Khemraj Harryram’s bogus presentation of polling data makes a mockery of statistical validity

Aug 29, 2025 Letters

---

Dear Editor,

A frequently used cautionary adage among competent analysts is that “statistics don’t lie, liars use statistics.” Quite often, this adage also helps to remind people of the GIGO (Garbage In Garbage Out) statistical analytical claptrap. That is, if you put incomplete, inaccurate, or biased data into the computer, you would get false (bogus) results, or simply put, garbage’ for results.

Given the adage, an overall examination of Khemraj Harryram’s presentation of voting data (KN 8/27/25) compels one to question whether it is true, or bogus. He did state that he “was privileged to review the results” of the polling data which reveals that he is either not proficient in statistical analysis, or objective in his presentation. Here are some reasons why. He said that:

1. The polling was funded by “a private New York-based organisation currently exploring investment opportunities in the country.” The obfuscation of who funded the polling is questionable, if not suspicious, since investors do not use polling data to tell them what areas are best for investments. Instead, they rely on demographic “economic impact analysis,” a methodology commonly utilised by economists to provide reliable and valid information on preferable areas for investments. Harryram claims that his presentation was based on data collected by a group of University of Guyana students. This again is suspicious and compels one to ask: (i) Who was the lead researcher (supervisor) for the polling project? (ii) How many students participated in the data collection process? (iii) Their academic fields of study that prepare them for research; (iii) Who trained the students for data collection field work? (iv) What specific dates during which students collected the data? (v) Was the questionnaire pre-tested for reliability and accuracy? Having taught social science research methods and statistics for over 25 years, I had to provide answers to these questions to validate every research undertaken. Hence, Harryram’s failure to provide information on these questions creates suspicion regarding his analytical competency, and the data collection competencies of the students.
2. According to Harryram, “A total of 4,900 individuals were interviewed, with demographic weighting guided by data from the Bureau of Statistics.” This raises three very important concerns: (a) In elections polling, respected statisticians recommend that a sample size of 1,000 to 1,500 is required to provide accurate and meaningful results. Why wasn’t this done? This is necessary to reduce the “margin of error.” Simply put, to let the reader be aware that the results are inaccurate five per cent of the time, 10 per cent, or more? (b) Did the lead researcher do a “stratified weighted sample,” or just weighted sample? Both sampling techniques are necessary for accurate, meaningful and valid results since Guyana’s ten regions are comprised of different racial groups (must be weighted to ensure the various races are equally represented in the sample), the same with varying numbers of people living in the area (proportionate by population size). (c) Verification checks are essential to determine the accuracy, and reliability, of secondary data. That is, data already collected. In this case the Bureau of Statistics data. Anyone familiar with government data should know that the Bureau of Statistics collects and stores data primarily to inform public policy decisions and track strengths and weaknesses in areas of nation building. Almost always such data set is not readily amenable to secondary analysis without verification and/or veracity testing for alternative purposes/analyses. When used, the shortcomings must be noted.

In addition to the above, the biases in Harryram’ presentation are glaringly evident in his presentation of “Key Findings,” for which he provided the actual percentages of votes earned by the various political parties. Yet, in his presentation of “Voters Main Concerns” he failed to provide any statistical evidence (percentages as he did in his Key Findings).

Perhaps the most laughable of Harryram’s presentation is his tabular presentation “Total Expected Valid Votes,” for 2025, and his “Total Valid Votes from 2020. Here is a recap.

Regions	Total Valid Votes 2020	Total Expected Valid Votes (i.e 2025)	Difference
1	12,111	14,121	2,010
2	26,621	28,621	2,000
3	72,592	79,725	7,133
4	202,077	209,192	7,115
5	33,119	33,531	412
6	64,567	64,645	78
7	9,592	11,259	1,667
8	4,665	5,165	500
9	12,261	13,922	1,661
10	22,747	24,227	1,480
Total	460,352	484,408	24,056

 

As one can see from the figures in the above table, all the regions have increases in the voting population of over 1,000 voters. Regions Five and Six have less than 500 additional voters. If we accept Harryram’s explanation that Region’s Six voters of only 78 individuals is due to migration to North America, this seems to suggest that voters woke up one morning and said, “I am migrating to North America.” How ridiculous is this explanation when no data is provided to verify emigration from these areas?

Currently, with 757, 690 registered voters, and the expression of voters’ enthusiasm evidenced by PPP/C campaign attendances, it is ridiculous to project a three-way split in the electoral race, something Harryram claims without valid data. And in concluding his presentation with, “I trust this piece can be debated constructively” to which I say, “I did.”

Regards,

Narayan Persaud, PhD

Professor Emeritus

---

• bogus, Khemraj Harryram, Narayan Persaud, polling data, Professor Emeritus, Statistical

---