorpency of the napkins willl decrease because larger napkins are more dense and do not have enough space within them to retain water.
thx!

That would be an acceptable experiment. Just as the bounciness of different types of balls would be acceptable.

hi everybody! ok i really need help with this one topic. i am so stuck! any ideas?
1 buzzer
1 lemon
1 tomato
1 battery
1 penny
1 paper clip
1 ice bath
1 hot plate
1 thermometer
i was thinking since the only measurable dependent variable seems to be temperature or time, the experiment could be about heating the ice bath using three different heating methods - the hot plate, the battery, make a lemon battery. but then i thought those independent variables aren't really at 3 levels...so then i thought the battery, make a lemon batter, and make a tomato battery. but how in the world do you make a lemon battery and tomato battery using only the materials they give you? so then i was thinking maybe we can heat the ice bath using the hot plate and time how long it takes for the ice to melt using the different levels/powers of the hotplate (like you know how you can switch the hotplate to different levels) but then again, we only have one ice bath container probably, and heating the hot plate to differentlevels would take too much time
also, do you think they would give us an outlet to plug the hotplate into, or are we just given a plain hotplate?
so if you could give me some help with ideas, that would be fantastic!
thanks!

I got one! Does the surface tempurature of a fruit effect the inner tempurature? It's sorta obvious, but if this is what I came accross in competition than that is what I would go with. And yes, you would probably get an outlet for your hotplate !

[letsallbefriends]

What does it mean on the rubric where it says "At least three levels of IV given?"
Also, I read the wiki, but I still don't really understand the analysis part in general. How exactly are you supposed to discuss your data?

This means that You have to test with three diffferent strenghts 0f the IV. say you were testing Height of ball dropped vs. how high it bounced back. You would have to test at say, one foot, two feet and three feet. if you were doing a different experiment were you tested ball material vs. bounce height, you would have to use at least three differnt types of balls. hope this helps!

hi! i've got a quick question:
does our independent variable necessarily have to be connected, or show a relationship or trend? for example, could we have different brands of napkins as the independent variable? cuz the different brands of napkins have no correlation, they are simply different brands. so our graph would be a bar graph, and is this an okay hypothesis for that: if we drop water onto different brands of napkins, then the brand browny will be the most absorbent because browny is the thickest, therefore it will have the strength to hold the water.
is that still considered to show a relationship or trend? cuz typically, our hypotheses look like this (the format, not the content): if we increase the size of the napkin, then the absorpency of the napkins willl decrease because larger napkins are more dense and do not have enough space within them to retain water.
thx!

That would be an acceptable experiment. Just as the bounciness of different types of balls would be acceptable.

hi everybody! ok i really need help with this one topic. i am so stuck! any ideas?
1 buzzer
1 lemon
1 tomato
1 battery
1 penny
1 paper clip
1 ice bath
1 hot plate
1 thermometer
i was thinking since the only measurable dependent variable seems to be temperature or time, the experiment could be about heating the ice bath using three different heating methods - the hot plate, the battery, make a lemon battery. but then i thought those independent variables aren't really at 3 levels...so then i thought the battery, make a lemon batter, and make a tomato battery. but how in the world do you make a lemon battery and tomato battery using only the materials they give you? so then i was thinking maybe we can heat the ice bath using the hot plate and time how long it takes for the ice to melt using the different levels/powers of the hotplate (like you know how you can switch the hotplate to different levels) but then again, we only have one ice bath container probably, and heating the hot plate to differentlevels would take too much time
also, do you think they would give us an outlet to plug the hotplate into, or are we just given a plain hotplate?
so if you could give me some help with ideas, that would be fantastic!
thanks!

I got one! Does the surface tempurature of a fruit effect the inner tempurature? It's sorta obvious, but if this is what I came accross in competition than that is what I would go with. And yes, you would probably get an outlet for your hotplate !

Ahh...but what would the three levels of independent variables be? What exactly would you test that shows a correlation between variables? Thanks!!

[siciscio]

The three levels of the independent variable, since you are testing for temperature, would be the tomato or lemon on a cold area (ice) a hot area. (Hot plate) and some where in between ( heat the ice with the hot plate) this way you produce three different surface temperatures for the fruit. ( don't do room temperature, you need it for the SOC) as for the correlation I can't help you there since I don't do that part but I'd think you'd just puncture a hole into the fruit and measure the internal temperature If I'm reading this correctly. Hope it helps

[nejanimb]

How do you operationally define a varible?

This was the most common error I saw when I supervised this event last week, and not one team got any points for it.

To operationally define something is to explain what PRECISELY you mean when you identify the variable. So say you're doing the "height of ball being dropped vs. bounce height" experiment. Your IV is "the height of the ball when it is first dropped." You operationally define that as "the shortest distance, as measured with a meter stick in centimeters, between the surface of the ball and the floor." That has a low enough level of ambiguity, at least in my opinion. An operational definition like "the height of the ball above the floor in centimeters" leaves huge room for interpretation: do you measure from the top of the ball? From the center? From the bottom? What do you use to measure? Do you measure straight down? Sometimes it can be difficult to come up with a good operational definition, even when the intuitive version is very clear. An operational definition is not "the time it takes for something to drop," it's "the time, measured in seconds (to 0.01s) using a stopwatch, between when the object is released by the experimenter's fingers and when the outer surface of the object first comes into contact with the floor"

Other common problems I saw:

- Zero teams did regression analysis.
- Zero teams broke out their tables into multiple categories.
- Many teams confuse what should go in the "analysis" section and what should go in the "Conclusion"
- Many teams gave me a "mode" for the entire table of results. That is not a particularly meaningful statistic, since that's not the purpose of the mode. Really, the mode is almost never relevant for this event.
- Many teams were tricked into doing experiments with categorical variables, which almost always results in a worse lab.

[magicalforest]

Can you elaborate when you say "Many teams confuse what should go in the 'analysis' section and what should go in the 'Conclusion.'"

From our group's understanding,they seem to have similar information. In the Analysis section, you analyze the data and point out trends, while in the Conclusion you similarly have to discuss and analyze the data in order to support your hypothesis. One year I did both the Data Analysis and Conclusion sections, and I felt like I was recopying my data analysis into my conclusion with a few additional sentences.

Can you explain the differences between the sections regarding use of the data? Thanks.

[nejanimb]

Can you elaborate when you say "Many teams confuse what should go in the 'analysis' section and what should go in the 'Conclusion.'"

From our group's understanding,they seem to have similar information. In the Analysis section, you analyze the data and point out trends, while in the Conclusion you similarly have to discuss and analyze the data in order to support your hypothesis. One year I did both the Data Analysis and Conclusion sections, and I felt like I was recopying my data analysis into my conclusion with a few additional sentences.

Can you explain the differences between the sections regarding use of the data? Thanks.

For sure.

Analysis is a deep dive into the data. You mention outliers (either along the trend line or among the trials) and discuss why they might have occurred (reference to the Experimental Errors section is sometimes helpful here). You discuss overall shape of the data and comment qualitatively on the trend. Then you discuss your regression analysis and how you create a model for the data (in raw numbers and in context), and explain the significant of the intercept and the slope. If you did transformations, you explain how you did those. If you compared multiple models (linear and square root, for instance), you explain the difference in fit. You discuss the variation around the centers of each of your trials, and why certain ones might have higher variance than others, etc. Deep dive into data and the statistics. You do not, however, make any reference to your hypothesis or expectations. It is purely descriptive statistics.

The conclusion is very simple and often short. You first restate the hypothesis, for which you literally take the sheet you wrote at the beginning and copy it down. You then say we choose to reject or accept/fail to reject (depending on how you've structured). Explain why the data does/does not fit the hypothesis (if it doesn't, explain what sort of data or relationship would have fit the hypothesis and note that yours was different). You don't have to say a whole lot though - one or two sentences relating it to the expectations is all you need.

Does that make sense? The analysis is stuff you could plausibly come up with looking only at the data and not at the context of the experiment or knowing what the expected relationship was. The conclusions evaluates the hypothesis based on the data and analysis just done. The analysis is generally 3-4x as long as the conclusion.

[mnstrviola]

@nejanimb wow, I always thought you only had to put units to operationally define variables. Thanks!

For " other appropriate statistic", what did a lot of teams do for it/you guys do for it that works and doesn't involve a lot of math?

[nejanimb]

@nejanimb wow, I always thought you only had to put units to operationally define variables. Thanks!

I should say, please don't consider what I say the final word. Unless you're in Virginia, I won't be the one scoring your labs, so your milage may vary! If you are in VA though and I'm your supervisor, now you know what to do!

For " other appropriate statistic", what did a lot of teams do for it/you guys do for it that works and doesn't involve a lot of math?

So many options, so long as you label it. Coefficient of variation, correlation statistics, medians, transformed regression, adjusted means... lots of reasonably easy options. It will depend on the lab, often. If you do a categorical lab, the F statistic from your ANOVA test is a reasonable way to go; if you do something that ends up with significant outliers, you can use the median or the trimmed mean; if you have lots of repeated trials, you can use coefficient of variation; if you have many levels, you can use some smoothing techniques (moving average is the easiest). You should have someone who knows a little bit about statistics and they should have some intuition about what would be a good way to take the analysis. Of course, to further my above discussion, the Analysis section is the place where you should not only point to what your bonus statistic was, but justify why it was appropriate!