7. Gauging Performance and Impact /

7.2 Analysing performance information


We regularly assemble and analyse information about our performance.

To meet the standard in full, you regularly look at the information you have collected and simplify the results to aid clarity. You ensure that information is analysed with appropriate rigour. As part of this, you look for patterns in the information that show important differences in the experience or effects of your work.

The information you collect will be in a raw form until you analyse it.

Analysis is simply the process of inspecting, comparing, and using data in order to come up with interesting insights that will enable you to make better decisions.

Before you start your analysis it’s best to ensure your raw data is as good as it can be – accurate, complete and consistent. If it’s not, then try to correct or remove inaccurate data first. You can find out more about data cleaning and preparation here.

Data then needs to be organized. Course 7 from the Impact Practice series examines ways in which you can analyse and present data in an orderly way.

Some of the most interesting analysis can be the most simple. Some basic descriptive statistics you can use include:

  • Count. Count the number of sources supplying information (e.g. the number of respondents to a survey).
  • Range. Assess the variation in results (e.g. the difference between the lowest and highest satisfaction rating).
  • Frequencies. Identify how frequently a result was evident (e.g. how often a behaviour or activity was observed).
  • Averages. Consider the mean, median, or mode (e.g. average improvement in self-reported skills).
  • Percentages. Work out what proportion of beneficiaries that experienced the change (e.g. the percentage of cases were signs of recovery were observed).
  • Ratio. Express and compare quantities relative to each other (e.g. the cost per outcome).

Try to ensure that the descriptive statistics you produce, describe the change in your agreed performance indicator.

Descriptive statistics help you describe what’s going on in the data but can only take you so far. Headline numbers may disguise important differences between groups.

This is where cross-tabulation can be helpful. For example, you could compare data to help you understand whether your work was more successful for men or women. Or you could use it to compare how people in different locations rated different activities. You can do this using pivot tables in everyday spreadsheet software.

While cross-tabulations can be helpful, it may require some further work to establish a statistically relevant correlation. This relies on advanced procedures such regression analysis to produce inferential statistics. These procedures are more common in specially commissioned impact studies, or as part of large-scale program evaluations.

Your choice between basic or more advanced analytical techniques is likely to depend on the quality of data you have, the resources you can commit to the task, and the standard of evidence required by others.