Sequential Behavior Analysis in Sport
Also known as: Lag Sequential Analysis, Sequential Pattern Analysis, Transition Probability Analysis, T-Pattern Analysis
Sequential behavior analysis treats a sporting performance not as a bag of independent events but as an ordered stream in which what happens next depends on what just happened. Drawing on Roger Bakeman and John Gottman's authoritative 1997 text Observing Interaction: An Introduction to Sequential Analysis, the method codes play into a time-ordered sequence of mutually exclusive events, builds a transition matrix counting how often each event is followed by each other event at a given lag, and converts these counts into conditional transition probabilities. Crucially, it tests those probabilities against what would be expected by chance, so that genuinely recurrent patterns of play — the move that reliably leads to a shot, the defensive action that triggers a turnover — can be distinguished from coincidence. Hughes and Bartlett's performance-indicator framework supplies the bridge from these tested sequences to actionable tactical knowledge.
Key highlights
- Captures the order and contingency of play that frequency counts and aggregate indicators discard.
- Tests observed sequences against chance, distinguishing genuine dependencies from base-rate artifacts.
- Identifies both excitatory links (sequences that recur) and inhibitory links (sequences that are avoided).
- Rests on the well-established, statistically grounded framework of Bakeman and Gottman, transferable across sports and interaction settings.
Intuition
This section is available to Pro members. Upgrade to Pro
How it works
This section is available to Pro members. Upgrade to Pro
When to use it
Use sequential behavior analysis when the order and contingency of events matters and you want to know which behaviors reliably follow which others, beyond what their base rates would predict. It is appropriate when performance can be coded into a single time-ordered stream of mutually exclusive categories, when you have enough events for stable transition estimates, and when your questions are about patterns and dependencies — build-up play, coach-athlete interaction sequences, momentum runs — rather than about totals. It is less suited to sparse data where many transition cells are empty, to questions that concern aggregate volume rather than sequence, or to settings where the relevant codes cannot be made mutually exclusive, since overlapping categories violate the method's core assumption.
Strengths & limitations
- Captures the order and contingency of play that frequency counts and aggregate indicators discard.
- Tests observed sequences against chance, distinguishing genuine dependencies from base-rate artifacts.
- Identifies both excitatory links (sequences that recur) and inhibitory links (sequences that are avoided).
- Rests on the well-established, statistically grounded framework of Bakeman and Gottman, transferable across sports and interaction settings.
- Requires mutually exclusive, exhaustive codes and an accurately ordered stream, which can be hard to construct for fluid play.
- Needs a large number of events for stable estimates, so rare but important sequences are poorly captured.
- Higher-order and time-varying dependencies are awkward to handle within simple lagged transition matrices.
- Significance testing across many cells inflates false positives unless multiple comparisons are controlled.
Common pitfalls
This section is available to Pro members. Upgrade to Pro
Applications
This section is available to Pro members. Upgrade to Pro
Frequently asked
Why isn't a high transition probability enough on its own?
Because a transition probability ignores how common the target event is. If shots follow ball recoveries 30 percent of the time, that sounds meaningful — but if shots follow almost everything 30 percent of the time, the recovery tells you nothing special. Bakeman and Gottman therefore compare the observed transition to what base rates alone would produce, using z-scores or adjusted residuals. Only transitions that occur reliably more (or less) often than chance indicate a genuine sequential dependency worth interpreting.
What does the 'lag' in lag sequential analysis mean?
The lag is the number of steps between the two events whose association you are testing. Lag 1 examines immediately consecutive events — what follows directly after a given action. Higher lags examine events separated by that many steps, letting you ask whether an action reliably precedes an outcome two or three events later. Examining several lags reveals not just immediate contingencies but the longer chains through which, say, a recovery leads to a shot after a few intervening passes.
How is sequential analysis different from ordinary notational analysis?
Notational analysis typically reports how often events and outcomes occur and builds performance indicators from those frequencies, treating events largely as a collection. Sequential analysis adds the dimension of order: it asks which events follow which, at what lag, and whether those contingencies exceed chance. The two are complementary — sequential analysis usually operates on a notationally coded event stream — but it answers questions about the temporal structure of play that frequency-based indicators cannot.
Sources
- 1.Bakeman, R., & Gottman, J. M. (1997). Observing Interaction: An Introduction to Sequential Analysis (2nd ed.). Cambridge: Cambridge University Press.ISBN 9780521574273
- 2.Hughes, M. D., & Bartlett, R. M. (2002). The use of performance indicators in performance analysis. Journal of Sports Sciences, 20(10), 739-754.
You have read it. What now?
Cite this page
ScholarGate. (2026, June 23). Sequential Behavior Analysis in Sport. ScholarGate. https://scholargate.app/sport-leisure-studies/sequential-behavior-analysis-sport