brewmaster15
02-17-2013, 05:25 PM
Hi all,
I'm not going to go over the huge topic of experimental Design. If anyone really wants to read up on, the web is full of resources but I warn you, the statistical element to is really a source for headaches!
You can still accomplish alot with just some simple and basic science and observation.
I suggest a good basic starting place is understanding the Scientific Method...
http://en.wikipedia.org/wiki/Scientific_method
This process allows you to really dig into a subject and evaluate it... Hows it work?...Its actually pretty straight forward and my outline here should help make it even more straightforward.
1) Formulate a question.
Example ...What happens if I feed my discus alot of food verses smaller amounts as they grow.
2) Hypothesis
Example..... The amount of food fed should affect the growth rates. This is established for many animal models and makes sense here as well.
3) Prediction
Example... The Discus Fed the most food will grow the biggest and fastest.
4) Test.... Set up an experiment to test the Hypothesis and prediction.
Example... Take a group of Fry from one pair, randomly split the fry into 2 groups.. Heres where it gets tricky... you could test frequency of feeding....2 groups, one gets fed 2 times a day, 1 gets fed 5 times a day , both groups get the exact same amount each feeding.... then you record the growth of each fish over time.
You may also want to test amounts fed... this time you set up 2 groups and each gets fed the same number of times, but you have group 1 get "X"grams of feed per feeding and and the other group gets 2 times "X" again you measure and record the growth over time.
Its also a great idea if you have a control group that is used a reference. It gets whats normally done. Sometimes when a control group is not possible, you would use published data for related experiments.
Heres an important aspect... When you test anything, you need to minimize the variables that could effect the results.. So in the above case, you need to treat both groups identical except for what you are testing...the food amounts or frequency of feeding. That means, same size tanks, same water parameter, same equipment, same kind of food, same lighting, same original group of fish. ...you get the point? You also have to minimize your own Biases.... Its important that the group is randomly chosen, You can't let your preconceived ideas influence you as you set up the groups.. Often in animal research they would take the weight of each animal and assign the animal a number. The groups are then set up so you have an equal distribution of animals (by weight) in each test group.
5) Analysis.
Example... You take the weights of the animals/fish above and plot them on a graph or average them per group. You can do the same for the physical lengths of the fish. This will give you hard numbers you use in representing the differences between the groups... Its also really helpful and in hard science, its necessary to run statistics on the results. This goes back to the original huge topic of experimental design... you need to be sure you have enough "N" numbers per group. An "N" number is how many fish in this case,per study group. The bigger the "N" number, the less effect individual variability will have on the results. You would not want to draw a conclusion based on an "N" of 1 fish per group, but an "N" of 7 per group may be good. Some things may require an "N" of 15,20,50..what ever..It depends on what you are testing.
At this point you have an idea of if your hypothesis is right...so logically you...
6) Report results
What good is the data and experiment if you don't share it.
7) Re-test and validate
Either you or someone else will often try to repeat what you have done....if its repeatable..its considered proven... if its not...well then its back to the drawing board.
I simplified this procedure quite a bit, and theres some hardfast methodology of how the Scientific Theory works....What I wrote here is more of a "Experimenting For Dummies Guide"
Hth,
al
Ps.. You don't need to do all of this to share your experiments and observations in the Laboratory section of the forum. But it would be really cool if you did, and alot harder for someone to refute what you are writing about!:D
I'm not going to go over the huge topic of experimental Design. If anyone really wants to read up on, the web is full of resources but I warn you, the statistical element to is really a source for headaches!
You can still accomplish alot with just some simple and basic science and observation.
I suggest a good basic starting place is understanding the Scientific Method...
http://en.wikipedia.org/wiki/Scientific_method
This process allows you to really dig into a subject and evaluate it... Hows it work?...Its actually pretty straight forward and my outline here should help make it even more straightforward.
1) Formulate a question.
Example ...What happens if I feed my discus alot of food verses smaller amounts as they grow.
2) Hypothesis
Example..... The amount of food fed should affect the growth rates. This is established for many animal models and makes sense here as well.
3) Prediction
Example... The Discus Fed the most food will grow the biggest and fastest.
4) Test.... Set up an experiment to test the Hypothesis and prediction.
Example... Take a group of Fry from one pair, randomly split the fry into 2 groups.. Heres where it gets tricky... you could test frequency of feeding....2 groups, one gets fed 2 times a day, 1 gets fed 5 times a day , both groups get the exact same amount each feeding.... then you record the growth of each fish over time.
You may also want to test amounts fed... this time you set up 2 groups and each gets fed the same number of times, but you have group 1 get "X"grams of feed per feeding and and the other group gets 2 times "X" again you measure and record the growth over time.
Its also a great idea if you have a control group that is used a reference. It gets whats normally done. Sometimes when a control group is not possible, you would use published data for related experiments.
Heres an important aspect... When you test anything, you need to minimize the variables that could effect the results.. So in the above case, you need to treat both groups identical except for what you are testing...the food amounts or frequency of feeding. That means, same size tanks, same water parameter, same equipment, same kind of food, same lighting, same original group of fish. ...you get the point? You also have to minimize your own Biases.... Its important that the group is randomly chosen, You can't let your preconceived ideas influence you as you set up the groups.. Often in animal research they would take the weight of each animal and assign the animal a number. The groups are then set up so you have an equal distribution of animals (by weight) in each test group.
5) Analysis.
Example... You take the weights of the animals/fish above and plot them on a graph or average them per group. You can do the same for the physical lengths of the fish. This will give you hard numbers you use in representing the differences between the groups... Its also really helpful and in hard science, its necessary to run statistics on the results. This goes back to the original huge topic of experimental design... you need to be sure you have enough "N" numbers per group. An "N" number is how many fish in this case,per study group. The bigger the "N" number, the less effect individual variability will have on the results. You would not want to draw a conclusion based on an "N" of 1 fish per group, but an "N" of 7 per group may be good. Some things may require an "N" of 15,20,50..what ever..It depends on what you are testing.
At this point you have an idea of if your hypothesis is right...so logically you...
6) Report results
What good is the data and experiment if you don't share it.
7) Re-test and validate
Either you or someone else will often try to repeat what you have done....if its repeatable..its considered proven... if its not...well then its back to the drawing board.
I simplified this procedure quite a bit, and theres some hardfast methodology of how the Scientific Theory works....What I wrote here is more of a "Experimenting For Dummies Guide"
Hth,
al
Ps.. You don't need to do all of this to share your experiments and observations in the Laboratory section of the forum. But it would be really cool if you did, and alot harder for someone to refute what you are writing about!:D