Correlation of Leading & Lagging Indicators


I am working on a organizational performance dashboard for my company. I have identified lagging and leading indicators in five domains of the organization that are critical to ongoing success of the firm. These indicators are essentially organizational metrics. I want to determine the strength and the reliability of the relationship between the lagging and leading indicators.

I think that calculating the correlation coefficient of pairs of lagging and leading indicators will answer my question about the strength of the relationship. However, I am not sure how to determine how reliable the relationship (or how to decide how large my sample size must be).

Any guidance you can offer is greatly appreciated. Bonus question: I am using SAS for this analysis and if you happen to know which functions I should use, I would appreciate the advice.


Less is more. Stay pure. Stay poor.
This could be a terminology issue (for me), but are you saying that you have an outcome (DV) and predictors of success. Some predictors are leading, they increase success - while others are lags, decreasing success? How do you define success (Success Y/N [binary], or is it continuous [95%])?

Please provide more details, initially it seems like it might be logistic regression that you need (proc logistic), but more information is needed.


Less is more. Stay pure. Stay poor.
Also, are leading and lags the same thing, but on other ends of a spectrum. So a continuous variable can be defined as leading with a high value and lag with a low value?
Leading and lagging are really just labels I am assigning to metrics. It is not meant to imply that one is value is higher or lower. For example, a lagging metric is salary costs as a % of total expenses. Its lagging because it indicates how money was spent in the past. A leading indicator of how money will be spent in the future would be something like change in accounts payable. Accounts payable being a accounting term for money that is owed but not yet paid.

I don't know that the leading or lagging labels matter for my analysis. Really I just need to understand how to determine the strength and reliability of the relationship between two variables.


Less is more. Stay pure. Stay poor.
If they are continuous, a good start is just plotting correlations. The magnitude and signage (negative or positive) with tell you the potential relationship and directionality. You can also typically get a p-value (on whether the correlation is significantly different from zero.

If the variables are continuous normal (Pearson correlation), continuous non-normal (Spearman correlation), and a combination of categorical and continuous (biserial rank).