DIF Analysis

Previously, I had mentioned DIF (difficulty, importance, frequency) Analysis as a tool that I used in the military to determine if job tasks required training. I finally got around to creating the expanded model in a digital format, so here it is.


In making these tools available online some people ask if I’m giving away some secrets to the trade. I don’t think so, because these are pretty basic tools which I’ve been using for over a decade and many others use as well. Also, the world of work is getting to a point where performance improvement may not be the best approach. In knowledge-intensive workplaces, procedures and tasks can’t be easily quantified. Tools like DIF analysis only work when there are similar jobs done by several people. They won’t help in a creative work environment like a design shop.

My own interest is to develop new tools and methods, beyond human performance technology and instructional design. Methods like online personal knowledge mastery are of current interest.

18 Responses to “DIF Analysis”

  1. Karyn Romeis

    This is a good, common sense model, Harold, and I guess it would benefit many people to organise their thoughts like this when prioritising learning under pressure.

  2. Harold

    It’s a good tool when you have hundreds or thousands of tasks to analyse, as we did in the military.

  3. Cameron Bales

    Is it intentional that difficult important infrequent things are a higher training priority than frequent things – maybe self discovery or peer training might happen for frequent, but formal training is needed for infrequent?

  4. Harold

    The rationale is that if they are difficult & important, the training is necessary. You probably need to couple this with re-certification or annual checks.

  5. Harold

    If the task is performed on the job infrequently then it will likely be forgotten over time. Therefore, training is necessary to develop this difficult skill, and some type of reinforcement activity should be scheduled at regular intervals. A military example is shooting your rifle. For people in support trades who do not shoot frequently, there is annual refresher training, because this is an important skill.

  6. Clive Booley

    I agree that DIF is an excellent algorithm for complex task analysis.

    One bright spark in the Training Development Team I worked for in the UK Military, set it up in Excel so that as the results of the questionnaires came in, we could download the scanned data straight into the spreadsheet and generate the training category in an instant.

    I only wish I had lifted the spreadsheet before I promoted out!

  7. Harold

    It shouldn’t take that much effort to replicate it, Clive. Maybe someone will open source or Creative Commons license it 🙂

  8. Jake

    We use DIF ratings to decide: No training, initial training only, initial and continuing (refresher). We then started added “refresher every 3 {or 4 or 5} years. Unfortunately it has no empirical basis. Do you know of any way to empirically prove that a certain type of task needs to be refreshed every X period of time?

    • Harold Jarche

      I don’t think that the diversity of human learning and “empirical” go together. I would suggest internal studies to see what works as well as looking at current practices in specific fields for refresher training.

  9. Ed Hughes

    I am certain it is merely an oversight, but your last case (Not Difficult; Not Important) has the training priorities reversed. It should be None-None-Low, from top to bottom. As you know, a more frequently performed task will always require less training than one performed less frequently.

    Thanks for the diagram. It is a useful model first introduced to me at Fort Sill, Oklahoma in 1991 in their local training developers course.

  10. Denis Forest

    A better example of a high trg priority for an infrequent task are Aircraft Emergencies, e.g., engine failure. That happen very rarely (a good thing ;-)) in real life, however they are a high if not over train priority given their importance and difficulty.


Leave a Reply

  • (will not be published)