Validity, reliability and accuracy 4.14 e) performing first-hand investigations evaluate and modify experimental procedures Prior experience and knowledge for strategy • Identify variables that are being kept constant and those being changed in a first-hand investigation • Variables must be identified and controlled for an investigation to be valid. • The independent variable is the factor that is purposely changed or manipulated by the experimenter. • The dependent variable is the factor that is measured by the experimenter to find the effect of the independent variable. Classroom activities In this activity, students recall explicit metalanguage related to variables then consider the importance of validity and reliability in scientific investigations. Learning metalanguage A valid investigation tests the hypothesis fairly. It is reliable in its design and accurate in the measurements made. In a valid investigation, all variables are kept constant (as much as possible) apart from those being tested. For example, in the investigation following: • the independent variable is the type of liquid used to water the sprouts • the dependent variable is the height of the sprouts • some of the variables held constant include: – the amount of liquid given to each set of sprouts – the environmental factors of light, temperature, wind or air movement – the medium for growth (cotton wool or soil) – the age of sprouts at the start of the investigation. Experimental design influences the reliability of experimental results. A reliable investigation typically requires: • using the most appropriate measuring instruments correctly • taking the mean (average) of multiple measurements or trials. A reliable investigation is highly repeatable and results are consistently demonstrated. Accurate results are repeatable and are close to the true value for measured or calculated quantities. Accurate measurements typically require: • using the most appropriate measuring instrument correctly, for example, accounting for zero error • recording measurements and performing calculations correctly using appropriate significant figures • using correct values for constants. Strong internal validity means that not only are there reliable measures of the independent and dependent variables but also a strong justification that causally links the independent variable with the dependent variable. At the same time, all the extraneous variables can be ruled out because they have been kept constant. Any alternative, often unanticipated, cause for changes in the dependent variable can be identified and explained. Thus, strong internal validity refers to the unambiguous assignment of cause to effect. Internal validity is about causal control. Importance of designing a valid investigation Students use a think-pair-share strategy to analyse and assess the reliability and accuracy of the investigation using the following worksheet. Students think about, and perhaps draft, responses alone, then discuss them in pairs. Once students have agreed on their responses, they add their responses to a black/white board for class discussion. Can you help Abbey design a valid investigation? Abbey wanted to grow bean sprouts which were larger than those available in the supermarket. She thought she would try changing the liquid used to water the bean sprouts. In thinking about what she could use, Abbey decided to test plain water, mineral water, lemonade, tea and a liquid fertiliser. 1 What is Abbey’s independent variable? .................................................................................................................................................... 2 What is Abbey’s dependent variable? .................................................................................................................................................... Accurate measurements can be made by using the most appropriate measuring equipment and using correct technique for reading the equipment (such as avoiding parallax error). 3 What equipment should Abbey use to ensure that she makes accurate measurements of her dependent variable? How should she use it? .................................................................................................................................................... .................................................................................................................................................... .................................................................................................................................................... A reliable investigation often includes averaging of results from the same investigation. This helps to produce results that are highly reproducible, which means that the same experiment will produce the same results each time it is repeated. 4 What else should Abbey include in her method to ensure that she produces reliable results? What should she average? .................................................................................................................................................... .................................................................................................................................................... .................................................................................................................................................... .................................................................................................................................................... A valid investigation fairly tests the hypothesis. It must also be accurate and reliable. 5 How would you know if Abbey’s design has led to a valid investigation? .................................................................................................................................................... .................................................................................................................................................... .................................................................................................................................................... .................................................................................................................................................... ....................................................................................................................................................