By Dr. Assan Jallow.
Concerning the recent cepRass poll, I would like to thank the researchers for presenting for the very first time a polling study in the Gambia’s history. This is an excellent start in the evolution of our nation’s political development, which helps gauge public opinion about what influences citizens’ decisions during elections and which parties’ programs and policies they believe best represent the needs and aspirations of the people. A poll is a great initiative to gather public opinion as a means of expressing opinions about The Gambia’s forthcoming presidential election on December 4, 2021. However, although the poll appeared to be reliable, I questioned its reliability because data could have been skewed and I’m most interested in the margin of error and if it is within the range of 1-5, that isn’t bad. In addition, it would equally interest me to know the targeted respondents used to determine whether the study was nationally representative. This is the maiden poll and we cannot determine the consistency of the results obtained based on expressed opinions unless if there is a similar study of this kind. We would like to know what where the region of acceptance and rejection on the sampling distribution of the study, if there were any, and why?
We need to perform these critical assessments, and the best way to do this is by using the dataset to run a similar study or use their questionnaire to see if any noticeable differences exist between theirstudy and a counter-survey exercise in the form of replication or replicability. There are some striking factors in the 40 percent identified by the pollsters as undecided voters is too high a number to accept. To me, what is presumed to be the undecided votes cannot be ascertained as the assumption could be that among this identified or presumed category of the voter population, there are those who do not wish to reveal their choice of candidate or share it with the pollsters, until the election nears.
Additionally, I wonder if simple random sampling was used based on data analysis because it seems to me that the researchers used a textural format to place a greater emphasis on their study based on the altar of convenience to disguise the structural flaws of the poll. Based on that, I believed they used the non-probability sampling method through the convenience sampling and not the simple random sampling. Going by their submission, one can deduce a notice of visibility of convenience sampling which is an indication that their population was made up of volunteers holding all factors constant due to the use of data collection method of the study (i.e., telephone interview). Let me provide a context on convenience sampling. A convenience sampling is a type of a nonprobability sample made up of those volunteers or others who are readily available and willing to participate. It is obtained in a variety of ways and by its by its design, it has its advantages and disadvantages. That is obvious and it happens to any form of sampling design in any given study. It is wise, however, to avoid these types of overtly convenient samples because they may overrepresent one group in the population. This is particularly problematic if your sample is composed of those who have similar views on the subject one is investigating. It is, therefore, best advised that one need to make the convenience sampling a little bit more inconvenient by obtaining data from various places at various time to ensure global representativeness of the population understudy for a particular phenomenon.
I have not seen any strong correlation between the issues they highlighted in their study regarding the citizens needs and aspirations in regard to development prioritization. It beats my imagination based on their analysis that the surveyed respondents placed more emphasis on youth employment at 34% against education which represent 8%. I believe that if those are the expressed opinions of the respondent, then we have set the priorities incorrectly since the importance here is to protect a damaged good that cannot move Gambia from its current state of destitution, and from the poverty of ideas and underdevelopment it experiences.Youth employment cannot be addressed without a prosperous economy and a serious investment in relevant, quality, and practical education to provide a critical mass of our youth with the skill sets and knowledge to enter the job market and contribute effectively to society. This suggest that we need a shift in policy orientation, training, and integration to getting our priorities right as a country, thus, setting our youthful folks on the pedestals to organic growth to succeed in their endeavors. In terms of the construct of items in their study, I’m concerned, particularly about the grouping of items together, such as (term-limit, simple majority, diaspora voting, age limits for presidents, and paper ballots). Using that approach, one cannot determine what the study is trying to accomplish based on the number of variables highlighted in the study.
In my opinion the construct of the study fails the test -retest and the dynamics of construct validity and content validity. Construct validity indicates whether a measure mirrors the characteristics of a hypothetical construct, can be assessed in multiple ways and content validity on the other hand refers to an inclusion of all aspects of a construct by items on a scale or measure. Therefore, the studies unit of measurement is weak and cannot be substantiated unless the issues raised are addressed to ensure the reliability of the study. This is because sampling bias, interviewer bias and social desirability bias are all inevitable in any form of study or surveys, but we cannot just accept results or move to the alter of convenience in staying mute on issues of critical concerns as we move pass to December 4 and elect a continuing or a new government. I believed the study falls under the effects of serious sampling bias as the sample used does not represent the population of the country. Social desirability bias refers to how participant may respond based on how they want to be perceived or what is socially acceptable, and an interviewer bias happens when an interviewer provides a verbal or nonverbal cues that may impact how the participant responds.
This was a great effort and a commendable initiative from cepPrass, but we hope that they will be able to improve on the gaps and provide better polls that reflect the views of the people and not just a selected few.
To conclude, it is often said that in elections, the winner is determined before the election itself and not on the actual voting day, and this is especially true in elections in Africa, where incumbency can play a crucial role in winning elections.