I am new to experimental design and would like to perform the following experiment but I am not sure if my setup is correct or not so far. My idea is below.

Let's say you want to optimise the growth conditions for the small fish. You have been using a standard type of food that has worked reasonably well. But now, three new brands of fish food have become available and you want to conduct an experiment to test which food works best and whether it is worth changing to one of the new food types. Based on previous experiments, you reckon that you will need 10 replicates to have a good chance of finding meaningful differences. There are aquaria and tanks of different sizes available so you are not constrained by the equipment.

For this experiment, I assume there will three treatments (the three new brands of fish food). I will call them A, B and C. Since we are testing three different brands of fish food, I will replicate the brands of fish food three times so there will be nine tanks used in this experiment in total. I will label them from 1 to 9. Each tank will randomly receive different food brands of food every day.

But now, how will I find meaningful differences in the growth of the fish if I apply different treatments randomly on the tanks? Also, how many fish do you recommend I use for this experiment and how often should I monitor the growth of the fish? Or do you have another suggestion of how I could set up this experiment, help. Your help will be much appreciated. Thanks in advance.

  • $\begingroup$ If this is a homework question, please consider using the 'homework' tag, and consult the homework policy here (biology.stackexchange.com/help/homework). It will help people give more suitable answers. If you are actually designing this experiment in real life, it would help if you included information about the species and the purpose for which they are being raised. $\endgroup$ – MikeyC May 5 '20 at 21:40

This is a good experiment!

  1. 3 types of food: ok

  2. 9 tanks: ok

  3. Apply different treatment randomly - no! You are right - no way to tell with random. It is like comparing McDonalds to vegan diet - you have to stick with your diet to be able to tell a difference in the end. So Tank 1,2 and 3 get food A. Tank 4,5,6 get food B. Tank 7,8,9 get food C. Each tank gets only that food type.

  4. You need to add the same weight of food to each tank. More food = more weight. I am not sure how much is a reasonable amount; it depends on your fish. Get out what looks like a reasonable amount and weigh it. That is how much food each tank gets each day. You could weigh a bunch of quantities and set them aside - maybe in an old egg carton? That will make it easier to feed the fish each day.

  5. Have the same number of fish in each tank. 3? Ideally fish are same species and all about the same size to start because different size fish might not all gain weight at the same rate. I am picturing 11-cent goldfish. 3 per tank makes 9 / food type and 27 in all which is good.

  6. You can gauge growth by fish weight. When it is time to weigh the fish, fill a bowl with aquarium water, put it on the balance and zero the balance. Now you will get the weight of whatever you add to the cup because balance is ignoring weight of cup and water. Sweep a fish out of the tank into the bowl and weigh it. Then dump it back in and repeat with next fish.

If you can tell the fish apart each one can be weighed separately and I think should also be named with a name that starts with the letter of its tank because that is fun. For example TankB4 will contain Burr, Black Beauty and Buzz. Tank C7 will have Clyde, Caesar and Coco.

  • Keep track in your lab notebook of each fish and its weight
  • Take a photo of each one when you start and one when you are done. This will make a great poster.

    1. You can weigh fish every day or maybe every 3 days if you think handling them is too stressful.
      You can track weight for each fish, average weight for each tank, and average weight for all 9 fish (3 fish / tank x 3 tanks / food type) over time. You will graph each of these. You can use Excel or do by hand.

If a fish or 2 die your data will still be good.

Good luck! I hope you will come back to this idea and post an image of Aloysius, Angel and Aesop, the fish from tank A1.

  • 3
    $\begingroup$ It would still strengthen the design for treatments to be randomly assigned among the tanks, ideally in a way that is blinded to the person weighing the fish (you keep a key to which tank gets which treatment, decode after data collected). This helps guard againt measurement bias and other possible confunding variable, like where the tanks are located in the room, or the order in which they are fed. You'll also want way more than 3 fish per tank, depending on the stage of development they are starting at. Some fish have very low rates of survival to adulthood. $\endgroup$ – MikeyC May 6 '20 at 19:59

Per your prompt, there are three new brands of fish food, but...

You have been using a standard type of food that has worked reasonably well.

I suggest you include the "standard" fish food as a control, so that you will have four treatments in all. Besides that, the experimental framework suggested by Willk is spot-on.


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.