Because it is a model that predicts the future. It is not a metric, it is not direct data or a measurement. It is a calculated estimate which is very sensitive to initial conditions and assumptions.

And assumptions are informed by the biases and ideology of those crafting the model.

If everyone’s biases and ideology were perfectly aligned then modeling concerns might be limited to issues such as privacy and transparency. However education reforms of recent decades articulate a competitive marketplace system of winners and losers, defined by these models directly. Controlling the biases that inform these models, is tantamount to controlling the landscape of what constitutes winners and losers in that marketplace. Controlling the assumptions that guide the growth models is therefore imperative to the neoliberal vision of education reform as a marketplace.

Such underlying bias is referenced in this article written by an Aspire charter executive and ideologue, for CORE-PACE, one of the myriad hidden constituents of LAUSD’s “portfolio” of research partners:

…trading … growth models for proficiency models … will not necessarily reflect the values of those tasked with creating accountability systems in education.”

https://edpolicyinca.org/sites/default/files/PB_Cremata_Sept19.pdf

Those who set the presumptions of the model, control its consequent decisions.

More, while our public, democratic school system is supposed to be transparent – operated for, by and of the people, because the calculations and assumptions that constitute “growth data” are treated as proprietary secrets, its inherent biases are untraceable. We the public are not stakeholders in use of this data in any stage of its processing, from collection, harvesting, manipulation or consequent decisions.

We do not substantially know who has our children’s data, how it is manipulated, who is paying for – or being paid to – perform these calculations, who has authorized or overseen their use, how to characterize or adjudicate between competing model presumptions, or what decisions adopted modeling informs.