When using Duolingo to learn German, I'm often torn between the choice of reviewing existing language skills that I've already learned and continuing on to new skills. Is it more effective to review or learn new concepts? Generally, I learn new lessons for 2-3 days, then have a review session for about 2 days. Is it more effective if I learn new skills and review during the same session every day?
A lot of research has been done into the ideal spacing of revisions - just try to search for topics such as "spaced repetition" and "forgetting curve"! A rule of thumb: the most effective strategy is to review each item when the chances that you have forgotten it are 5-15%. In other words, revising so soon that you still remember everything is a waste of time (you spend unnecessary amounts of time revising); revising so late that you have forgotten more than every sixth item is also a waste of time (you spend unnecessary amounts of time relearning things that you once knew). A lot of language learning / spaced repetition software (duolingo included, https://www.quora.com/Does-Luis-Von-Ahn-have-any-plans-for-optimizing-Duolingos-vocabulary-learning-using-spaced-repetition) tries to set up the learning intervals so that you revise items in the right interval, neither too soon nor too late. The main question is how good their algorithms are at estimating when the right moment comes. It is a hard task, partly because users may not feel comfortable with whatever strategy is theoretically optimal (forgetting 10% of items may be too frustrating to many users).
Another thing to consider is this: it has been hypothesized that one of the advantages of using spaced repetition is that you end up revising items from several different lessons. In this way, your brain has a chance to create new connections that would be much harder to spot if you were always revising all items from one lesson, then all items from another lesson etc.
There are advantages to either "repeating" or "leapfrogging."
The advantages of "repeating" existing skills are fairly obvious. You can become more sure that you've actually learned the material.
The advantages of "leapfrogging" are harder to quantify. Under this scheme, you cover more ground, and gain more "passive" knowledge, quicker. The issue then becomes, can you take advantage of your greater quantitative knowledge to make up for what you lack in "practice."
Over a very long period of time, your knowledge of a language comes together (at least that has been my experience), once you reach "critical mass." The danger with "leapfrogging" is that if you don't reach that critical mass, you will end up with too weak a foundation. For "foundational" purposes, the repetitive method is slower but surer unless you have a plan, or at least a knack, for quickly arriving at critical mass.
"Repetition is the mother of learning." It applies to all kinds of learning, including learning a language.
In my experience (when I was learning English and German), reviewing the already learned subjects is very important in the beginning, but once you come to a basic/intermediate level you are able to progress with new material much quicker.