go to the AUS-e-TUTE homepage
home game Join AUS-e-TUTE test contact
 

 

Experimental Design

Key Concepts

  • An experiment is conducted to test an hypothesis.

  • An experiment that is well designed has these features:

    1. A well defined aim (goal or objective).

    2. Results that can be reproduced (precision).

    3. Errors that can be analysed.

  • A well-defined objective is an hypothesis that can be disproved by experiment.

  • Results will most likely to be reproducible if

        (a) the independent and dependent variables are well chosen

        (b) all other variables are held constant

        (c) you repeat the same experiment a number of times until the results agree

  • Understanding the sources of error in an experiment BEFORE you conduct it can help you to

        (a) minimise the error inherent in the experiment

        (b) choose the most appropriate apparatus with which to conduct the experiment

Defining the Aim

The aim, goal or objective of an experiment is to test the validity of an hypothesis.
An hypothesis is a possible answer to a scientific question.
So, an experiment is conducted to decide whether the possible answer to the question is most likely to be "true" or not.

In science, it is very difficult, almost impossible, to "prove" an hypothesis, that is, it is almost impossible to say that the answer to the question is "true".
Example:

Scientific Question:What temperature does water boil at?
Possible Answer:Water boils at 100oC

If you issued a thousand city-dwellers all over the world with digital thermometers that read to the nearest 10oC, then they would all report back that water boils at 100oC.
But, will you have proved that the hypothesis is correct?
No. The best you could say is that the results of this experiment support the hypothesis.
If you preformed the same experiment using digital thermometers that read to the nearest 0.1oC you would find a huge range of temperatures reported for the boiling point of water.
Have you disproved the hypothesis?
Yes. The results of this experiment disprove the hypothesis.

An hypothesis is useful when it is a statement that can be disproved*.
Results of an experiment may support the hypothesis, but can rarely definitively prove the hypothesis to be true.
Results of an experiment that do not support the hypothesis can be said to disprove the hypothesis.

A well-defined aim, goal or objective, will be one which can be tested experimentally and can be disproved.

Achieving Precision

Controlling Variables

The first step in ensuring that the results of your experiment can be reproduced is to reduce the number of factors that can change during the experiment.
The different factors, or quantities, that can change are known as variables.
Before you even begin to think about how you will conduct the experiment, you will need to make a list of all the variables that you can think of that could affect the results of the experiment.
This is really hard to do without prejudice! You can NOT afford to make any value judgements about the "quality" of the variables at this stage, you need to list all the things that could change, even if you think it is unlikely or a bit silly.

Would you like to see what this means using an example?
Click this link to go to the complete tutorial if you are an AUS-e-TUTE member.
Not an AUS-e-TUTE Member?

  • Find out how an AUS-e-TUTE Membership can help you here.
  • Become an AUS-e-TUTE member here.

Once you think you have listed all the possible variables, you need to make 2 very important decisions#:


  • Which variable will you change deliberately? (called the independent variable)

  • Which variable will you measure as it responds to the change you make? (called the dependent variable)

Once you have made these decisions, you must then control ALL the other variables so that they do not change during the experiment. (called constant variables)

Would you like to see what this means using an example?
Click this link to go to the complete tutorial if you are an AUS-e-TUTE member.
Not an AUS-e-TUTE Member?

  • Find out how an AUS-e-TUTE Membership can help you here.
  • Become an AUS-e-TUTE member here.

Where you conduct the experiment, the time, the type of day etc, are important, they lay the foundations for the conditions under which the experiment is conducted.
Where you conduct your experiment can then be thought of as the height above sea level condition.
The time/type of day can be thought of the temperature condition, the humidity condition, and the atmospheric pressure condition.

Reliability of Your Results

How will you know if the results of your experiment are reliable?

There really is only one way to determine how reliable your results are. You will need to repeat the experiment several times and see if the results for each experiment agree with each other.**
How many times should you repeat your experiment?
A general rule of thumb for High School Chemistry students is to do one quick experiment to determine the approximate result of the experiment, then perform as many careful experiments as is required to get 3 results within 2% agreement.##

Estimating Errors

Before you begin an experiment, you should be aware of the types of errors that are possible so that you can reduce their impact on your results.
There are two types of errors; random and systematic.
You have no control over random errors such as fluctuations in the temperature of a bunsen burner flame caused by fluctuations in the gas supply to the burner for instance. Repeating the experiment in order to get results in close agreement with each other should compensate for these random errors.
You do have some control over systematic errors, so you need to think about the types of errors that are inherent in the experiment you are designing.

In the boiling water example, we have decided to use the same amount of water for all the experiments.
The most common ways of measuring an "amount" of substance in Chemistry is to weigh out a mass of the substance or to measure out a volume of the substance.
If we use a constant volume of water, say 50 mL, there will be different systematic errors depending on the apparatus we use to measure out the volume:

  • 50.00 mL transfer pipette might have a tolerance of ± 0.03 mL so with the bottom of the miniscus sitting on the mark of the pipette the volume could be as low as 50.00 - 0.03 = 49.07 mL or as high as 50.00 + 0.03 = 50.03 mL

  • 50.00 mL burette might have a tolerance of ± 0.06 mL so with the bottom of the meniscus sitting on the 50.00 mL mark, the volume in the burette could be as low as 50.00 - 0.06 = 49.04 mL or as high as 50.00 + 0.06 = 50.06 mL

  • If you used a measuring cylinder graduated in 1 mL divisions, you can only feel certain about the measurement to the nearest mL, but the amount of uncertainty in the measurement is half the limit of reading, that is half of 1 mL. So with the bottom of the meniscus sitting on the 50 mL mark, the volume could be as low as 49.5 mL or as high as 50.5 mL.

The transfer pipette would result in the smallest amount of systematic error, so it might be considered the best option for measuring out 50.00 mL of water.

But, the transfer pipette has been designed to deliver 50.00 ± 0.03 mL ONLY at 20oC. Many substances, like water, expand as they get warmer, so the volume of water increases as it gets warmer. If the temperature is not 20oC you would need to calibrate the transfer pipette in order to determine the volume of water being delivered at that temperature.

Instead of using 50.00 mL of water, we could instead use 50.00 g of water.
If we used an electronic balance that measures mass to the nearest 0.0001 g there is a large degree of uncertainty in the last decimal place. Is the mass really 49.9999 but was rounded up to 50.0000 g? Or was the mass 50.0001 g that was rounded down to 50.0000 g ? We don't know, so we might say that the uncertainty in the measurement is ±0.0001 g.
Compared with the systematic error involved in measuring the volume of water, the systematic error in measuring the mass of water appears to be lower. We might therefore decide to weigh out 50.0000 ±0.0001 g of water instead of measuring out a volume of 50.00 ±0.03 mL of water in order to reduce the systematic error.

Similarly, we would need to determine what sort of thermometer to use. One graduated in 1oC? One graduated in 0.1oC?
The limit of reading on a thermometer graduated in 1oC divisions is 0.5oC.
The limit of reading on a thermometer graduated in 0.1oC divisions is 0.05oC.
We might decide that it is important to have greater certainty in the temperature measurement and opt for the thermometer graduated in 0.1oC divisions in order to reduce systematic errors in the experiment.

Thinking about sources of error BEFORE you set up an experiment enables you to:

  1. reduce the systematic error in an experiment before it is conducted

  2. select the most appropriate apparatus with which to conduct the experiment

What would you like to do now?

I wou

  • read the complete tutorial (AUS-e-TUTE Members ONLY)

  • *Principle of falsification.

    #"One Variable At a Time" (OVAT) approach assumes the 2 variables are independent of each other.

    **We do not propose to discuss the statistical significance of results here.

    advertise on the AUS-e-TUTE website and newsletters
     
     

    Search this Site

    You can search this site using a key term or a concept to find tutorials, tests, exams and learning activities (games).
     

    Become an AUS-e-TUTE Member

     

    AUS-e-TUTE's Blog

     

    Subscribe to our Free Newsletter

    Email email us to
    subscribe to AUS-e-TUTE's free quarterly newsletter, AUS-e-NEWS.

    AUS-e-NEWS quarterly newsletter

    AUS-e-NEWS is emailed out in
    December, March, June, and September.

     

    Ask Chris, the Chemist, a Question

    The quickest way to find the definition of a term is to ask Chris, the AUS-e-TUTE Chemist.

    Chris can also send you to the relevant
    AUS-e-TUTE tutorial topic page.

     
     
     

    Share this Page

    Bookmark and Share

     
     

    © AUS-e-TUTE