News Categories

Calculating Total Weighted Scores for RFx Responses

Some questions and sections in an RFx survey or questionnaire are more important than others. This relative importance should be reflected when total scores are calculated for the vendors being assessed. PostRFP allows evaluators to set weights for both questions and sections, and can then calculate weighted totals using one of two possible scoring formulas.

In PostRFP, all questions are scored on the same scale (1-10 by default). Each question is then given a weighting which can be any positive whole number. By default, the weighting for a question is 1. Most questionnaires are organised into a structure of sections and sub sections, so PostRFP allows further weightings to be allocated to sections.

When a survey consists of sections and subsections (nested to any depth), with weights applied at both question and section level, the mathematics for deriving a total score for each vendor becomes surprisingly complex and error prone when calculated in a spreadsheet. Two alternative scoring formulas are supported by PostRFP.

For example, consider the following example RFP questionnaire for evaluating Hotels:

1 Location  (1)
    1.1 Distances From Key Locations (2)
    1.2 Mini bus collection (1)
 2 Services  (1)
    2.1 Swimming Pool? (1)
    2.2 How many bars? (1)
    2.3 Health Club? (1)
  3 Pricing  (2)
    3.1 Standard Room Rate (1)
    3.2 Deluxe Room Rates (1)
The figures in brackets show the weighting allocated to each section and question. These weightings are applied when calculating total scores for each Supplier's answered response. PostRFP supports two different formulas for calculating Total weighted scores:
  1. Arithmetic
  2. Normalised
The difference between these formulas is that, in Normalised scoring, changing the weight of a question will only change the contribution of that question to the total score for the parent section, not to the Total Score for the entire RFx questionnaire. By contrast, if you increase the question weighting in Arithmetic scoring this will have the effect of increasing the parent section's contribution to the Total Questionnaire Score.

Arithmetic Scoring Formula

This is the simplest formula. The total score for a respondent is found as follows:
  1. Calculate the question score X question weight for each answered question
  2. Sum the weighted question scores for each section, then multiply by the section weight
  3. Sum the weighted section scores to find the total score.
This approach is illustrated in the following screenshot.

Arithmetic Weighting Calculation

This is a simple way to calculate the totals. In summary, questions scores are multiplied by question weights and section weights, and then summed to produce a total score.

As described above, the crucial thing about Arithmetic scoring is that increasing the weighting for a question has the effect of increasing the contribution that the parent section makes to the total questionnaire score. In other words, the Section weighting does not alone describe the contribution of that section to the questionnaire as a whole. In order to separate the effect of question and section weighting we need to normalise the scores for each section and subsection

Normalised Scoring Formula

Normalising the scores converts the weighted score for each answer in a section to the percentage of the total possible score for that section. So the total possible score for a section is 100%. Therefore, with Normalised totals, Section weightings define and cap the contribution that a given section can make to the total questionnaire score.

Normalised Calculation

Switching Scoring Formulas

Which scoring formula to use is determined on a per Project basis. There is an option to switch between formulas under the "Manage Weightings" link in the questionnaire tree.

Alternative Scoring Algorithms

For organisations that wish to follow their own scoring calculation formula PostRFP provides the ability to download all scoring data into spreadsheets for offline calculation.

Comments