Reproducibility of statistical data, academic publications and policy implications: Evidence from Ghana