1.6 Million “Left Behind”: Errors in Ghana 2021 PHC report “exonerate” U.S. Census Bureau

Billed as the country’s first fully digital census, the 2021 PHC has been applauded for presenting timely data within three months of exiting the field, though it is yet to be disclosed when the full results will be made available to the general public. The 2021 PHC Press Release on Provisional Results published by the Ghana Statistical Service (GSS) asserts that the de facto population in Ghana on Census Night was 30,792,608 being made up of 15,610,149 females and 15,182,459 males, situating the current national population density to 129 persons per square kilometer. It goes further to show the cumulative population figures for each of the sixteen (16) regions and remarks that the four most populous regions include, in descending order of magnitude: Greater Accra Region (5,455,692), Ashanti Region (5,440,463), Eastern Region (2,925,653), and Central Region (2,859,821), leaving the Northern region to the fifth position with 2,310,939 persons. But that is not all.

While declaring that the new results show a 24.9% increase in population from the previous census, the report further compares the national results with a range of benchmark extrapolations to test GSS’s degree of predictive accuracy with those of the United Nations Statistical Division and the U.S. Census Bureau. It claims that all the projections have deviated in the negative, namely the GSS 2021 mid-year population projections deviated by -2.7% of 31,636,804, the United Nations 2021 mid-year medium variant population projections by -3.0% of 31,732,129, and the U.S. Census Bureau projections by -4.9% of 32,372,889. By this conclusion, the report implies that the deflections in the U.S. Census Bureau’s extrapolations are the furthest from statistical reality, followed by those of the United Nations Statistical Division, making those of GSS the most reliable in the contest.

But impressive as it may sound to the casual reader, this conclusion has hefty implications and calls for objective scrutiny to establish its credibility. While it appears that the ‘Prophetic eye of the government’ may have taken the supposition for granted, a meticulous study of the report suggests that it may be far from right, albeit unwittingly. And unless otherwise indicated, there are gross data discrepancies, the analysis of which produce two irreconcilable outcomes. There is therefore a need for GSS to either provide some footnotes of palpable scientific justification to show that its results are beyond impeachment or, in the absence of any, respectfully revise same, as this may have grave ramifications for policy analysis, planning, research and development for stakeholders.

In fact, it can be deduced from the General Report Vol. 3A[ii] that the actual results of five pairs of extended tables do not tally with the totals supplied in all published reports. They simply do not add up, suggesting that as far as the 2021 PHC is concerned, the results that have been published so far are defective and untenable. Consider the table below and compare it with the totals supplied in the extended tables in the original document for verification.

TABLE 1: COMPARISON OF PUBLISHED AND CORRECTED NATIONAL AND REGIONAL POPULATION RESULTS

Finally, in testing the confidence levels in the provisional results, using the benchmark comparisons, GSS 2021 mid-year population extrapolations will deviate by +2.4% of 31,636,804, the United Nations 2021 mid-year medium variant population projections will deviate by +2.1% of 31,732,129, and the U.S. Census Bureau projections by +0.1% of 32,372,889 as shown in the Table 2. Per this analysis, since the U.S. Census Bureau extrapolations represent the smallest deflections from statistical reality as far as the 2021 PHC projections are concerned, it stands exonerated, followed by the United Nations. And that will mean also that the GSS projections are the furthest from the reality and therefore not reliable.

TABLE 2: COMPARISON OF PUBLISHED AND CORRECTED NATIONAL POPULATION WITH BENCHMARK VALUES