Observations of pulsar timing provide strong constraints on scalar–tensor theories of gravity, but these constraints are traditionally quoted as limits on the microscopic parameters (like the Brans–Dicke coupling, for example) that govern the strength of scalar–matter couplings at the particle level in particular models. For binary pulsars whose all five post-Keplerian parameters have been measured, we present fits to timing data directly in terms of the phenomenological couplings (masses, scalar charges, moment of inertia sensitivities and so on) of the stars involved, rather than to the more microscopic parameters of a specific model. For instance, for the double pulsar PSR J0737-3039A/B, we find at the 68% confidence level that the masses are bounded by 1.28 < mA/m⊙ < 1.34 and 1.19 < mB/m⊙ < 1.25, while the scalar charge-to-mass ratios satisfy |aA| < 0.21, |aB| < 0.21 and |aB − aA| < 0.002, independent of the details of the scalar–tensor model involved, and of assumptions about the stellar equations of state. Whenever it is possible to do so, we urge observers to express their results in this more model-independent way, which potentially can then be used by theorists to constrain a great variety of specific models by computing the fit quantities as functions of the microscopic parameters in any particular model. For the Brans–Dicke and quasi-Brans–Dicke models, the constraints coming from the double pulsar obtained in this manner are consistent with, and slightly weaker than, those quoted in the literature.