Analysing evidence and drawing conclusions

Overview

Difference between describing, analysing and interpreting evidence and difference between a conclusion and an evaluation.

What you need to know

Answer each one before you reveal it.

Questions from this subtopic

Each one with the model answer underneath.

1. 1

   What is the difference between describing, analysing and interpreting evidence?

   Model answer

   Describing says what the data show - the values, the trend, the highest and lowest, quoted with figures. Analysing breaks the pattern down and examines it: how strong the trend is, where and by how much cases depart from it, how one variable relates to another, what the statistics say about it. Interpreting explains it geographically, linking the pattern to processes, to theory and to the characteristics of the place. Answers that stop at description are the commonest weakness; the marks lie in moving from what to how much and then to why.

2. 2

   How do you describe a pattern shown on a graph or map properly?

   Model answer

   State the overall trend or distribution first in a single sentence, then quantify it with figures taken from the diagram - the values at each end, the size of the change, the highest and lowest. Then identify any departures from the general pattern and locate them. Naming places and quoting numbers is what separates a description that could apply to any data set from one that is evidence, and every claim should be supported by a figure that a reader could find on the diagram.

3. 3

   What is an anomaly, and how should it be dealt with?

   Model answer

   An anomaly is a value that departs markedly from the pattern the rest of the data follow - a site well off the line of best fit, a reading far outside the range of the others. It should be identified explicitly, its size and direction stated, and a geographical explanation sought, since anomalies usually reveal a local factor: a tributary entering just upstream, a car park generating footfall away from the centre, a resistant band of rock, or a measurement taken in changed conditions. It should not be quietly deleted, though a value shown to be a recording error can be excluded provided that is stated.

4. 4

   How do you link results back to the hypothesis?

   Model answer

   By restating the hypothesis and then saying, with figures and with the statistical result, whether the evidence supports it, supports it partly, or contradicts it. Partial support is the usual honest answer, and it should be specified - the predicted increase held over the first six sites but reversed beyond the confluence, or the relationship held for one transect but not the other. Where a statistical test was applied, the conclusion should reference the null hypothesis and say whether it was rejected and at what level.

5. 5

   What must a conclusion do?

   Model answer

   It must answer the original question or hypothesis directly, in a sentence that could be quoted on its own, and support that answer with the key evidence and the statistical result. It should say how far the answer extends - to which area, which conditions and which period - and acknowledge the degree of confidence justified by the data. It should not introduce new data, repeat the whole analysis, or drift into an evaluation of the methods, which belongs in its own section.

6. 6

   Why must a conclusion be supported by the data rather than by what you expected to find?

   Model answer

   Because the entire value of an investigation lies in the evidence being allowed to decide the outcome. Writing a conclusion that matches the textbook model when your own measurements contradict it means the fieldwork was decorative, and a reader checking the figures will find the mismatch. Where the results depart from expectation, that departure is the most interesting part of the enquiry and should be reported and explained, not concealed.

Other ways to work this

The same content, four other ways.

What is inside Levo

Six ways to work the same material.