Everyone loves to whip out a statistic in a debate over moral issues, but we need to improve our statistical awareness to maintain credibility and consistency in our argument.
In all systems and businesses, statistics guide judgements and policies to further improve any aspect of that system or business. Statistical awareness involves people understanding the role and importance statistics have on our lives as well as how and why those who produce and quote the statistics use them (Almofarrij, 2020). In addition, like a fair test, statistics can not possibly control all of the variables involved in the social dilemma they are aimed at; they select one variable to measure and compare the results (in many cases this may be someone’s gender or race). Thus, statistical analysis of any data comes with an acknowledgement of an incomplete picture.
The use of statistics within the realm of social-political issues adds depth and credibility to one’s opinion. Having evidence to support your point of view puts you ahead in the battle for moral high ground. Subsequently, statistics are now being thrown around: out of context, from unreliable sources and without an acknowledgement of other variables and factors at play as well as discrepancies with each study. This has created a need for increased statistical awareness, which will be addressed in this article through the highlighting of key concerns surrounding this topic.
It probably does not shock people to hear that the media each have a political bias. It has been found that the political stance of each media outlet’s audience possess the same political values (Morris, 2007). Thus, people often only watch one news source. This has detrimental impacts upon the understanding of the events that occur in our society as this political bias often obscures the true details. A perspicuous example of this comes from a contradiction in the reporting of hydroxychloroquine by CNN. In May, when President Trump mentioned the treatment, CNN slandered President Trump for the mere suggestion with an unethically released study (which has now been retracted) stating that use of this drug increases the risk of death (Houck, 2020). Yet, less than a month later, CNN reported on how the drug positively treats people – hypocrisy.
If people were more inclined to consume news from a wide variety of sources, they will be able to interpret more accurate details from contradictions and the understanding of bias within what they are reading. Finding contradictions and diverse views and claims, will only entice people to pursue the truth and will logically support the reduction of widespread misinformation and unreliable statistics.
Sweeping statements ignore all variables
The quote “secretive and illegal pay culture” was used by Carrie Gracie (BBC news editor) back in 2018 to describe how the BBC pay their staff. This is what would be classed as a sweeping statement due to the claim being made based on a singular comprising factor that has been used to measure the pay differences between men and women – their gender.
Many claims are based upon one variable. Jordan Peterson in his 2018 interview with Cathy Newman spoke of at least 20 different variables that affect how women are paid. As it turns out, the more differing variables you analyse, the smaller the gap between the pay gets. Some of the other variables involved are as follows: career choice, qualifications, prior experience, maternity leave and personality traits. This is not to say that some of these factors are not immoral, but choosing a singular factor to generalise a social issue hinders discussion and reduces the credibility of the argument.
Unfortunately, this use of statistics to push agendas is far too common. You also see this in: permanent exclusions in schools, stop and searches, police brutality and the classification of hate speech. Each issue is comprised of a variety of factors. The claim that ‘police are racially targeting ethnic minorities’ fails to include crime stats, it just looks at the differences with arrest rates. The claim that ‘black children are treated unfairly in school’ fails to look at home background factors in to behaviour as well as what type of behaviour got them excluded. This use of statistics is only used to further political positions and win that battle for the moral high ground; it does not help the issue that is being claimed.
When these issues and claims are properly addressed and discussed, everyone should be looking at all of the possible variables and attempt to spot any correlations. If a majority of variables correlate then it is obvious that this is an issue that can be solved internally within systems and organisations.
Multivariate is better than univariate data
Claims are made to state a current problem within society. The issue with the majority of these claims is that they are based on univariate data. Univariate data looks at one variable at a time and therefore univariate analysis often leads to incorrect or inaccurate conclusions (CAMO, 2011). Many studies and collecting of data is univariate, but it would need to be accompanied by collected data of other factors to allow a thorough multivariate analysis. The example below is a univariate data collection of police interactions, where force was involved, based on the race of the individual having force used against them.
This in recent times especially, has been the leading source (there are many other sources that have measured in the exact same way) of the argument in police brutality and the mistreatment of BAME people by police forces. Laws for police use of force vary between countries; in the USA they can vary between states. While this statistic is accurate in its recording, it excludes other factors that contribute to the use of force within interactions with police. According to a UK police article on the use of force (TNS, 2015), police force can be used in circumstances of:
defence of another;
defence of property;
prevention of crime; and
The police report in 2015 expands on this stating the Crown Prosecution Service implore that the following factors must be considered when using force:
the nature and degree of force used;
the seriousness of the offence which is being prevented or in respect of which an
arrest is being made; and
the nature and degree of any force used against an officer by a person resisting
Information about guidance or training which an officer has received may be used to
assist in determining what is reasonable.
From this, it seems illogical to make the claim based on a univariate study. Logically, to make the claim that police brutalise BAME individuals or treat them unfairly compared to their white counterparts, you would need to analyse the causes for use of force – the type of crime being committed which requires use of force is one of these factors. In this study, if the majority of crimes being committed are violent, then you need to look at who is committing the most violent crime.
The same can be said in claims made about the unfair treatment of black boys in schools, white privilege and the gender pay gap as I mentioned earlier. These claims get massive backing from a large proportion of the population, yet very few take the time to analyse or even think of other factors that are intertwined with it. Thus, you will see this type of statistical analysis used in political campaigns and political arguments that are based upon the lack of statistical awareness of a large proportion of people in society.
There have been many debates where statistics have been used but do not directly prove or disprove the topic of debate, only support the direction of it. An article published by VICE (2020) was addressing the concerns with ‘Joint Enterprise’.
‘Joint Enterprise’, as defined in an article published by The Gaurdian (Bowcott, 2016) “involves crimes where more than one person takes part. The evidence rules enable those who did not strike the fatal blow or pull the trigger nonetheless to be convicted of murder.” It is used to hod accomplices accountable for their involvement in a crime – particularly gang crime in the last few decades.
In this article, the primary point is addressing that people are being wrongfully sentenced through this law. It provides anecdotal cases (this is not to say these people were not wrongfully sentenced) and a statistic to support the direction this article wants to follow – this law should be scrapped. Taking a step back and looking at the claim, it is certainly possible that it should be. It is true that disproportionately, black people are subjected to this law more than any other race – 37.2% (Bowcott, 2016) – despite the article not explicitly mentioning the statistic.
The statistic that was mentioned relates to the stigma of gang culture placed upon BAME males under this law. It reports that 80% of BAME males feel that gang narratives were invoked upon them compared to 38% of white males (Longley, 2020). While this may be true, it ignores the fundamental aspect of the direction of the article. Was this law successful and lawful in prosecuting individuals for the crimes they are claimed to have committed or had involvement in? To measure this, a success rate would be needed. Similar to stop and searches, the success rate is near 30% and has been shown to have no impact on crime rates and therefore poses the question of should we be continuing to carry this action out? While including feelings and perceived thoughts, it has no impact on whether the individual is guilty or not and therefore has no impact upon whether the law should be scrapped. Thus, making this particular statistic irrelevant to the article’s direction and purpose.
Surveying for thoughts and feelings around social and political issues has no baring on whether the argument is right or wrong. It only creates a moral atmosphere in the attempt to make something appear a certain way to the reader. This use of irrelevant statistics is used to push agendas in the battle for moral high ground. Though it can only be successful through poor statistical awareness.
Sample size of many surveys and studies bring about many problems. The construction of the sample size must be done appropriately based on the population specific to the topic so that more accurate inferences can be drawn and people do not lead to wrong conclusions (Sarmah and Hazarika, 2012). For a sample size that can be representative of the population, the minimum amount would need to be 384 (The Research Advisors, 2006). This being said, a sample size exceeding 400 would only increase the validity and the strength of its representation.
So it is important to understand the sample size of the statistic you are looking at. A study produced by the Northamptonshire rights and equality council (2014) produced the results claiming that overall, stop and searches was ineffective and that the authority police had was abused. The sample size of this study was 120 people and therefore not even close to being fairly representative of the population. On top of this, the study has more black males than white males in the study, meaning the black community are overrepresented in the study. This is not only inaccurate, it is bias due to the common knowledge and widely reported fact that blacks are disproportionately affected by stop and search powers. Not only this, but all participants in the study had been stopped and searched and there is no mention anywhere as to whether they had been carrying anything illegal, giving the presumption that all 120 in every case were innocent, despite the stop and searches having a 25% success rate. These three factors alone affect the credibility of the study despite the possibility of it being correct and stop and searches needing to be scrapped (as mentioned in an earlier section).
What needs to change?
This article was not set out to change people’s personal opinions nor beliefs, it was created to inform people of the factors involved in statistics and how you can build and support your opinions and beliefs more comprehensibly. If each person improved their statistical awareness, it is logical to think that mediocre statistics will not be forced upon us to push ideologies and political agendas. Each individual needs to think independently and assess what has been put in front of them before making their political stance.
Almofarrij, T., 2020. Statistical Awareness.. To Be Connected. [online] General Authority for Statistics. Available at: <https://www.stats.gov.sa/en/statistical-knowledge/51> [Accessed 3 August 2020].
Morris, J., 2007. Slanted Objectivity? Perceived Media Bias, Cable News Exposure, and Political Attitudes. Social Science Quarterly, 88(3), pp.707-728.
Houck, C., 2020. CNN Gave Over 90 Minutes in One Day to Now-Retracted Hydroxychloroquine Study. mrcNewsBusters, [online] Available at: <https://www.newsbusters.org/blogs/nb/curtis-houck/2020/06/04/cnn-gave-over-90-minutes-one-day-now-retracted-hydroxychloroquine> [Accessed 4 August 2020].
CAMO (2011) What is Multivariate Analysis? CAMO Analytics, [Online] Available at: https://www.smitconsult.nl/assets/Uploads/white-paper-MVA.pdf [Accessed 6 August 2020].
Bowcott, O., (2016). Joint enterprise law: what is it and why is it controversial?. The Gaurdian, [online] Available at: <https://www.theguardian.com/law/2016/feb/18/joint-enterprise-law-what-why-controversial> [Accessed 6 August 2020].
Longley, O., 2020. Black People In The UK Are Going To Jail For Crimes They Didn’T Commit. Here’S Why. [online] Vice.com. Available at: <https://www.vice.com/en_uk/article/n7wqqg/what-is-joint-enterprise-impact-black-bame> [Accessed 6 August 2020].
Sarmah,H. Hazarika,B. (2012) Importance of the size of Sample and its determination in the context of data related to the schools of greater Guwahati. [Online] Available at: https://www.researchgate.net/publication/306099484_Importance_of_the_size_of_Sample_and_its_determination_in_the_context_of_data_related_to_the_schools_of_greater_Guwahati#:~:text=The%20sample%20size%20is%20an,may%20lead%20to%20unreliable%20conclusions. [Accessed 6 August 2020]
Northamptonshire Rights and Equality Council, 2014. Stop And Think Northamptonshire. [online] Northampton. Available at: <https://northantsrec2013.files.wordpress.com/2012/12/northantsssreport-final.pdf> [Accessed 7 June 2020].
The Research Advisors (2006). Sample Size Table. [online] Available at: <https://www.research-advisors.com/tools/SampleSize.htm> [Accessed 7 August 2020].