How To Properly Conduct An Experiment

[drobbyb]

I was thinking to myself today about possibly conducting an experiment...only I have forgotten exactly how. I remember a bit about having a hypothesis. Where I go a little foggy is exactly how I should document my steps and findings so that they would be scientifically solid. If someone could possibly enlighten me I would be very grateful!

 

[rdd1952]

It really depends on what your experiment is and how complicated it is. For instance, if your theory was that a large marble dropped from 10 foot in height would shatter or crack when it hits a concrete surface, then all you need is the marble, a ladder and a concrete floor. On the other hand, if your theory is that bacteria-in-a-bottle products don't work, then you need lots of data on cycling a tank without the products before you can even try testing them.

 

[Bignose]

In addition to what rdd said, it also depends on what your goal for this is. If you just want to provide evidence for an idea you had, all you need to do it write down your method of testing and what the results were. If you are aiming to publish a paper in a scientific journal, you probably need to be more careful -- keep a data log, sign it everyday, etc. Even then, that level of "security" is only truly important if you think someone else is going to try to take credit for or steal your work.

In terms of a general overview, you might want to read through [URL="http/en.wikipedia.org/wiki/Scientific_method"]http/en.wikipedia.org/wiki/Scientific_method[/URL]

 

[drobbyb]

Thank you very much. I'm still in the planning stages with nothing set in concrete yet. I'm not looking to have anything published, so I'm not worried about having anyone steal it. It's more to satisfy my own curiosity and possibly share with others. Regardless of the result, I want to do it right.

 

[OldMan47]

I find the hardest thing to do when designing an experiment, just for my own use, is to anticipate all of the things that may affect the outcome so that I can control those to be the same in my experimental setup and my control setup. It can be a truly daunting task to be able to say with some confidence that this factor affects that one.

 

[rdd1952]

I find the hardest thing to do when designing an experiment, just for my own use, is to anticipate all of the things that may affect the outcome so that I can control those to be the same in my experimental setup and my control setup.

I agree 100%. I tried to do some cycling experiments a while back and when I started thinking through all the things that needed to be controled or the same for each tank, it was almost overwhelming. Even the initial fill water for instance. Since the thought is that the initial bacteria are present in the water, my thought was that if I filled one tank first and then the other, that one could possibly contain more bacteria than the other thus affecting the cycling time. Depending on what the experiment is, there are so many things that can have an effect on the outcome thus making the experiment less than perfect.

 

[andywg]

I tried to do some cycling experiments a while back and when I started thinking through all the things that needed to be controled or the same for each tank, it was almost overwhelming. Even the initial fill water for instance. Since the thought is that the initial bacteria are present in the water, my thought was that if I filled one tank first and then the other, that one could possibly contain more bacteria than the other thus affecting the cycling time.

Which is why any experiment should ideally be repeated a number of times (I recall three being a minimum) to ensure that the potential for a freak result is removed.

In an ideal situation you will isolate the point you are conducting the experiment on and document your methods and findings in such a way that someone else can come and repeatedly obtain similar results repeating what you did.

 

[OldMan47]

I must agree completely with that Andy.

 

[Spiceweasel69]

Hi

From experience, 5 times would be better. Omit the highest and lowest results and average the 3 remaining to get what is still a rough result. Remember to control all variables that you can. Even little variables such as conducting an experiment at 24 degrees and 26 degrees can skew a result if temp is a sensitive variable.

 

[rdd1952]

And in our hobby there are a lot of variables: water temperature, ambient air temperature, type of filter, flow rate of filter, type of media, location/number/wattage of heaters (filters too), and on and on.

 

[drobbyb]

Very good info. It would seem that it's not as cut and dry as I thought it would be.

 

[TigerIssey]

Brief outline of a write up can help your experimental design

AIM - What you want to learn

HYPOTHESIS - What you think will happen ie nothing = H0 = null hypothesis, something = H1 = something will happen
H0 and H1 only really needed for advanced statistcs

ABSTRACT - Written after completing everything - a brief summary

INTRODUCTION - Includes background research and explains what others have tested

METHODS AND MATERIALS - List all equipment reagents etc used step by step procedure safety discusses three types of variables ie Controlled (what you will keep constant) Independent (what you want to change) dependent (what changes as a result of changing the independent variable)

RESULTS OBSERVATIONS - Graphs etc (not raw data though this goes in appendix at back and can be referred to) Statistics

DISCUSSION - Compare your findings to others, suggest improvements to the experiment etc)

CONCLUSION - State what you have found and whether or not you proved or disproved hypothesis

REFERENCES - Any work you have quoted done by others

APPENDIX - Raw data safety sheets etc.

I know you don't want to publish in a journal however following these basic sections can break the experiment down into smaller chunks that you can deal with seperately and can help greatly in experimental design.

Hope it helps

Tina