Research methods common mistakes

Common mistakes

• Confusing primary and secondary data

  Assuming that data collected from a previous study is primary data, or that data collected by the researcher is secondary data.

  Fix itClarify that primary data are those collected directly by the researcher for the current study, while secondary data are previously collected and published by others.

• Confusing mean with median

  Assuming the mean always represents the central tendency, even when the distribution is skewed.

  Fix itExplain that the mean is sensitive to extreme values and may not reflect the typical value in a skewed distribution; the median is often more appropriate in such cases.

• Confusing Range with Mean

  Treating the range as a measure of central tendency.

  Fix itRemember that the range only describes spread; the mean or median describe central tendency.

• Confusing correlation with causation

  Assuming that a significant correlation means one variable causes the other.

  Fix itClarify that correlation indicates association only; causation requires experimental manipulation or longitudinal evidence.

• Confusing histograms with bar charts

  Using a bar chart to display the distribution of a continuous variable.

  Fix itA histogram should be used for continuous data; a bar chart is appropriate for categorical data or group comparisons.

• Confusing the direction of skewness

  Assuming that a long right tail indicates a left‑skewed distribution.

  Fix itA long right tail is a positive (right) skew; the mean will be greater than the median. A long left tail indicates negative (left) skew.

• Confusing correlation with causation

  Assuming that a significant correlation means one variable causes the other to change.

  Fix itExplain that correlation only indicates association; causation requires experimental manipulation, control of confounds, and ruling out alternative explanations.

• Confusing nominal and ordinal variables

  Treating gender as an ordinal variable and applying rank‑based analyses.

  Fix itGender is a nominal variable; it should be analysed with methods appropriate for categorical data, such as chi‑square tests.

• Confusing coding with categorisation

  Treating coding as simply assigning labels without a clear, predefined coding scheme.

  Fix itBefore coding, develop a detailed coding frame that specifies each code’s definition, inclusion and exclusion criteria, and ensure it is exhaustive and mutually exclusive.

• Confusing statistical significance with practical significance

  Assuming that a statistically significant result automatically means the effect is large or important in real‑world terms.

  Fix itExplain that statistical significance relates to the likelihood of chance, while practical significance concerns the magnitude and real‑world impact of the effect. Researchers should report effect sizes and confidence intervals to assess practical importance.