The second-order statistics of transionospheric RF scintillation are described by the two position, two-frequency, two-time mutual coherence function of the received signal. The derivation of the mutual coherence function from Maxwell's equations requires a model for the temporal and spatial variations of the electron density fluctuations in the ionosphere. Under Taylor's frozen-in hypothesis the electron density fluctuations are described as a rigid structure that drifts past the line of sight. There is then a deterministic relationship between spatial and temporal fluctuations in the electron density. This model is accurate when the ionization has formed a thin layer of striations aligned with the geomagnetic field lines. Before striations have formed or when there are multiple scattering layers in the ionosphere with different relative velocities, a turbulent model may be more appropriate. In the fully turbulent case the spatial and temporal fluctuations of the electron density are uncorrelated. Reality should lie somewhere between these two limiting models. This paper describes a general model which varies smoothly between the frozen-in and turbulent models. Examples are given which illustrate the effects on the received signal as the channel varies from the turbulent to frozen-in models.