Working Scientifically

A study funded by a sugary drinks company concludes that sugar does not contribute to obesity. The study used only 20 participants, all of whom were regular exercisers. Evaluate the objectivity of this study.

Model response: This study has multiple objectivity concerns. First, funding bias: the company profits from people believing sugar is harmless, creating a financial incentive for favourable results. Second, selection bias: using only regular exercisers is not representative of the general population — exercise may offset the effects of sugar, masking the true relationship. Third, the sample size is far too small to draw reliable conclusions about a complex relationship. Fourth, there may be confirmation bias in how the researchers interpreted their data, given the funder's expectations. For this study to be credible, it would need to be replicated by independent researchers with a larger, randomly selected sample, and subjected to rigorous peer review before publication.

A student investigates how temperature affects enzyme activity. At 30°C they measure the reaction rate as 2.4, 2.5, and 2.4 cm³/min. At 35°C they get 2.6, 2.7, and 2.5 cm³/min. They conclude that temperature significantly increases the rate. Evaluate this conclusion in terms of accuracy and precision.

Model response: The measurements at each temperature are precise (small spread: ±0.1 cm³/min). However, the difference between the two means (approximately 2.43 vs 2.60 cm³/min) is only 0.17 cm³/min, which is barely larger than the measurement uncertainty (±0.1). This means the apparent increase could be due to random measurement error rather than a genuine temperature effect. The conclusion is not well-supported because the precision of the measuring technique is insufficient to detect such a small difference reliably. To strengthen the conclusion, they would need either a more precise measurement method or a larger temperature difference to produce a clearer signal. They should also check for systematic error — for example, whether the thermometer is reading correctly.

A team publishes a study claiming that playing classical music to plants increases growth by 15%. Another team in a different country repeats the exact experiment and finds no significant difference. What does this tell us, and what should happen next?

Model response: The failure to reproduce the result is significant because reproducibility is a cornerstone of reliable science. There are several possibilities: the original result could be a false positive caused by uncontrolled variables, small sample size, or confirmation bias. Alternatively, the replication team may have used slightly different conditions that affected the outcome (different plant species, different music volume, different growing conditions). The next step would be for multiple independent teams to replicate the study using a standardised protocol with pre-registered methods. If the majority cannot reproduce the result, the original claim is likely not valid. This is how science self-corrects — through replication and peer scrutiny. A single study, no matter how well-designed, is never sufficient to establish a scientific fact.

Wegener proposed continental drift in 1912, but it was not accepted until the 1960s when it became plate tectonics theory. Why was there a 50-year delay?

Model response: Wegener had compelling evidence: the jigsaw fit of continents, matching fossil distributions across oceans, and similar rock formations on separated coastlines. However, he could not explain the mechanism — how continents could move through solid ocean floor. Without a mechanism, the scientific community rightly demanded more evidence. The theory was also resisted because Wegener was a meteorologist, not a geologist, and there was institutional bias against outsiders. The breakthrough came in the 1960s when new technology (sonar mapping of the ocean floor) revealed mid-ocean ridges, seafloor spreading, and magnetic stripe patterns that provided both the evidence and the mechanism. This demonstrates that scientific acceptance requires: sufficient evidence, a plausible mechanism, the right technology to gather evidence, and overcoming social biases within the scientific community. The delay was not a failure of science — it was science working as it should, demanding strong evidence before accepting a radical new idea.

Some scientists argue that the current peer review system is under strain because reviewers are unpaid, journals have publication biases, and negative results are rarely published. Evaluate these concerns and suggest how the system could be improved.

Model response: These concerns are valid. Unpaid reviewers may rush or decline, reducing review quality. Publication bias means journals prefer exciting positive results over important negative results, which distorts the literature — a drug that does not work may never be reported, leading other scientists to waste resources testing it again. The 'publish or perish' culture incentivises quantity over quality. Improvements could include: pre-registration (researchers publish their method before collecting data, so the results are published regardless of outcome), open peer review (reviewers identified, increasing accountability), registered reports (journals commit to publishing based on the question and method, before seeing results), and funding agencies recognising negative results as valuable contributions. The replication crisis in psychology shows what happens when these issues are not addressed. Science is self-correcting, but the correction is faster and more efficient when the peer review system is well-designed.

The UK government must decide whether to approve a new vaccine. Clinical trials show it is 95% effective but causes mild side effects in 1 in 100 people and serious side effects in 1 in 100,000. The disease it prevents kills 1 in 1,000 infected people. Using your understanding of risk, advise the government.

Model response: This is a risk-benefit analysis. Without the vaccine, the disease kills 1 in 1,000 infected people — a high-severity, moderate-likelihood risk. With the vaccine, serious side effects occur in 1 in 100,000 — a high-severity but very low-likelihood risk. The vaccine reduces the disease risk by 95%, so the benefit (preventing deaths) far outweighs the risk (very rare serious side effects). The mild side effects (1 in 100) are low-severity and acceptable. I would advise approval because the risk of the disease is orders of magnitude greater than the risk of the vaccine. However, informed consent is essential — people should know the risks. Ongoing monitoring (pharmacovigilance) should continue to detect any rare side effects not seen in trials. This illustrates that 'zero risk' is impossible — all decisions involve comparing risks, not eliminating them.

You read that antibiotic resistance in bacteria is increasing worldwide. Generate three different scientific questions arising from this observation and evaluate which would be most feasible for a school investigation.

Model response: Question 1: Does the concentration of antibiotic affect the size of the inhibition zone around a bacterial culture? (Fair test — feasible in school with standard aseptic technique.) Question 2: Is there a correlation between a country's antibiotic prescribing rate and the prevalence of resistant bacteria? (Pattern-seeking using secondary data — feasible as a research project.) Question 3: How quickly do bacteria develop resistance when exposed to sub-lethal antibiotic doses over multiple generations? (Long-term experiment — less feasible in school due to time, containment, and ethical constraints.) Question 1 is most feasible because it can be completed in one lesson using standard school equipment and follows CLEAPSS guidelines. Question 2 is feasible as a data analysis project using published WHO data. Question 3, while scientifically important, requires facilities and timescales beyond school resources.

Two students make different predictions about what will happen to the rate of photosynthesis as light intensity increases. Student A predicts the rate will increase indefinitely. Student B predicts the rate will increase then level off. Design an experiment to test which prediction is correct, and explain which you think is right and why.

Model response: Student B's prediction is more scientifically grounded. At low light intensities, light is the limiting factor, so increasing light increases the rate. But at high light intensities, another factor (such as CO₂ concentration or temperature) becomes limiting, so the rate plateaus. Student A's prediction ignores limiting factors. To test this, I would use an aquatic plant (Elodea) in sodium hydrogencarbonate solution, placing a lamp at different distances to vary light intensity. I would count oxygen bubbles per minute at each distance (at least 5 different intensities, 3 repeats each). If Student B is correct, a graph of rate vs light intensity will curve and plateau. If Student A is correct, it will be a straight line that continues upwards. I would control temperature and CO₂ concentration (using the same concentration of sodium hydrogencarbonate throughout). This experiment can distinguish between the two predictions because it tests a wide enough range of light intensities to detect the plateau.

A company claims their new insulating material keeps drinks hotter for longer than a standard mug. Design a rigorous experiment to test this claim, identifying all variables and potential confounders.

Model response: Independent variable: type of container (new insulating material vs standard ceramic mug). Dependent variable: temperature of the water after 30 minutes (measured with a digital thermometer, ±0.1°C resolution). Control variables: same volume of water (250 ml), same starting temperature (80°C, measured with the thermometer), same room temperature, same lid (or both without lids). Potential confounders: (1) Different surface areas — if the containers are different sizes, the one with larger surface area will cool faster regardless of insulation. I would choose containers of similar size and shape. (2) Evaporation — if one container has a wider opening, more heat is lost through evaporation. I would use identical lids. (3) Room air currents — a draught could cool one container faster. I would place both in the same location. I would repeat the test 5 times to ensure reliability and calculate means. To strengthen the evidence further, I would test at multiple time intervals (every 5 minutes for 60 minutes) to plot a cooling curve.

You want to investigate whether there is a relationship between air pollution levels and the diversity of lichen species on trees in your local area. Design an appropriate investigation.

Model response: This requires a pattern-seeking investigation rather than a fair test, because I cannot experimentally control air pollution levels. I would identify 10 sampling sites at increasing distances from a busy road (0m, 50m, 100m, 200m, 500m, etc.). At each site, I would select 3 trees of similar species, size, and aspect. On each tree, I would use a 10x10 cm quadrat on the north-facing side at 1.5m height and count the number of different lichen species present (species diversity). I would also record potential confounding variables: tree species, tree diameter, aspect, proximity to other pollution sources. I would use secondary data from local air quality monitoring stations to estimate pollution levels at each distance. Limitations: this is a correlational study — I can identify a pattern but cannot prove causation because other factors (e.g. microclimate, tree age) may vary with distance from the road. The sample size (30 trees) gives reasonable statistical power but may not capture all variation. I would present results as a scatter graph (distance from road vs lichen diversity) and describe the correlation.

You are asked to determine the vitamin C content of different fruit juices but the school only has basic equipment (measuring cylinders, pipettes, DCPIP solution, and test tubes). Describe how you would adapt a titration technique and evaluate its limitations.

Model response: I would use DCPIP (dichlorophenolindophenol) solution, which is blue but turns colourless when vitamin C (ascorbic acid) is added. Method: Place 1 cm³ of DCPIP in a test tube. Use a pipette to add fruit juice drop by drop, counting drops, until the DCPIP just turns colourless. The fewer drops needed, the higher the vitamin C concentration. Repeat 3 times for each juice and calculate the mean. Limitations: Without a burette, I am measuring in drops rather than cm³, which is less precise — drop sizes vary depending on technique. I could improve this by using a graduated pipette if available. The endpoint (when the colour disappears) is subjective — different people may judge it differently. A colorimeter would give an objective measurement. The DCPIP method also responds to other reducing agents in the juice, not just vitamin C, so it may overestimate the vitamin C content. Despite these limitations, this method gives useful comparative data if I am consistent in my technique.

You need to measure the thickness of a single sheet of paper, but your ruler only measures to the nearest millimetre. How would you obtain a more precise measurement?

Model response: A single sheet of paper is approximately 0.1 mm thick, which is below the resolution of a standard ruler (1 mm). To measure it, I would stack 100 sheets of the same paper, measure the total thickness of the stack with the ruler (giving approximately 10 mm), and then divide by 100. This gives a thickness per sheet of approximately 0.10 mm — a measurement 100 times more precise than directly measuring one sheet. The uncertainty of ±1 mm in the stack measurement becomes ±0.01 mm per sheet when divided by 100. This technique works because the random errors in the thickness of individual sheets average out over 100 sheets. I would also repeat with a different batch of 100 sheets to check reproducibility. This is an example of a general strategy: when a single measurement is too small for your instrument, measure many identical items together and divide.

A student concludes that copper is a better thermal conductor than aluminium because a copper rod melted wax further along its length after 5 minutes of heating. Critically evaluate whether this conclusion is justified by the method used.

Model response: The conclusion may not be valid because the method does not adequately control for confounding variables. First, the rods may have different cross-sectional areas or lengths — a thicker rod conducts more heat regardless of material. Second, the initial temperature of the rods must be the same. Third, 'distance wax melted' depends on how much wax was applied and how evenly — this is a crude measure. More critically, thermal conductivity depends on both the material and the geometry. To validly compare conductivity, the student would need rods of identical dimensions (same length, same diameter), would need to ensure the same amount of heat is delivered to each (using identical Bunsen burner settings), and would need to measure temperature rather than wax melting (using thermocouples at fixed distances along each rod). Even then, the conclusion that copper conducts better than aluminium contradicts the established data (aluminium has higher thermal conductivity per unit mass, though copper has higher thermal conductivity per unit volume). The student's method is too imprecise to distinguish between these materials reliably.

You want to investigate how plant species diversity changes with distance from a river. Design a sampling strategy and explain how you would ensure your data is valid and representative.

Model response: I would use a belt transect perpendicular to the river, because a transect captures the continuous change in conditions with distance. At regular intervals (every 2 metres from the riverbank to 30 metres inland), I would place a 0.5 m² quadrat and record: species present, percentage cover of each species, and soil moisture (using a moisture meter). I would repeat this using 3 parallel transects at least 20 metres apart to ensure my results are representative and not affected by local anomalies (a fallen tree, a path, etc.). For validity, I would control the transect direction (always perpendicular to the river), record at consistent distances, and identify species using a field guide. I would calculate a diversity index (e.g. Simpson's Index) at each distance to quantify diversity rather than just counting species. Sources of bias: seasonal variation (different species visible at different times), subjective percentage cover estimates, and the transect may cross micro-habitats that confound the distance variable. To address these, I would sample at the same time of year, have two people independently estimate cover, and record any local features that might explain anomalies.

A 1,200 kg car accelerates from rest to 30 m/s. Calculate the kinetic energy gained. If the car travels 450 m during this acceleration, calculate the average force applied. State any assumptions.

Model response: Step 1: Kinetic energy gained: Ek = ½mv² = ½ × 1200 × 30² = ½ × 1200 × 900 = 540,000 J = 540 kJ. Step 2: Work done equals energy transferred: W = F × d, so F = W ÷ d = 540,000 ÷ 450 = 1,200 N. This is the average net force. Assumptions: (1) I have assumed all the work done goes into kinetic energy, with no energy lost to friction or air resistance. In reality, resistance forces would mean the engine must produce more force than 1,200 N. (2) I have calculated the average force — the actual force may vary during the acceleration. (3) I have assumed a straight-line journey. A reality check: 1,200 N is roughly the force that a small car engine produces, which is physically reasonable.

You have data showing UK energy generation by source (coal, gas, nuclear, wind, solar) for the years 2000, 2010, and 2020. What is the best way to present this data and why?

Model response: I would use a stacked bar chart or a stacked area chart with years on the x-axis and percentage of total energy on the y-axis, with each source as a different colour. This format shows both the overall trend (how the total mix changes over time) and the individual contribution of each source. An alternative would be a multiple line graph showing each source as a separate line — this is better for comparing rates of change but does not show proportions as clearly. I would avoid a pie chart because, while it shows proportions at a single time point, comparing three pie charts side by side is cognitively difficult. A table of raw numbers would be included as a supplement for readers who want precise values, but the visual representation communicates the trends more effectively. I would include a key, appropriate labels, and a descriptive title.

A dataset shows the relationship between CO₂ concentration in the atmosphere (measured in ppm) and global average temperature from 1960 to 2020. The correlation coefficient is 0.95. Evaluate this data and discuss what it tells us about climate change.

Model response: A correlation coefficient of 0.95 indicates a very strong positive correlation between CO₂ concentration and global temperature. However, correlation alone does not prove causation. The evidence for causation comes from the known mechanism: CO₂ absorbs infrared radiation, which is well-established physics (demonstrated in laboratory experiments since the 1850s by Tyndall and Arrhenius). The correlation supports the mechanism, and the timing matches — CO₂ began rising sharply after the Industrial Revolution when fossil fuel burning increased. Potential confounders include solar activity and volcanic eruptions, but climate models that include only natural factors cannot explain the observed warming — only models that include human CO₂ emissions match the data. The strength of the climate change evidence comes from this combination: a strong correlation, a known physical mechanism, temporal alignment, and the inability of alternative explanations to account for the observations. However, the relationship is not perfectly linear — feedback loops (e.g. melting ice reducing albedo, permafrost releasing methane) mean small increases in CO₂ can trigger amplified warming.

A student finds that plants watered with 'music water' (water exposed to classical music for 24 hours) grow 2 cm taller after 3 weeks than plants watered with regular water. They conclude that music makes plants grow better. Evaluate this conclusion.

Model response: The conclusion is not well-supported by the evidence for several reasons. First, the difference (2 cm over 3 weeks) is small and may be within the range of natural variation between individual plants — without knowing the standard deviation and sample size, we cannot assess whether this difference is statistically significant. Second, there is no known scientific mechanism by which sound waves could alter the chemical composition of water. Third, there are multiple confounding variables: the 'music water' was stored separately for 24 hours, which could have changed its temperature, dissolved gas content, or allowed chlorine to dissipate — any of these could affect plant growth. A valid experiment would need: larger sample sizes (at least 20 plants per group), a control group with water stored for 24 hours without music, measurements of water properties (temperature, pH, dissolved oxygen), and statistical testing to determine whether the difference is significant. Without these controls, the most parsimonious explanation is that the difference is due to chance or an uncontrolled variable, not the music.

Two students disagree about why a can of fizzy drink feels cold when opened. Student A says 'the gas escaping takes heat with it.' Student B says 'the dissolved CO₂ coming out of solution is an endothermic process.' Evaluate both explanations using your scientific knowledge.

Model response: Both explanations have some validity, but they operate at different levels. Student A's explanation (gas carrying heat away) is partially correct: when CO₂ gas escapes, it carries some thermal energy with it, slightly cooling the can. However, this effect is small because the mass of gas is tiny compared to the liquid. Student B's explanation is more scientifically precise: dissolved CO₂ coming out of solution is endothermic — energy is absorbed from the surrounding liquid to break the intermolecular attractions between CO₂ molecules and water molecules, causing the liquid to cool. Additionally, the rapid expansion of gas from high pressure (inside the can) to low pressure (atmosphere) causes adiabatic cooling — the gas does work expanding against the atmosphere, reducing its internal energy. The dominant cooling effect is probably the combination of endothermic dissolution and adiabatic expansion, rather than just 'heat escaping with the gas.' This demonstrates that everyday phenomena often have multiple contributing mechanisms operating at different scales, and a complete explanation requires identifying and evaluating the relative importance of each.

A student measures the density of aluminium by finding mass (27.0 ± 0.1 g) and volume (10.0 ± 0.5 cm³). Calculate the density and evaluate whether the uncertainty in the result is small enough to distinguish aluminium (2.70 g/cm³) from iron (7.87 g/cm³) and magnesium (1.74 g/cm³).

Model response: Density = 27.0 ÷ 10.0 = 2.70 g/cm³. To estimate the uncertainty, I combine the percentage uncertainties: mass uncertainty = (0.1/27.0) × 100 = 0.37%, volume uncertainty = (0.5/10.0) × 100 = 5.0%. Total percentage uncertainty ≈ 0.37% + 5.0% = 5.37%. So density = 2.70 ± 0.14 g/cm³ (range: 2.56 to 2.84 g/cm³). This is easily distinguishable from iron (7.87) and magnesium (1.74) because neither falls within the uncertainty range. The volume measurement contributes the most uncertainty (5% vs 0.37%), so to improve precision, I should focus on measuring volume more accurately — perhaps by using a more precise measuring cylinder or water displacement with a burette instead. The mass measurement is already sufficiently precise. This analysis shows that improving the least precise measurement gives the greatest improvement in the final result.

Your class found that different soil types support different plant communities. Design a series of three linked investigations that would build on this finding, explaining how each builds on the previous one.

Model response: Investigation 1: Characterise the soil — measure pH, moisture content, nutrient levels, and particle size at each site. This establishes the specific abiotic factors that differ between soils, moving beyond the vague category of 'soil type'. Investigation 2: Controlled greenhouse experiment — grow the same plant species in each soil type under identical conditions (same light, water, temperature). This tests whether the soil properties directly affect plant growth, controlling for other field variables like aspect, shade, and grazing. Investigation 3: Manipulation experiment — take one soil type and systematically alter one factor (e.g. add lime to increase pH, add fertiliser to increase nutrients) and observe the effect on plant community composition. This moves from correlation (Investigation 1) through controlled comparison (Investigation 2) to causal manipulation (Investigation 3). Each investigation builds on the previous one: the first identifies patterns, the second tests whether the relationship is causal in controlled conditions, and the third identifies the specific mechanism. This mirrors how scientific research programmes develop — from observation to controlled experiment to mechanistic understanding.

The speed of light is 300,000,000 m/s and a hydrogen atom has a diameter of approximately 0.000000000106 m. Express both in standard form. Then explain why consistent units matter in the equation E = mc².

Model response: Speed of light: 3.0 × 10⁸ m/s. Hydrogen atom diameter: 1.06 × 10⁻¹⁰ m. Standard form makes very large and very small numbers manageable and reduces errors. In E = mc², the units must be consistent: mass in kg, speed in m/s. If mass is in kilograms and speed is in metres per second, energy comes out in joules: [kg] × [m/s]² = kg·m²/s² = J. If I accidentally used grams instead of kilograms, my energy value would be 1,000 times too small. Unit consistency is not just about convention — it is about getting the correct numerical answer. This is why scientists always convert to SI units before substituting into equations.

A 60 kg skydiver falls from rest. After falling 500 m (ignoring air resistance), what is their speed? Use Ep = mgh and Ek = ½mv². Take g = 10 N/kg.

Model response: By conservation of energy, the gravitational potential energy lost equals the kinetic energy gained: mgh = ½mv². The mass cancels: gh = ½v², so v² = 2gh = 2 × 10 × 500 = 10,000. Therefore v = √10,000 = 100 m/s. This is approximately 360 km/h or 224 mph — which is realistic for a skydiver in a streamlined position without air resistance. In reality, air resistance would mean their actual speed is lower, reaching a terminal velocity of approximately 55 m/s. The key insight is that mass cancels, meaning all objects fall at the same rate in a vacuum (as Galileo demonstrated). This derivation also shows that speed depends on the square root of distance — to double the speed, you need to fall four times as far.

Group A (n=10) tested plant growth with Fertiliser X and got a mean height of 15.2 cm (range 12-19 cm). Group B (n=10) used Fertiliser Y and got 16.8 cm (range 11-22 cm). A student concludes Fertiliser Y is better. Evaluate this conclusion.

Model response: The conclusion is not well-supported. The difference in means (1.6 cm) is much smaller than the ranges of both groups (7 cm for A, 11 cm for B). The ranges overlap substantially (12-19 for A, 11-22 for B), meaning many individual plants in Group A grew taller than many in Group B. With only 10 plants per group and such large variation, the 1.6 cm difference could easily be due to natural variation between individual plants rather than the fertiliser. To strengthen this conclusion, the student would need: (1) a much larger sample size (at least 30 per group) to reduce the effect of individual variation, (2) a statistical test (such as a t-test) to calculate the probability that the difference is due to chance, and (3) controlled conditions to ensure the only variable is the fertiliser. As it stands, the data is consistent with both fertilisers having the same effect — the apparent difference is not statistically significant.