Student no:

Module EPM 202

Assessed Assignment

Markers agreed grade: 2

Below are some specific comments about your work that you will hopefully find useful.

  • You started well with a clear explanation of the study question and the outcome and exposure.
  • While your categorisations of household size and education were valid, it is not appropriate to say that this makes the analysis easier. Better to give a more appropriate reason for categorising, for example to reduce the problems due to small numbers in some categories. Further, you should be careful of over-collapsing categories as this can vastly reduce the information available to the analysis and in some cases can result in a significant finding being missed (as was the case with your treatment of education as a binary variable).
  • Although age was left in 4 categories, you only presented the results of a trend test in the tables. It would have better to have the categories given in table 1, with a comment in the text regarding a test for trend, and also a test of linearity.
  • You correctly identified the missing data on indebtedness and correctly removed those individuals from the analysis. One thing to note in the future is whether the missing data could lead to a bias.
  • You adequately described the statistical methods used.
  • Table one should have also included the numbers in each category, rather than just the results. At no point in your report do we have enough detail regarding the numbers in groups.
  • Although your results correctly described significant associations, you should have mentioned the directions of said associations and could have commented on the non-significant findings.
  • You correctly identified an interaction of borderline significance between toilet access and reproductive history on mood. The next step would have to do to a stratified analysis to allow fuller interpretation of this potential interaction.
  • The discussion was lacking.You started well with the correct interpretation of the odds ratio, but much more was required in terms of interpretation. For example, topics that could have been discussed include potential biases, measurement error, residual confounding, missing data in terms of potential bias, reverse causality and the generalisability of the study. It is also a little strong to say that the study calls for a probable causal relationship, particularly as you state it is ‘plausible’ without offering any evidence to back that statement up.

In summary, although your report lacked attention to detail and, in particular, interpretation of results, you generally used a correct approach to the statistical analyses.