REF-1610 · Remmert et al., 2023 · Perceptual and Motor Skills · Paper · PubMed

Six weeks of practice did not significantly improve how accurately trained lifters called reps in reserve: nine trained men benched three times a week, taking the last set of each session to momentary failure while calling 4 and 1 reps left, and their absolute error did not change significantly; only the signed error drifted slightly (−0.077 reps per session) toward underestimating reps left in later sessions and in longer sets. Small, uncontrolled, from the abstract

Main caveat: Nine participants, no control group, no power calculation in the abstract; a true improvement could be missed.
limited, mixed evidenceMeasured on recreational to trained athletesRead from the abstract only

Remmert and colleagues tested whether reps-in-reserve calls made during a set get more accurate with weeks of training. After a week of familiarisation, nine trained men did three bench press sessions a week for six weeks; the final set of each session went to momentary muscular failure, and during it they said when they felt 4 and then 1 rep were left.

The difference between called and actual reps was analysed both with its sign and as an absolute error, in mixed models with session and proximity to failure as fixed effects and reps as a covariate. The signed error fell slightly over time and fell as sets grew longer; the absolute error showed no significant effect.

The authors conclude that accuracy did not significantly improve over time, although in later sessions and higher-rep sets there was a greater tendency to underestimate reps in reserve.

Participants

n = 9 trained men. Age, strength and training history are not in the abstract.

Protocol

Design: one familiarisation week, then six weeks of three bench press sessions a week.
Calls: in the last set of each session, taken to momentary muscular failure, verbal calls at perceived 4 RIR and 1 RIR.
Outcomes: raw (signed) and absolute differences between called and actual reps.
Analysis: mixed-effects models with session and proximity to failure as fixed effects, reps as a covariate, random intercepts per participant; p ≤ .05.
Interests: seven of the authors (JFR, ZPR, JCP, SRH, EE, ERH, MCZ) declare that they are coaches and writers in the fitness industry.

Key Results

Signed error over time: significant main effect of time (p < .001), estimated marginal slope −0.077 reps, a slight decrease.
Signed error with set length: estimated marginal slope of reps −0.404, the signed error falling as more reps were done.
Absolute error: no significant effects of time or of the other terms.
Authors' reading: accuracy did not significantly improve; underestimating reps in reserve (believing fewer were left than there were) became more likely in later sessions and in higher-rep sets.
Not in the abstract: loads, rep counts, the size of the error at the start or the end, and whether the 4 RIR and 1 RIR calls differed.

Practical Application

In nine trained men, six weeks of bench press with a failure set each session, calling 4 and 1 reps left, did not make the calls significantly more accurate. For an already experienced lifter, more weeks of the same practice may not sharpen the call much; the improvement in REF-1609 came in people new to the method, and a non-significant change in nine men is not proof that none occurs.
The small drift toward saying fewer reps were left than there were means the calls got slightly more conservative over the weeks (our reading of the authors' wording).
Wiedenmann et al. read this result and REF-1485 as consistent with a ceiling: lifters whose starting error is around one rep or less have little room to improve (REF-1609; their interpretation, not tested).
Seven of the thirteen authors declare roles as coaches and writers in the fitness industry.

Limitations

Nine participants, no control group, no power calculation in the abstract; a true improvement could be missed.
Abstract only: loads, rep ranges and baseline error are unknown.
One lift, men only, already trained.
Calls were checked in one set per session only.
Authors' declared fitness-industry roles.
How this entry was checked and changed
Why this level

an uncontrolled repeated-measures study is B; one step down because n = 9 and only the abstract was read, which gives no loads, rep targets or starting accuracy and no power calculation, so its null result has little behind it