Welcome to Data Analysis for AS Physics!
Welcome to one of the most rewarding parts of your AS Physics course! In Paper 3 (AS 3: Practical Techniques and Data Analysis), you get to step into the shoes of a real research scientist. In science, gathering data in an experiment is only half the journey. The real magic happens during data analysis, where you turn raw numbers into meaningful discoveries, calculate physical constants, and evaluate how reliable your results actually are.
Don't worry if maths or graph work feels intimidating at first. We will break everything down into clear, manageable steps so you can tackle any data analysis question with complete confidence!
1. Handling Raw Data and Tables
Before you draw a single graph, you must record and present your data accurately. Examiners look for very specific standards in your tables of results.
Key Rules for Presenting Tables
• Column Headings: Every column heading must clearly state the quantity and the unit, separated by a solidus (forward slash) or brackets. For example: \(v / \text{m s}^{-1}\) or \(t \text{ (s)}\).
• Raw Data Consistency: All raw readings in a single column must be recorded to the same precision (the same number of decimal places), which reflects the resolution of the measuring instrument used. For instance, if you use a standard ruler with millimetre divisions, all lengths must be recorded to the nearest \(0.1\text{ cm}\) (e.g. \(12.0\text{ cm}\), \(12.5\text{ cm}\), \(13.0\text{ cm}\)). Never write \(12\text{ cm}\)!
• Calculated Values: Values that you calculate (such as averages, squares, or logarithms) should generally be recorded to the same number of significant figures (s.f.) as the raw data with the fewest significant figures, or at most one extra significant figure.
• Repeating Readings: Repeating an experiment at least 3 times helps you spot anomalies (outliers). When calculating the mean, ignore any obvious anomalies and divide by the number of valid trials remaining.
Analogy: Think of significant figures like measuring ingredients for a recipe. If your scale only weighs to the nearest gram, writing down that you have \(10.000\text{ g}\) of sugar doesn't make your measurement more precise—it just misrepresents your tool!
Quick Review: Tables
Key Takeaway: Always include Quantity / Unit in headings, keep decimal places consistent for raw data, and discard anomalies before calculating the mean.
2. Mastering Graph Skills
Graphs are the most powerful tool in experimental physics because they average out random errors and allow you to visualise relationships between variables.
Step-by-Step Guide to Perfect Graphs
1. Choosing the Axes:
• Independent Variable: The variable you change goes on the horizontal axis (\(x\)-axis).
• Dependent Variable: The variable you measure goes on the vertical axis (\(y\)-axis).
• Label both axes clearly with the quantity and unit (e.g. \(\text{Force } F / \text{N}\)).
2. Selecting the Scale:
• Choose scales that are easy to read, using sensible steps such as \(1\), \(2\), or \(5\) (and their multiples of \(10\), like \(0.1\), \(0.2\), \(0.5\), \(10\), \(20\), \(50\)).
• Avoid awkward scales like multiples of \(3\), \(7\), or \(6\), as they make plotting and reading points very difficult and often lead to errors.
• Your plotted points must occupy more than half of the grid in both the \(x\) and \(y\) directions.
3. Plotting the Points:
• Use a sharp pencil.
• Mark each point with a small, neat cross \(\times\) or a small dot with a circle around it \(\odot\). Each point must be accurate to within half a small square.
4. Drawing the Line of Best Fit:
• For linear relationships, use a clear ruler to draw a single, unbroken straight line.
• The line should follow the overall trend, having a balanced number of points evenly distributed above and below it.
• Do not simply connect the dots! If there is an obvious outlier, ignore it when positioning your line.
Calculating the Gradient
To find the gradient (\(m\)) of a straight-line graph:
\(m = \frac{\Delta y}{\Delta x} = \frac{y_2 - y_1}{x_2 - x_1}\)
• Draw a large triangle: The hypotenuse of your triangle should cover at least half of the drawn line.
• Read points from the line: Never use original data table points to calculate the gradient unless they lie directly on your line of best fit! Read coordinates \((x_1, y_1)\) and \((x_2, y_2)\) straight from the line.
• Include units: The unit of the gradient is simply \(\frac{\text{unit of } y}{\text{unit of } x}\). For example, if \(y\) is in \(\text{m}\) and \(x\) is in \(\text{s}\), the gradient unit is \(\text{m s}^{-1}\).
Finding the \(y\)-intercept
The \(y\)-intercept (\(c\)) is the point where the line crosses the vertical axis at \(x = 0\).
• Common Trap: If your horizontal axis does not start at zero (a "false origin"), you cannot simply read \(c\) off the vertical axis! Instead, substitute the gradient (\(m\)) and the coordinates of a known point \((x, y)\) on the line into the equation:
\(c = y - mx\)
Quick Review: Graphing
Key Takeaway: Ensure your graph fills at least half the page, use sensible scale steps (\(1, 2, 5\)), draw a large gradient triangle (\(\ge 50\%\) of line length), and remember that false origins require calculating \(c = y - mx\).
3. Linearising Non-Linear Equations (\(y = mx + c\))
In physics, many equations are not naturally linear. However, by comparing a theoretical formula to the general equation of a straight line, \(y = mx + c\), we can transform curved relationships into straight lines to find unknown constants.
The Matching Method
Follow these three steps:
1. Rearrange the physics equation to isolate the dependent variable (or a function of it) on the left-hand side.
2. Align it directly beneath \(y = mx + c\).
3. Identify what corresponds to \(y\), \(x\), \(m\) (gradient), and \(c\) (\(y\)-intercept).
Example 1: Simple Pendulum
The time period \(T\) of a pendulum of length \(l\) is given by:
\(T = 2\pi \sqrt{\frac{l}{g}}\)
Squaring both sides gives:
\(T^2 = \left(\frac{4\pi^2}{g}\right)l\)
Now compare this to \(y = mx + c\):
• Plot on \(y\)-axis: \(T^2\) (unit: \(\text{s}^2\))
• Plot on \(x\)-axis: \(l\) (unit: \(\text{m}\))
• Gradient \(m = \frac{4\pi^2}{g}\)
• \(y\)-intercept \(c = 0\) (the line passes through the origin)
To determine the acceleration of free fall \(g\) from your experimental gradient \(m\):
\(g = \frac{4\pi^2}{m}\)
Example 2: Internal Resistance of a Cell
The terminal potential difference \(V\) of a battery of EMF \(\mathcal{E}\) and internal resistance \(r\) delivering current \(I\) is:
\(V = \mathcal{E} - Ir \implies V = -rI + \mathcal{E}\)
Comparing this to \(y = mx + c\):
• Plot on \(y\)-axis: \(V\)
• Plot on \(x\)-axis: \(I\)
• Gradient \(m = -r\) (a negative slope gives the internal resistance)
• \(y\)-intercept \(c = \mathcal{E}\) (gives the EMF directly)
Quick Review: Linearising Equations
Key Takeaway: Rearrange the equation to match \(y = mx + c\). What you plot on the axes determines what the gradient and intercept represent physically.
4. Uncertainties and Errors
No measurement is 100% exact. Understanding the limits of your measurements is essential for evaluating experimental evidence.
Types of Errors
• Random Errors: Cause readings to scatter unpredictably above and below the true value. Caused by human reaction time, slight temperature fluctuations, or reading a scale from slightly different angles. How to reduce: Take repeated measurements and calculate a mean.
• Systematic Errors: Cause all readings to differ from the true value by a consistent amount in the same direction every time. Caused by poorly calibrated instruments, zero errors (e.g. a balance reading \(0.2\text{ g}\) when empty), or background interference. How to fix: Recalibrate instruments, check for zero errors, or adjust technique. Taking repeats does not eliminate systematic errors!
Quantifying Uncertainty
1. Reading Uncertainty of an Instrument:
• Analogue scale (e.g. ruler, liquid thermometer): Uncertainty is typically taken as \(\pm\) half the smallest division (or \(\pm\) the smallest division when aligning both ends, like a ruler).
• Digital instrument (e.g. digital stopwatch, digital multimeter): Uncertainty is \(\pm 1\) in the last displayed digit (e.g. a balance reading \(4.52\text{ g}\) has an uncertainty of \(\pm 0.01\text{ g}\)).
• Repeated readings: Uncertainty in the mean \(= \frac{\text{range}}{2} = \frac{x_{\text{max}} - x_{\text{min}}}{2}\).
2. Absolute, Fractional, and Percentage Uncertainty:
For a measured value \(x \pm \Delta x\):
• Absolute Uncertainty: \(\Delta x\) (carries the same unit as the measurement)
• Fractional Uncertainty: \(\frac{\Delta x}{x}\) (has no units)
• Percentage Uncertainty: \(\% \text{ uncertainty} = \left(\frac{\Delta x}{x}\right) \times 100\%\)
Example: A length is measured as \(L = (20.0 \pm 0.2)\text{ cm}\).
• Absolute uncertainty \(= 0.2\text{ cm}\)
• Fractional uncertainty \(= \frac{0.2}{20.0} = 0.01\)
• Percentage uncertainty \(= 0.01 \times 100\% = 1.0\%\)
Combining Uncertainties (Rules to Memorise)
When you combine measurements in calculations, their uncertainties propagate according to these rules:
Rule 1: Addition or Subtraction (\(y = a + b\) or \(y = a - b\))
Always ADD the ABSOLUTE uncertainties.
\(\Delta y = \Delta a + \Delta b\)
Example: If initial temperature \(T_1 = (20 \pm 1)^\circ\text{C}\) and final temperature \(T_2 = (65 \pm 1)^\circ\text{C}\):
Change in temperature \(\Delta T = 65 - 20 = 45^\circ\text{C}\)
Absolute uncertainty \(= 1 + 1 = 2^\circ\text{C}\)
Result \(= (45 \pm 2)^\circ\text{C}\)
Rule 2: Multiplication or Division (\(y = a \times b\) or \(y = \frac{a}{b}\))
Always ADD the PERCENTAGE (or fractional) uncertainties.
\(\% \text{ unc}(y) = \% \text{ unc}(a) + \% \text{ unc}(b)\)
Example: A voltage \(V = (10.0 \pm 0.5)\text{ V}\) (which is \(5\%\)) produces a current \(I = (2.0 \pm 0.1)\text{ A}\) (which is \(5\%\)).
Resistance \(R = \frac{V}{I} = \frac{10.0}{2.0} = 5.0\ \Omega\)
Percentage uncertainty in \(R = 5\% + 5\% = 10\%\)
Absolute uncertainty in \(R = 10\% \times 5.0\ \Omega = 0.5\ \Omega\)
Result: \(R = (5.0 \pm 0.5)\ \Omega\)
Rule 3: Powers (\(y = a^n\))
MULTIPLY the percentage uncertainty by the power \(n\).
\(\% \text{ unc}(y) = n \times [\% \text{ unc}(a)]\)
Example: If radius \(r = (3.0 \pm 0.1)\text{ cm}\) (uncertainty \(= \frac{0.1}{3.0} \times 100\% = 3.33\%\)):
Area \(A = \pi r^2\). Here \(n = 2\).
\(\% \text{ uncertainty in } A = 2 \times 3.33\% = 6.67\%\)
Memory Aid:
• Addition / Subtraction \(\rightarrow\) Add Absolute uncertainties.
• Product / Quotient / Power \(\rightarrow\) Add Percentage uncertainties (or multiply by power).
Quick Review: Uncertainties
Key Takeaway: Absolute uncertainties add for \(+\) and \(-\). Percentage uncertainties add for \(\times\) and \(\div\), and multiply by the exponent for powers.
5. Error Bars and Graphical Uncertainties
To show how uncertain each data point on a graph is, we draw error bars. These look like small "I" shapes extending above and below (or to the left and right of) a plotted point.
Drawing Error Bars
• A vertical error bar for a point \((x, y)\) with uncertainty \(\pm \Delta y\) extends from \(y - \Delta y\) up to \(y + \Delta y\).
• The total length of the error bar is \(2\Delta y\).
Finding Uncertainty in the Gradient Using Worst Acceptable Lines
To determine the uncertainty in your calculated gradient:
1. Draw the Line of Best Fit (LOBF) that passes evenly through the data points.
2. Draw a Worst Acceptable Line (WAL). This can be either the steepest possible line or the shallowest possible line that still passes through all the error bars.
3. Calculate the gradient of the best fit line (\(m_{\text{best}}\)) and the gradient of the worst acceptable line (\(m_{\text{worst}}\)).
4. Calculate the uncertainty in the gradient (\(\Delta m\)):
\(\Delta m = |m_{\text{best}} - m_{\text{worst}}|\)
(Alternatively, if you draw both steepest and shallowest lines: \(\Delta m = \frac{m_{\text{steepest}} - m_{\text{shallowest}}}{2}\)).
Finding Uncertainty in the \(y\)-Intercept
In the same way, compare the \(y\)-intercept of the best fit line (\(c_{\text{best}}\)) with that of the worst acceptable line (\(c_{\text{worst}}\)):
\(\Delta c = |c_{\text{best}} - c_{\text{worst}}|\)
Did You Know?
Professional astronomers used error bars and line-of-best-fit analysis on the brightness of distant Type Ia supernovae to discover that the expansion of the universe is actually accelerating—an achievement that won the 2011 Nobel Prize in Physics!
Quick Review: Graphical Uncertainties
Key Takeaway: The uncertainty in a gradient or intercept is the absolute difference between the Best Fit Line and the Worst Acceptable Line passing through your error bars.
6. Essential Exam Tips and Common Pitfalls
Keep this checklist handy whenever you practise past paper questions:
• Don't forget units: Always quote the correct units for gradients, intercepts, and calculated constants.
• Significant figures in final uncertainties: An absolute uncertainty is usually stated to 1 significant figure (e.g. \(\pm 0.2\text{ m s}^{-1}\), not \(\pm 0.245\text{ m s}^{-1}\)). The final value must then be quoted to the same decimal precision as the uncertainty (e.g. \(5.8 \pm 0.2\text{ m s}^{-1}\)).
• Check scale markings: Double check every grid reading carefully. The most common mark lost in practical papers is misreading coordinates from non-standard scales!
• Pencil for graphs: Always use a sharp pencil for plots, lines of best fit, and error bars so that adjustments can be made neatly.