THE ART OF SCIENTIFIC INVESTIGATION 



" These results show how important it is when carrying out a 

 controlled investigation on human subjects to do everything 

 possible to ensure that the vaccinated and control children are 

 similar in every respect, including such factors as age, race, sex, 

 social, economic and housing conditions, intellectual level and 

 co-operativeness of the parents, risk of exposure to infection, 

 attendance at infant welfare or other clinics and treatment when 



in." 106 



Professor Wilson has pointed out to me in conversation that 

 unless decisive experiments are done before an alleged remedy 

 is released for use in human medicine, it is almost impossible 

 subsequently to organise an experiment with untreated controls, 

 and so the alleged remedy becomes adopted as a general practice 

 without anyone knowing if it is really of any use at all. For 

 example, Pasteur's rabies treatment has never been proved by 

 proper experiment to prevent rabies when given to persons after 

 they are bitten and some authorities doubt if it is of any value, 

 but it is impossible now to conduct a trial in which this treatment 

 is withheld from a control group of bitten persons. 



Sometimes it is a necessary part of a field experiment to keep 

 the groups in different surroundings. In such experiments one 

 cannot be sure that any differences observed are due to the 

 particular factor under scrutiny and not to other variables 

 associated with the different environments. This difficulty can 

 sometimes be met by replicating both test and control groups 

 so that any effects due to environment will be exposed and 

 perhaps cancel out. If variables which are recognised but thought 

 to be extraneous cannot be eliminated, it may be necessary to 

 employ a series of control groups, or carry out a series of experi- 

 ments, in order to isolate experimentally each known difference 

 between the two populations being compared. 



Whenever possible the results of experiments should be assessed 

 by some objective measurement. However, occasionally this 

 cannot be done, as for instance where the results concern the 

 severity of clinical symptoms or the comparison of histological 

 changes. When there is a possibility of subjective influences 

 affecting the assessment of results, it is important to attain 

 objectivity by making sure that the person judging the results 

 does not know to which group each individual belongs. No 



i8 



