Science Buddies: "Ask an Expert"

My hypothesis is that people are usually balanced in their brain dominance until around age 15 when they become right, left, or balanced. I have sent four tests to around 100 people of all ages by email. I also tested some 6th-9th graders at their classroom. I am now compiling the results of the tests to see what brain dominant they are. My question is what is my control variable, independent variables, and dependent variables. I thought my control would be the tests, the independent would be the test population, and the dependent would be the dominant brain hemispehere. Any other ideas?? How could I make this a better project?

Hi Mauvemary,

Lets try and break this down some more because this is a fairly complex experiment and there are many aspects that could be considered variables and you might want to discuss this further in your paper. If you can break it down it might make you more confident of how you classified your variables in your posting and you can answer your own question.

So you mention that you think your test was a controlled variable, why is it controlled? - was the test the exact same questions for everybody or was the test different for different groups? did they all get the same amount of time to answer the test? did some people study for the test?

also, where did you get the test? is there back ground work that was done by another group about this test? what did they find? what was it about your background reading that made you come up with the hypothesis?

It sounds like your hypothesis involves ages and you know the ages of the individuals being tested. Are they all the same age or is this a variable?
what else might be different about the people taking the test? do you know what gender they are?

I think this is a great project, you just arent letting the judges see inside your head and all of the ideas you are having.

best of luck,
caroline

carolinethorn wrote:Hi Mauvemary,

Lets try and break this down some more because this is a fairly complex experiment and there are many aspects that could be considered variables and you might want to discuss this further in your paper. If you can break it down it might make you more confident of how you classified your variables in your posting and you can answer your own question.

So you mention that you think your test was a controlled variable, why is it controlled? - was the test the exact same questions for everybody or was the test different for different groups? did they all get the same amount of time to answer the test? did some people study for the test?

also, where did you get the test? is there back ground work that was done by another group about this test? what did they find? what was it about your background reading that made you come up with the hypothesis?

It sounds like your hypothesis involves ages and you know the ages of the individuals being tested. Are they all the same age or is this a variable?
what else might be different about the people taking the test? do you know what gender they are?

I think this is a great project, you just arent letting the judges see inside your head and all of the ideas you are having.

best of luck,
caroline

Thanks for your reply. So I conclude from this that the tests (which were all the same questions, all blind tests, no study was allowed, no conferring with other people, etc.) were the control. (Do you think the tests I gave to children in a classroom setting should be excluded from the project since they were in a different setting?) The variables were the different age people and different genders. Can we have more than one variable at a time? I am also having doubts about the problem this project should be solving. Do you have any suggestions? I know I can not have too broad a problem or a problem that I can not really prove. I'd rather not just have my problem be the ratio of right brained to left brained or male to females.

Doing experiments with humans always makes it hard to control variables, there is always more than one uncontrolled variable because people have such variable lives. A lot of the time when the study is designed the researcher tries to make sure they have the same numbers of people in each group, and within that the same number of each gender etc. But since you have already done the experiment you will need to just be aware of these possible limitations and try various different analyses and see if the results are different. Don't worry, this happens with real clinical trials too, an investigator may suspect that a variable they didnt control for, like smoking or something, may have had an impact on the results of the study so they go back and re-analyse the data.

I would describe your study design as a cohort study - you have a cohort of younger people, who you hypothesise will test differently to cohort 2, the cohort of over-15 aged people. I would not exclude the tests taken in a different environment yet.

Divide up your tests into the 2 cohorts and work out the percentage of left brainers in each cohort.

Do you see a difference in the percentage of left brain people between the 2 groups - is it significant? can you do a statistical test (such as a t-test or a chi square test) to prove this?
You can also do the statistical test to show if the percentage of males is significantly different between the two cohorts. This might give you an indication about whether the difference you see might be correlated with gender.
Then you could go back and divide out the results and just look at males, or just look at females, or exclude the tests taken in a different setting and see if your results change.

It does seem pretty complicated but if you have difficulties there are people here to help.
good luck,

caroline

Hi mauvemary,

I would echo everything that Caroline posted, and would also recommend including your data from the classroom setting - at least at this point. Were the classroom students ages completely different to those who you emailed to other people, was there overlap, or were they in the same range of ages? You may want to look at those cases where the ages were the same and the setting changed.

I would suggest that if they were the same age, and the only difference was that this was in a classroom setting versus an email, that you examine and compare the two relevant sets of results and see what you find. If there is an appreciable difference, maybe the classroom setting does have an influence - in which case, I would still compile and interpret the data, but not necessarily include it in your total scores. Alternatively, if there is no appreciable difference in the results, you might be able to conclude that the setting is immaterial - in which case you can definitely include the results in your overall conclusions.

This sounds like a very interesting study - and I, for one, would be very interested to hear your results! As Caroline said, though, there are an awful lot of variables involved, so just be sure to take your time and work through it methodically. Although statistics is not my area, I know there are plenty of people on this forum who will be able to help you interpret your results.

Good luck!
Geoff.

Hey, again
I am having problems trying to figure out how to insert my data into the chi square and T-tests and correctly interpret the data. I found these two sites that explain them, but am still having a hard time with it. I was wondering if anyone could help me out here? Thanks!
http://www.georgetown.edu/faculty/ballc ... i_tut.html
http://home.clara.net/sisa/t-thlp.htm

Hi mauvemary,

My statistics is a little rusty also but i think we can manage. Its nice that you found a website that will do the calculations for you. It might be good to see if you can find something that has a simpler explanation of what the tests measure.

Lets start with working on the chi square.

How were you planning to set up your chi square table? I looked at the first link you posted and would suggest you think of it like this...

Look at their table 1d.
For the rows instead of Male and Female you would have instead Left brained and Right brained.
For the columns you would have YoungFemale, YoungMale, Over15Female, Over15Male

then you would put in the number of people you saw in each category into the table.

For table 1e you need to figure out what your expected outcomes are, which are based on your hypothesis. Your hypothesis is that for the young group they are equal numbers of Right and Left brainers in each gender. So if you had 20 young females you would put 10 in the square for Right and 10 in the square for left.
What was your hypothesis for the over 15 age group?

-Caroline

Caroline.
Thanks for your reply. My only problem is that I didn't just group the people into left and right brained categories, but I also had balanced as a choice. Does that matter in using the Chi square and T-tests? My hypothesis for the children was not that there would be an equal amount of left brained and right brained, but that there would be more balanced than specific right or left brained children. MY hypothesis for over 15 was that they would have a larger amount of left or right brained dominant people than balanced people.

HI mauvemary,

You can still do the chi square table, you just need to group according to your hypothesis - so it would be right/left for one row and balenced for the other row.

-Caroline

I'm now ready to do my charts and graphs. Can someone tell me what kind of graph to do--line, bar, etc.? What specifically should be on both sides to prove or disprove my hypothesis? I'm running short on time so would appreciate a quick response. Thanks for everything!!!

Maybe go for pie charts to show proportions of right, left and balenced in the older and younger groups. Then bar graphs for numbers of individuals in each breakdown category (eg young-male-left, young-male-right, young-male-balenced, young-female-left etc)

Do you also have some kind of chart that shows how score on test relates for whether you count someone as left, right or balenced?

good luck!
caroline

-Caroline
Ok, I thought I understood the Chi Square tests, but I have a question on the input and figuring out the expected results. So I was thinking I should put the Young Males,YOung Females, etc... In my rows and the Right/Left and Balanced in my columns instead of the opposite way around.....or not? I thought this just because I was thinking I needed to find the expected results of the amount of Right/Left in comparison to the Balanced. I was getting 15.5 for my Right/Left and 9.5 for my Balanced though...should I seperated the age groups into two seperate Chi Graphs? My expected results should be that the Balanced is greater right? So as you can see I am slightly confused. If you or anyone else could answer as soon as possible that would be great! Oh an another thing..do I need to do a T-test as well?

Hi mauvemary,

t-tests are a way to test significance between 2 or more sets of continuous data ie. data that is on a scale, eg. average test score, time to complete homework etc
chi square are a way to test significance between 2 or more sets of discreet data ie. data that can only be whole numbers, like number of people testing positive for rightbrain dominance

I think to do the test the way the website explains for the hypothesis that until age 15 "that there would be more balanced than specific right or left brained children" you need to have the rows be the values you observed.
so you need to have for the rows : number of balenced observed, number of balenced expected, number of right and left dominant observed, number of right and left dominant expected.
then the columns would be under15 and over 15.

then you could do additional tables where you add in gender as a column but you would need to think about how your expected might be different for testing the hypothesis that you would want to test. but lets try the simple one first and see what your chi squared and phi values come out to.

good luck,
-caroline

Ok, then for my expected results. Do I put any number in I want? It said something about taking the total of teh column times the total of the row divided by the total number. So if I did that for the first column, would that be finding the expected of balanced or right and left? OR am I all wrong?

Your expected results are based on your hypothesis. Its a really extreme version of what would happen if your hypothesis was right every single time. which can be true for some things in life but usually not for interesting science questions so your actual "observed" results will be different for the "expected".

So lets say you tested 30 people in total. 20 of them were under 15. 10 were over 15. You hypothesis is that under 15 they will be balenced. You then make up what you would expect to see if that rule was absolutely true all of the time.
So expected for under 15s would be 20 people in the balenced and then 0 in the right/left.
Your hypothesis is that over 15 will be left or right dominant. So assuming that is true you would expect 0 in the balenced and 10 in the right/left.

does that make sense?

the part about taking the total of the column times the total of the row divided by the total number is about determining how reliable your test is, or how many degrees of freedom you have.

-caroline

I'm sorry maybe i didn't explain the point of this test very well earlier. The way the test works is to assess how different or simlar your "observed" or real results are to what your hypothesis predicted or "exected" results.
If your results are similar enough to the hypothesis (using the phi score) then you can accept your hypothesis.

Thanks so much for your help Caroline! The School Science Fair was on the 16th and I qualified for regionals on March 2! I don't think I'll get much farther than that, but we shall see. Oh, and I was having trouble figuring out how to determine the results, but the judges told me that if I could explain it better at regionals it would be great. So that last comment did help. So if my phi value equals about 14% I'm assuming that it is close than since you can tell just by looking at the graphs that my hypothesis is correct.

mauvemary wrote:Thanks so much for your help Caroline! The School Science Fair was on the 16th and I qualified for regionals on March 2! I don't think I'll get much farther than that, but we shall see. Oh, and I was having trouble figuring out how to determine the results, but the judges told me that if I could explain it better at regionals it would be great. So that last comment did help. So if my phi value equals about 14% I'm assuming that it is close than since you can tell just by looking at the graphs that my hypothesis is correct.

Oh and I kind of didn't realize you had even posted that last comment until today because it was on teh beginning of a new page!

congratulations!
well done on getting to the regionals. don't be down on yourself about getting any further. you have a great project and should be proud of yourself. being interested and enthusiastic about your project and having confidence in yourself when you are talking about it with the judges can get you far.
so maybe you can find a math teacher at your school to talk about the statistical tests with so you get more confident about explaining it.

best wishes,
Caroline