Getting your experiments right

... first time, every time

This node pertains primarily to biological and chemical experiments, although some points might apply to other fields as well.

Getting your experiments right is not a matter of good luck. It is also not a matter of being extremely hardworking, or having a lot of prior knowledge or experience, although these are needed to a certain extent. No, getting your experiments right is a matter of planning, finding out and preparation. These are the notes I've gathered after watching some efficient colleagues at work. Although they've never performed a certain experiment or technique before, they still managed to get the experiment right and never panicked or floundered during the process. And they accomplished all these without needing people to watch over them every step of the way. Here are their secrets.


1)     Plan well in advance, and plan in detail.

i)     All experiments should have an aim. What is it that you're trying to find? A well-stated aim is paramount to performing a good experiment. It should be specific and to the point. "Alcohol and female judgement" is a poorly stated aim. "To determine the amount of alcohol a human female would require before she cannot punch a string of 5 numbers accurately in 1 minute" is a well-stated one. The statement leads directly to the question of the variable in the experiment.

ii)     All experiments should have 2 variables: one to change, and one to be measured (thanks Jembeth!). What is it that you will change during the course of the experiment? Is it the time period? Is it the concentration of the chemical? If two or more variables are needed, a way of accurately comparing these variables is needed. All other factors should be kept constant. E.g. if both human males and females were to participate, they should be of a certain age group only, like 18-35 years, not 8-80 years. Also they should be judged healthy, and a large enough pool is needed to minimise individual differences. Even the time period given between the drink and the test should be the same. Only the amount of alcohol imbibed should change. And the variable to be measured is the imput of the sequence of 5 numbers given to the participants.

iii)     All experiments should also have a control. A control gives an idea of what the baseline is, or how a normal situation would affect your experiment. Always plan for the control. The control in this case would be normal healthy females who have not imbibed any alcohol.

iv)     All experiments should be planned in general first, before the details come in. Do not forget to plan for the details, however. It doesn't hurt to plan how much of a certain chemical you need. Sometimes a lab will run out of chemicals. Knowing how much you need for the length of the experiment prevents you from running around the lab bawling injustice because you did not have enough to meet your need. With regard to the example, calculate the concentration and amount of alcohol each individual needs based on body weight. And don't forget the equipment too. You might want to calculate the number of stopwatches and volunteers needed.

In order to make sure that you've left nothing out, it is always a good idea to ask someone to review it. That person should be someone with more experience and more knowledge than you. Be humble to accept corrections. Humility makes you strong, and strength crushes enemies.

Occasionally you can pull a protocol or an experiment outline off the internet. Be careful of using what you find on the internet directly. It is not always suited to your purposes. You might want to analyse this with someone as well.


2)     Finding out

i)     You need to find out whether you have enough chemicals. This is why you need to plan how much you need. If necessary, reserve them for your own use. Always have excess in case of emergency. You never know when you will need it. Order in new stock if you need. Oh by the way, try not to change to a new batch of chemicals in the middle of an experiment. If you need new stock, order enough for your experiment and use only that. Old and new stock may vary to give different results, like how new wines and old wines differ.

ii)     Find out about the materials and equipment you’re going to use. Its too late if you realize that the tube you’re using isn’t resistant to the strong acid you intend to use. Know how to operate the equipment. “Oops, the machine broke down while I was using it” is never a good statement to hear.

iii)     You need to find out some background information. Have other people done a similar experiment? What were their results? Reading this would be a great help, and would give you some idea of what you're doing. For biological experiments, particularly in molecular biology, it is important to find out why you're using that chemical. Especially if you took that protocol off the Internet. Knowing why you're using it, and what effect it has on the sample, and in what concentration it's needed, helps contribute to a successful experiment.

After doing some online or paper research, it might be necessary to edit your experiment outline or protocol a little.


3)     Preparation

Finding out isn't the last step. Many people confuse preparation with finding out. Finding out getting more information, but preparation involves action and coordination. You see, unless you're a super-duper dude in an ultra-rich lab, your experiment will surely involve some other people other than yourself and/or equipment that's shared.

i)     Preparation of chemicals. Yes, we're back to chemicals. You see, I work in a lab, and we often face the problem of "Oh there's no more X!" or "How do you prepare Y?". It doesn't help that these question occur 5 seconds before the person needs to use it. Now you know the volume you need, so prepare it and place it somewhere no one will dare mess around with. A sign like "Off-limits, this is MINE!" helps. Although it seems a little selfish, you have to let your colleagues know that your stuff is not for common usage, because you made it yourself and you need it. Of course, if you stole it from another person, that's another story.

ii)     Preparation of materials needed. We’re in the run-up to performing the experiment. Tubes need labeling. Label everything before a major experiment. Plan and coordinate what symbols you’re going to use to label your specimens, various chemicals, etc.

iii)     Preparation of equipment. In most research institutes, there are shared equipment. Usually its because these equipment are _really_ expensive. These machines either work on a first-come-first-served basis, or a booking basis. Make sure you know which system its under, and book way in advance if you need to. It saves a lot of hassle. ("My experiment is delayed by 1 hour because someone else is using the machine. Argh!")

iv)     Preparation of other personnel. This is really important if other people will take part in this experiment. You need to coordinate your movements and responsibilities really well. For my lab, we'll plan when I should start Y, when Y should be stopped, when Y gets handed to my colleague, what he should do and what I should do when he passes it back to me. Please don't take this step lightly, 'cos if someone forgets, or misunderstands, there goes your experiment. ~Poof!~ You might as well spend that time frying your fats with a pina colada in the Bahamas. It's more rewarding.


4)     Performing the experiment

Are you ready? Are you really ready? No I mean are you _really really really_ ready? Go back and check again! Make sure everything is in place, solutions made, equipment booked, and everyone knows what they should do.

Most people like to jump into this step without going through the previous two steps, and I'd like to emphasise that It's Wrong, It's All Wrong, This Is Not The Way You Get An Experiment Right First Time Every Time. You might get lucky, you might spend all your time screaming and begging and finally getting your way with the solutions and equipment and people, but that's not the way to do it, no sirree.

But if you've made sure everything is as it should be, perfect. Just go ahead and follow all the instructions that you've written and gone through. A perfectly planned and coordinated experiment might still meet some problems (hey, we're human) but it’s probably nothing you can't handle. Such an experiment will give great pleasure upon being completed.

My boss emphasises that during an experiment, no one should be asking any questions any more, and that all should know what they should be doing at what time. That is correct. Although she doesn't have an attractive personality (sic), she does have a very scientific and methodical brain, and therefore what she says needs to be taken into account.

Although the first 3 steps seem to take a long time compared to the experiment, it is imperative that the planning stage doesn't drag too long, and that it's completed in a reasonable amount of time. Work fast, work smart and all the best.


5)     In summary

My boss also said something else worthy of note. She said that a good scientist makes a good cook. Although I don't cook (and thus have no personal experience), I believe this is absolutely true. See, cooking and performing an experiment are quite similar, though sometimes the sequence of events differs. If I decided to cook a sun-dried tomato risotto I'd have to get my hands on a recipe, plan and make sure I have all the ingredients needed for the number of people I would serve. Knowing how each ingredient functions in the dish enables me to substitute other ingredients if I run out of them, or at least I should know what the overall result should be. Before I start cooking, I need to chop some onions and mince the garlic, in order to prepare for the actual cooking. This is how experiments should be run as well.



Sources:
My large eyes - the better to see with.
My huge ears - the better to hear with.
My puny brain - almost competent enough to compute, assess and evaluate all the data that my large eyes and my huge ears provide. The points above are just the bare essentials.


Jembeth says Don't forget that every experiment has 2 variables: The one you're going to change, and the one you're measuring. Other things: Always do a review of the literature to find out what safety stuff is necessary (fume hood? dust mask? respirator? gloves?) and write down everything you did so it can be reproduced exactly by yourself, or someone else.