Our Impact Analysis (Driver Analysis) is a form of correlation analysis and it can get quite technical explaining how it is done. However, for those who'd like to understand a bit more (without too much math), here is an explanation using a simple real world example. Things to keep in mind:
- Scores on Culture Amp reports are shown as % Favorable (% of Agrees + % Strongly Agree)
- Correlation analysis uses the full scale to calculate the strength of the relationship
- Correlation analysis measures the consistency of any relationship between two variables to determine the 'strength' of correlation - 0 means zero relationship between them and 1 is the strongest relationship possible
FAQ 1.: "Enjoys drinking coffee" scored 33%, while "Wears sneakers" scored 0% and "Likes the color Red" scored 33%, how does "Wears sneakers" have a stronger relationship with "Enjoys drinking coffee"? A 1.: Correlation analysis uses the full scale to calculate the strength of the relationship, the 33%(coffee), 0% (red) and 33% (sneakers) are irrelevant. In this instance:
- Respondent A showed a strong pattern of: when they scored high on coffee, they responded lower on"Wears sneakers" and "liking the colour Red"
- Respondent C showed somewhat of a pattern of: when they scored low on coffee, they also responded low on ""Wears sneakers" and "liking the colour Red". Although "Wearing sneakers" and "liking the colour Red" were not rated lower, the important take out is that they weren't rated highly.
- However, the biggest difference comes from Respondent B where they rated 3 on coffee, then a 1 on "Wears sneakers" then a 5 on "Likes the color Red". The way that this respondent rated "Likes the color Red" weakens the relationship pattern between coffee and red.
Therefore the pattern of the way people rated to"Enjoys drinking coffee" and "Wears sneakers" showed a stronger relationship than the way that people rated "Enjoys drinking coffee" and "Likes the colour Red"
FAQ 2.: "Enjoys eating chocolate" (66%) scored a lot higher than "Enjoys eating cake" (33%), how can they both have the same driver strength (0.98, Very Strong)? A2.: Correlations look at the patterned relationship between two variables, again the scores aren't relevant. In this instance:
- Those that enjoy drinking coffee, are likely to also enjoy eating chocolate and eating cake (respondent A = 5, 5, 4)
- Likewise, those that don't enjoy drinking coffee' are likely to also not enjoy eating chocolate and cake (respondent C = 1, 2, 1)
- Similarly, if you score somewhere in between (respondent B = 3) on coffee, you are likely to score somewhere in between on chocolate and cake (respondent B = 4, 2)
The way that each respondent scored on the 3 questions were very similar, and the pattern of the relationship is very strong between "Enjoys drinking coffee" and "Enjoys eating chocolate" as well as between "Enjoys drinking coffee" and "Enjoys eating cake"
FAQ3.: "These two questions scored the same, but why is one a driver but not the other?" A3.: Correlations look at the patterned relationship between two variables, again the scores aren't relevant.
- "Likes the colour Red" scored 33% but the pattern strength is only 0.33 when correlated with "Likes to drink coffee"
- "Enjoys eating cake" also scored 33% but the pattern strength is really strong when correlated with "Likes to drink coffee" at 0.98
The driver analysis ignores the score of a question, and strictly looks at the pattern of responses, and as you can see, the pattern of how people responded to "Likes drinking coffee" and "Enjoys eating cake" are almost the same, where as "Likes the colour Red", despite scoring the same as "Enjoys eating cake" exhibits a totally different pattern of responses.
FAQ4.: "How do you determine the classification of the driver strength?" A4.: The strength of the relationship is scored between zero and one, where zero indicates no relationship and one indicates a perfect relationship.
- > .70 – Extreme
- >.50<.70 – Very Strong
- >.40<.50 – Strong
- >.30<.40 – Moderate
- >.20<.30 – Low
- <.20 n/a