Brent Cornell

The scientific method is a systematic process of experimentation resulting in the collection of empirical data 

The experimental methodology can be summarised according to three main objectives:

• Formulate a hypothesis that describes a relationship between an independent and dependent variable
• Design a methodology that controls extraneous variables and allows for collection of reliable data
• Collect and analysis the data to draw conclusions that either support or reject the original hypothesis

Hypotheses

A hypothesis is a testable prediction outlining the relationship between an independent variable and a dependent variable

• The independent variable is the condition that is manipulated (selected or changed) by the experimenter 
• The dependent variable is the variable that the experimenter measures, which is assumed to be affected by the independent variable

For an experiment to be considered valid, only the independent variable should be influencing the dependent variable

• Validity describes the extent to which a conclusion appropriately reflects the generated data (i.e. correctly addresses the hypothesis)
• Data (and associated conclusions) are not considered valid if extraneous variables are uncontrolled (could affect the dependent variable)

When forming conclusions, it is important to remember that a correlation does not automatically indicate causation

• In other words, just because two variables share a relationship, does not mean that one variable one directly impacts the other

Methodology

In order to determine the impact of an independent variable on a dependent variable, it is beneficial to have a baseline for comparison 

• A control group is not subjected to the independent variable and provides a baseline for comparing results from all experimental groups

To confidently determine the relationship between the two scientific variables, all other extraneous variables must remain unchanged 

• These conditions which are kept constant are called control variables (and allow for valid conclusions to be drawn)

Replication

Experimental procedures should be replicated multiple times in order to ensure that the collected data is consistent and reliable 

• Repeatability refers to the collection of multiple data sets via concurrent trials (i.e. same equipment, same experimenter, same time)
• Reproducibility refers to the collection of multiple data sets via disparate trials (i.e. different equipment or experimenter or times)

Reproducibility is preferred but not always achievable in the context of a scientific investigation (due to time constraints)

Data Collection

Experimental data can be either qualitative or quantitative 

• Data that is qualitative is non-numerical and presented as written observations (hence is usually more subjective in its interpretation)
• Data that is quantitative is numerical and can be represented by a variety of graphs (usually more objective in its interpretation) 

Bar charts are used when the data is categorical or discrete, while line charts and scattergrams are used when data is continuous 

• When graphing, the independent variable is shown on the horizontal axis and the dependent variable is shown on the vertical axis 
• Error bars can be included to show variability between the different trials (larger error bars indicate greater variability between trials)

Experimental Errors

Errors describe the imperfections in scientific measurements (i.e. difference between the recorded data and the ’true’ value)

• The true value is the value, or range of values, that would be found if the quantity could theoretically be measured perfectly

Two types of measurement error should be considered when evaluating the quality of data: systematic errors and random errors 

• The level of doubt associated with a measurement value may be represented as an uncertainty (e.g. ±0.05 units) 
• The larger the measurement uncertainty, the less exact the recorded value is likely to be (i.e. smaller = better) 

Random Errors

Random errors cause unpredictable variations in the measurement process and result in a spread of data 

• Random errors cannot be avoided and the displacement of data values is not consistent 
• Random errors can be limited by keeping all extraneous conditions constant and by taking repeated measurements (then averaging)

Random errors affect the precision of a measurement 

• Precision refers to the degree of similarity between multiple measurement values
• Error bars can represent precision by showing degree of spread around a mean

Systematic Errors

Systematic errors cause predictable variations in measurements, whereby the displacement of data is consistent and directional 

• Sources of systematic error include faulty calibrations of a measurement device or faulty readings by a user (e.g. parallax error) 
• As a systematic error produces a consistent variation, the impact cannot be reduced by repeating the measurement

Systematic errors affect the accuracy of a measurement 

• Accuracy refers to the degree of similarity between a measurement value and a reference (‘true’) value 
• The determination of accuracy requires the prior establishment of a correct measurement standard 
• Improving data quantitation (by making data less qualitative) will improve the accuracy of a given reading 

Personal Errors

Personal errors are not measurement errors but instead refer to mistakes or miscalculations made by an experimenter 

• Personal errors can be eliminated by performing the experiment again (but correctly) and should not be included in data analysis 

Outliers are any results that deviate by a significant margin from all other values and are typically reflective of personal errors 

• Outliers should be included in the raw data but can be removed from processing (but the reason for doing so should be explained)