Notes · Environmental ScienceUK · A-Levels

Research methods

This chapter develops the practical and quantitative skills that run through the whole subject. It covers the planning of environmental investigations, the sampling techniques used in fieldwork, the monitoring of the physical environment and pollution, the description of data through averages and spread including standard deviation, and the choice and interpretation of statistical tests such as Spearman's rank correlation and the chi-squared test.

5 sections~18 min reading time3 competenciesLevel Standard 3 · Advanced 2

T·161616 / 16

Exam profile

AO1 · Describe methods of sampling, monitoring and statistical analysisAO2 · Apply sampling designs and select and carry out statistical tests on dataAO3 · Analyse and evaluate methods, reliability and validity, and interpret statistical results

Operators:describeexplaincalculateanalyseevaluatejustify

basic level

AS-Level expects you to plan simple investigations, use quadrats and transects, and calculate means, ranges and percentage changes.

higher level

The full A-Level requires the selection and application of statistical tests, the calculation of standard deviation, and the evaluation of reliability, validity and significance.

Depth

Text

Contents · 5 sections

Research methods

Planning an investigation: variables, hypotheses, reliability and validity#

StandardAQA 7447 3.7

The structure of a fair investigation

Fig. 1A fair test changes one variable, controls the rest, and repeats measurements to improve reliability.

A sound investigation begins with clear variables. The independent variable is the one deliberately changed or selected, the dependent variable is the one measured in response, and the control variables are the factors kept the same so that they do not confound the result. Identifying these correctly is the foundation of a fair test, because only if other factors are controlled can a change in the dependent variable be attributed to the independent variable, an application of the correlation-and-causation reasoning met in the climate chapter.

Investigations are framed by hypotheses. A hypothesis is a testable prediction, and in statistics it is usual to state a null hypothesis, which predicts no effect or no relationship (for example that there is no correlation between two variables), against which the data are tested. The aim of a statistical test is to decide whether the data give enough evidence to reject the null hypothesis; if they do, the alternative hypothesis, that there is an effect or relationship, is accepted. Stating the null hypothesis clearly is a routine but essential step.

The quality of the data is judged by reliability and validity, which must not be confused. Reliability is the consistency of a measurement, whether repeating it gives the same result; it is improved by taking repeated readings and enough samples and calculating a mean, which reduces the effect of random error and of anomalies. Validity is whether the investigation actually measures what it claims to, which depends on controlling variables and using an appropriate method; a reliable measurement of the wrong thing is still invalid.

Accuracy and precision are two further, distinct ideas about measurement. Accuracy is how close a measurement is to the true value, and precision is how close repeated measurements are to one another (and how finely the instrument can measure); a set of readings can be precise but inaccurate if there is a systematic error. Being able to plan an investigation with the right variables, a clear hypothesis, and attention to reliability, validity, accuracy and precision, and to evaluate these in a given method, is assessed throughout the written papers.

Worked example

Setting up an investigation

An investigation tests whether distance from a road affects lichen cover on trees. Identify the variables, state a null hypothesis, and suggest how to improve reliability.

Variables
Independent variable: distance from the road; dependent variable: lichen cover; control variables: tree species, aspect, height sampled and time of year.
Null hypothesis
There is no relationship between distance from the road and lichen cover on the trees.
Improve reliability
Take several quadrats at each distance and calculate a mean, and sample enough trees, to reduce the effect of random variation and anomalies.

Result: The independent variable is distance and the dependent lichen cover; the null hypothesis is no relationship; repeats and adequate sampling improve reliability.

Typical mistakes

Confusing reliability (consistency) with validity (measuring the right thing); a consistent measurement of the wrong variable is invalid.
Confusing accuracy (closeness to the true value) with precision (closeness of repeats); readings can be precise but inaccurate.

Active revision

A student investigates whether soil moisture affects the number of a plant species. Identify the variables and state a suitable null hypothesis.

Active recall

Recall the key points — then reveal.

Sources: AQA AS and A-level Environmental Science (7447) specification (AQA) · GCE AS and A level subject content (Department for Education)

Sampling: quadrats, transects and abundance measures#

StandardAQA 7447 3.7

A belt transect up a shore

Fig. 2A belt transect: quadrats placed at intervals along a line record how the community changes along an environmental gradient.

Because it is rarely possible to count every organism, ecologists sample a representative part of an area and scale up. How the samples are placed matters. Random sampling, placing quadrats at coordinates chosen with a random-number generator, avoids bias and is used to estimate the abundance of an evenly varying area. Systematic sampling places samples at regular intervals, for example along a transect, and is used to study how a community changes along a gradient. Stratified sampling divides an area into sub-areas and samples each in proportion, ensuring that every distinct habitat is represented.

Quadrats are frames of known area used to sample the organisms within them. In a frame quadrat the organisms are counted or their cover estimated; a point quadrat uses pins to record which species each pin touches, giving percentage cover. Abundance is expressed as density (number per unit area), frequency (the proportion of quadrats in which a species occurs), or percentage cover (the proportion of the ground a species covers), with cover often recorded on an abundance scale for plants that are hard to count. Enough quadrats must be taken for the estimate to be reliable, which can be judged by seeing when the running mean stops changing.

Transects are used to study how a community changes along an environmental gradient, such as up a rocky shore or away from a path. A line transect records the organisms touching a line; a belt transect records the organisms in a series of quadrats placed at intervals along the line, giving abundance as well as presence. Because the community changes systematically along the gradient, a transect is the right tool for studying zonation, the banding of species according to their tolerance of the changing conditions, as met in the living-environment chapter.

Choosing the right sampling method is itself examined: random sampling for an even area, a transect for a gradient, stratified sampling for a patchy area with distinct habitats. The design must also sample enough to be reliable and must avoid bias, and evaluating a sampling method, identifying a source of bias or unreliability and suggesting an improvement, is a common higher-level task that applies the planning principles of the previous section to fieldwork.

Species zonation along a transect

Fig. 3Abundance of two species along a shore transect (illustrative data): each is zoned according to its tolerance.

Worked example

Choosing and using a sampling method

You want to investigate how the community of seaweeds changes from the low to the high shore. Describe an appropriate method.

Choose the method
Because the community changes along a gradient, use a belt transect running from the low to the high shore.
Take the samples
Place quadrats at regular intervals along the transect line and record the percentage cover of each seaweed species in each quadrat.
Ensure reliability
Lay several parallel transects and average the results, and use enough quadrats, so the pattern is not due to chance variation.

Result: A belt transect with quadrats at intervals, repeated and averaged, is the appropriate method for studying zonation up a shore.

Typical mistakes

Using a transect for an even area, or random sampling to study a gradient; the method must match the question.
Placing quadrats non-randomly (for example where the plants look interesting), which introduces bias.

Active revision

Describe how you would sample to find whether the abundance of a plant changes with distance from a hedge, naming the method and how you would ensure reliability.

Active recall

Recall the key points — then reveal.

Sources: AQA AS and A-level Environmental Science (7447) specification (AQA) · GCE AS and A level subject content (Department for Education)

Monitoring the physical environment and pollution#

StandardAQA 7447 3.7

Chemical and biological monitoring

Fig. 4Chemical and biological monitoring are complementary: one is precise but instantaneous, the other integrates over time.

Alongside sampling organisms, environmental investigations measure the abiotic conditions and levels of pollution, so that biological patterns can be related to their causes. Abiotic factors are measured with appropriate instruments: temperature with thermometers or probes, light with a light meter, soil and water pH with a pH meter or indicator, dissolved oxygen with an oxygen meter, water flow with a flow meter, and humidity, wind and soil moisture with their own instruments. Measuring the abiotic factors at the same points as the organisms allows the two to be correlated, as in relating species zonation to exposure or pH.

Pollution is monitored both chemically and biologically. Chemical monitoring measures the concentration of a pollutant directly, for example the dissolved oxygen or biochemical oxygen demand of water, or the concentration of a gas in air; it gives a precise value but only for the instant of sampling. Biological monitoring uses indicator species, organisms whose presence or absence reflects the conditions, such as the stonefly nymphs and sludge worms met in the pollution chapter; it integrates conditions over time and reveals effects on living things, but is less precise. The two approaches are complementary, and using both gives a fuller picture.

Modern monitoring increasingly uses continuous and remote methods. Automatic data loggers record a factor continuously over long periods, capturing variation that occasional readings would miss; and remote sensing from satellites and aircraft monitors large areas for changes such as deforestation, ice extent, sea temperature and algal blooms, at scales impossible on the ground. These methods provide the long-term, large-scale data on which the study of climate change and land-use change depends, linking fieldwork to the global datasets met earlier in the course.

Whatever the method, the same principles of reliability and validity apply: instruments must be calibrated and used correctly, enough readings taken, and the measurement matched to the question. Evaluating a monitoring method means asking whether it measures the right thing accurately and reliably, and recognising the trade-off between the precision of a chemical measurement and the integrated, biologically meaningful picture from indicator species, so that the choice of method suits the purpose.

Worked example

Choosing a monitoring approach

A conservation group wants to know whether a river is recovering from an old pollution problem over a whole year. Recommend a monitoring approach and justify it.

Biological monitoring
Survey indicator species such as mayfly and stonefly nymphs, whose presence reflects good oxygen levels integrated over time, showing whether conditions support sensitive life.
Chemical monitoring
Also measure dissolved oxygen and BOD at intervals, ideally with a data logger, to give precise values and capture variation through the year.
Justify combining them
Together they show both the exact conditions and their effect on living things over time, giving a fuller and more reliable picture than either alone.

Result: Combining indicator-species surveys with logged chemical measurements best shows recovery over a year, as the two approaches are complementary.

Typical mistakes

Assuming a single chemical reading represents conditions over time; it captures only the instant of sampling.
Overlooking indicator species, which reveal the effect of pollution on living things and integrate conditions over time.

Active revision

Explain why using both chemical measurements and indicator species gives a better assessment of river pollution than either alone.

Active recall

Recall the key points — then reveal.

Sources: AQA AS and A-level Environmental Science (7447) specification (AQA) · GCE AS and A level subject content (Department for Education)

Describing data: averages, spread and standard deviation#

AdvancedAQA 7447 3.7

Same mean, different spread

Fig. 5Two samples with the same mean but different standard deviations (illustrative): the average alone does not describe the data.

Once data are collected they are summarised. A measure of central tendency describes the typical value: the mean is the sum of the values divided by their number and uses all the data but is affected by extreme values; the median is the middle value when the data are ordered and is unaffected by extremes; and the mode is the most common value. Choosing the right average matters, for example the median is better than the mean for skewed data with a few very large values.

A measure of spread describes how variable the data are around the average. The range, the difference between the largest and smallest values, is simple but is set by the two extremes alone. The standard deviation is a much better measure because it uses every value: it is essentially the average distance of the values from the mean, and a large standard deviation means the data are widely spread while a small one means they are clustered near the mean. Two samples can have the same mean but very different standard deviations, so quoting the spread as well as the average is essential.

The standard deviation is calculated with the formula

s = \sqrt{\dfrac{\sum (x - \bar{x})^2}{n-1}}

, where

x

is each value,

\bar{x}

is the mean and

n

is the number of values: find the mean, find how far each value is from it, square these deviations (so positive and negative do not cancel), add them, divide by one less than the number of values, and take the square root. Reporting a mean together with its standard deviation shows both the typical value and how reliable and variable the data are, and allows two samples to be compared properly.

Presenting data well is part of the analysis. The choice of table and graph should suit the data: a line graph for a continuous relationship, a bar chart for categories, a scatter graph to show the relationship between two variables, and a histogram for a frequency distribution. Percentage change and rate of change are used to compare and to summarise trends. Being able to calculate a mean and standard deviation, choose an appropriate graph, and interpret the spread as well as the average, is directly assessed and underpins the statistical testing that follows.

\bar{x} = \frac{\sum x}{n}

Mean

The sum of the values divided by the number of values; it uses all the data but is affected by extreme values.

s = \sqrt{\frac{\sum (x - \bar{x})^2}{n-1}}

Standard deviation

A measure of the spread of the data about the mean; a larger value means the data are more variable. Divide by $n-1$ for a sample.

Worked example

Calculating a standard deviation

Five soil samples have nitrate readings of 4, 6, 8, 10 and 12 mg per litre. Calculate the mean and the standard deviation.

Find the mean
$\bar{x} = \dfrac{4+6+8+10+12}{5} = \dfrac{40}{5} = 8$ mg per litre.
Find the squared deviations
Deviations from the mean are $-4, -2, 0, 2, 4$ ; their squares are $16, 4, 0, 4, 16$ , which sum to 40.
Divide by n-1 and take the root
$s = \sqrt{\dfrac{40}{5-1}} = \sqrt{10}$ .
$s = \sqrt{\frac{40}{4}} = \sqrt{10} \approx 3.16$
Interpret
The mean is 8 mg per litre with a standard deviation of about 3.16, so the readings are moderately spread about the mean.

Result: The mean is 8 mg per litre and the standard deviation is about 3.16 mg per litre.

Typical mistakes

Dividing by $n$ instead of $n-1$ when calculating the standard deviation of a sample.
Quoting only the mean and ignoring the spread; two samples with the same mean can differ greatly in variability.

Active revision

Calculate the mean and standard deviation of the values 5, 7, 9, 11, 13, and state what the standard deviation tells you.

Active recall

Recall the key points — then reveal.

Sources: AQA AS and A-level Environmental Science (7447) specification (AQA) · GCE AS and A level subject content (Department for Education)

Statistical tests and interpreting significance#

AdvancedAQA 7447 3.7

A correlation to be tested

Fig. 6A scatter graph suggests a positive correlation (illustrative data); Spearman's rank test decides whether it is significant.

A statistical test decides whether a pattern in data is likely to be real or could easily have arisen by chance. It works by assuming the null hypothesis (no effect or relationship), calculating a test statistic from the data, and comparing it with a critical value from a table for the sample size and a chosen significance level, conventionally

p = 0.05

(a 5% probability). If the result shows that a pattern as strong as the one observed would occur by chance less than 5% of the time, the null hypothesis is rejected and the result is called statistically significant; otherwise it is not, and the pattern may be due to chance.

The choice of test depends on the kind of question and data. Spearman's rank correlation coefficient tests whether two variables are correlated, for example whether species abundance changes with an abiotic factor; the chi-squared test compares observed frequencies (counts) with those expected under the null hypothesis, for example whether organisms are distributed evenly between habitats; and a test of difference between the means of two samples, such as the Mann-Whitney U test or the t-test, is used to compare two groups. Selecting the appropriate test for the data is itself examined, so it is worth learning what each test is for.

Spearman's rank correlation is calculated with

r_s = 1 - \dfrac{6 \sum d^2}{n(n^2 - 1)}

, where the two variables are each ranked,

d

is the difference between the ranks of each pair, and

n

is the number of pairs. The coefficient runs from

+1

(perfect positive correlation) through

0

(no correlation) to

-1

(perfect negative correlation). The calculated value is then compared with the critical value from a table: if it equals or exceeds the critical value, the correlation is significant and the null hypothesis of no correlation is rejected.

Interpreting the result correctly matters as much as the calculation. A significant correlation shows that two variables are associated but, as stressed in the climate chapter, does not by itself prove that one causes the other; a mechanism is needed for that. It is also important to remember that failing to reach significance does not prove there is no effect, only that the evidence is insufficient, and that the 5% level is a convention that balances the risks of wrongly claiming and wrongly missing an effect. Being able to select a test, apply it, compare with the critical value, and interpret the outcome cautiously is the culmination of the quantitative skills of the course.

r_s = 1 - \frac{6 \sum d^2}{n(n^2 - 1)}

Spearman's rank correlation

$d$ is the difference between the ranks of each pair and $n$ the number of pairs; $r_s$ ranges from $+1$ (perfect positive) through $0$ (none) to $-1$ (perfect negative).

\chi^2 = \sum \frac{(O - E)^2}{E}

Chi-squared

$O$ is each observed frequency and $E$ the expected frequency under the null hypothesis; a larger value means a greater departure from what was expected.

Worked example

Carrying out Spearman's rank correlation

For six sites, light intensity and plant cover are each ranked from 1 to 6. The differences in rank d for the pairs are 0, 0, -1, 1, 0, 0. Calculate Spearman's rank correlation coefficient and interpret it, given that the critical value at n = 6 (5% level) is 0.886.

Find the sum of d squared
Squaring the differences gives $0, 0, 1, 1, 0, 0$ , so $\sum d^2 = 2$ .
Apply the formula
$r_s = 1 - \dfrac{6 \times 2}{6(6^2 - 1)} = 1 - \dfrac{12}{6 \times 35} = 1 - \dfrac{12}{210}$ .
$r_s = 1 - \frac{12}{210} = 1 - 0.057 = 0.943$
Compare and interpret
The calculated $r_s = 0.943$ exceeds the critical value of 0.886, so the correlation is significant at the 5% level: the null hypothesis is rejected. There is a strong positive correlation, though this shows association, not proof of causation.

Result: r_s = 0.943, which exceeds the critical value of 0.886, so there is a significant strong positive correlation (association, not proven cause).

Typical mistakes

Treating a significant correlation as proof of causation; it shows association only, and a mechanism is needed for causation.
Concluding that a non-significant result proves there is no effect, rather than that the evidence is insufficient.

Active revision

State which statistical test you would use to decide whether species abundance is correlated with soil pH, and explain how you would interpret a significant result.

Active recall

Recall the key points — then reveal.

Sources: AQA AS and A-level Environmental Science (7447) specification (AQA) · GCE AS and A level subject content (Department for Education)

References & sources

Sources

AQA

AQA AS and A-level Environmental Science (7447) specification

Department for Education

GCE AS and A level subject content

Research methods

5 sections~18 min reading time3 competenciesLevel Standard 3 · Advanced 2

Research methods

Key points

Setting up an investigation

Key points

Choosing and using a sampling method

Key points

Choosing a monitoring approach

Key points

Calculating a standard deviation

Key points

Carrying out Spearman's rank correlation

Sources

Research methods

Key points

Setting up an investigation

Key points

Choosing and using a sampling method

Key points

Choosing a monitoring approach

Key points

Calculating a standard deviation

Key points

Carrying out Spearman's rank correlation

Sources