The law of evidence is a system under which facts are used to determine the rights and liabilities of the parties. Evidence proves the existence of “facts in issue” or infers them from related facts.
One of the fundamental considerations of criminal evidence law is the balance of justice to the accused against wrongful convictions, ensuring a fair trial and fairness to the victim. Since stakes in criminal trials are generally high, prosecution would have to prove the accused’s guilt beyond a reasonable doubt.
Since similar fact evidence – as evinced in s 11(b), 14 and 15 of the Evidence Act(“EA”) – can be highly prejudicial to the accused, and yet play an important role in proving culpability, it should be reformed to clarify its scope of application under the EA and reflect the common law’s prejudicial – propensity balancing test.
2. Similar Fact Evidence
Similar fact evidence is not directly connected to the case but is admitted to prove the existence of facts in issue due to its general similitude.The worry that the trier of facts would convict the accused based solely on evidence of prior conduct has generally led to the inadmissibility of such evidence in the trial as this is severely prejudicial. This concern is amplified in bench trials. Studies have suggested that judges were as influenced as laypersons when exposed to inadmissible evidence, such as prior convictions, even when they held that these evidences were inadmissible.  This is despite strong confidence in the judges’ ability to remain neutral to the inadmissible facts. 
Hence, the accused’s past similar offence or behaviour is inadmissible as similar fact evidence based on pure propensity reasoning would “colour the court’s ability to assess the evidence [objectively]”. 
Nonetheless, the mere prejudicial effect of such evidence does not render it inadmissible. It may be highly relevant, especially when such evidence is used to determine whether the acts in question were deliberate or to rebut a defence that could have been available. 
Under s 5 of the EA, a fact is only relevant if it falls within one of the relevancy provisions  stated in s 6 to s 10 of the EA. These provisions govern specifically the facts in issue, i.e. determinant facts that would decide whether the accused is guilty according to the substantive law governing that offence.In addition, s 11(b) was enacted to be the “residuary category” for the relevancy provisions.
In contrast, similar fact evidence is primarily administrated by s 14 and s 15 as it is concerned with conduct that is “merely similar in nature to those facts in issue”  .
s 14 is applicable only when the state of mind of the accused is in issue. Similar fact evidence is admissible to prove mens rea or to rebut the defence of good faith. 
s 15 deals with similar fact evidence that forms a “series of similar occurrences” to prove the mens rea of the accused,  and enables the prosecution to produce evidence to rebut a potential defence otherwise open to the accused. 
Admission of similar fact evidence under the EA is based on a “categorization” approach  , where such evidence is admissible pursuant to the exceptions stated in the EA  . Hence, similar fact evidence should only be admissible to prove the mens rea of the accused under s 14 and 15. 
Significantly, Singapore, India, Malaysia and Sri Lanka have identical s 11, 14 and 15 provisions in their respective Evidence enactments. With the exception of the Indian Evidence Act  , the three sections have not been amended since the 19th century. 
C. Case Law Interpretation of Similar Fact Evidence in EA
Singapore courts have not strictly adhered to the draftman’s intention in recent cases.  The court has incorporated common law’s “balancing test”  where the judge would balance between the probative and prejudicial effect of the similar fact evidence. 
In Lee Kwang Peng v PP  , scope of similar fact evidence was extended to prove actus reus. Pursuant to s 11(b), the court was of the view that the section “embodied”  the “balancing test”. Hence, similar fact evidence is admissible to prove both mens reas and actus reus.  Although readily admitting that this would be contrary to the draftsman’s intention, then Chief Justice Yong held that the EA should be a “facilitative statute” rather than a “mere codification of Stephen’s statement of the law of evidence”  .
The courts, until Law Society of Singapore v Tan Guat Neo Phyllis  (“Phyllis Tan”), were generally of the opinion that they had the discretion to exclude evidence substantially unfair to the accused. 
D. Under the Criminal Procedure Code Act (2010)  (“CPC”)
s 266 of CPC, dealing specifically with stolen goods, allows previous convictions of the accused to be admitted to rebut his defence of “good faith” and/or prove mens rea. However, notice would have to be given to the accused before adducing such evidence  .
E. In Comparison to Common Law
Unlike the EA, admissibility of criminal evidence at common law is “exclusionary”. As long as the evidence is logically probative, it is admissible unless it contravenes clear public policy or other laws. 
In Makin v A-G for NSW  , Lord Herschell formulated the “two-limbed” rule governing the admissibility of similar fact evidence. Under the first limb, the prosecution is not allowed to adduce similar fact evidence for pure propensity reasoning. However, under the second limb, evidence of the accused’s past conduct is admissible if relevant to the facts in issue via the categorization approach. 
Boardman v PP  reformulated these rules by incorporating the “balancing test”. Under this test, the probative force of the similar fact evidence must outweigh the prejudicial effect. Furthermore, such evidence should be “strikingly similar”, such that it would offend common sense if the evidence is excluded. 
However, the requirement of “striking similarity” was deemed to be too narrow in DPP v PP  . Instead, sufficient probative force could be gleaned from other circumstances  .
Nonetheless, similar fact evidence used to prove identity must be “sufficiently special” to portray a “signature or other special feature” that would “reasonably” point to the accused as the perpetuator of the crime.  Hence, similar fact evidence is admissible to establish actus reus  .
F. Defects of EA
Singapore courts have incorporated common law principles and extended the applicability of similar fact evidence to prove actus reus. This has resulted in inconsistencies between the draftsman’s intention and the court’s approach.
Further, there are some ambiguities that require clarification.
First, there are difficulties superimposing the Boardman’s balancing test into s 11, 14 and 15  as the EA was drafted to suit the categorization approach. The court would have to admit evidence falling under either section even if it may not be very probative or is highly prejudicial. Hence, s 14 and s 15 address only the “probative” part of the balancing test and leave no room for prejudicial effect considerations.  As a result, judicial discretion to exclude very detrimental evidence was developed. 
Second, although allowing similar fact evidence to prove actus reus would require the judge to evoke an additional step of inference  , limiting the use of this evidence to prove mens rea would allow extremely probative evidence to be excluded  .
Third, admissibility rules under the EA do not distinguish situations where the accused adduced similar fact evidence unintentionally. An injurious consequence would result if the prosecution is entitled to use such evidence. 
Fourth, under s 15, a single act, no matter how probative, is inadmissible.  However, this does not take into account the consideration that an act may be capable of supporting the argument “based on the rarity of circumstances”  .
Fifth, the courts have admitted similar fact evidence pursuant to s 11 (b) although it may not be pertinent to the facts in issue. Hence, EA should be amended to provide safeguards against such usage of s 11(b).
Lastly, in light of Phyllis Tan  , more protection against admissibility of similar fact evidence should be incorporated into the EA.
3. Possible Options
A. Survey of the other Jurisdiction
(1) Australia – Uniform Evidence Acts  (“UEA”)
Under UEA, “propensity evidence” and “similar fact evidence” are governed by the “tendency rule” in s 97 and the “coincidence rule” in s 98.
(a) Tendency Rule
Evidence pertaining to the “character”, “reputation”, “conduct” or “a tendency” that the accused possessed is inadmissible unless (a) notice is given to the accused and (b) the evidence has “significant probative value”. Although UEA does not state how “probative” the evidence has to be, “probative value of the evidence” is defined “the extent to which the evidence could rationally affect the assessment of the probability of the existence of a fact in issue”.  Operation of s 97 would be invoked when the evidence is adduced for the “purpose” of proving the accused’s inherent tendency to act in a particular way. 
(b) Coincidence Rule
As a general rule, evidence of “related events” – which must be “substantially and relevantly similar”  – would be inadmissible unless notice is given and the court is convinced that it possesses “significant probative value”  .
(c) Criminal Proceedings
A safeguard was inserted in relation to criminal proceedings under s 101 where the probative value of either the tendency or coincidence evidence would have to “substantially outweigh any prejudicial effect”. However, the degree of “probativeness” need not  be of such a degree that only one conclusion, i.e. culpability of the accused, could be drawn  . (“No rationale explanation test”)
UEA clearly adopts the “balancing test” approach, where the probative effect of the evidence must “substantially” outweigh the prejudicial effect on the accused. Furthermore, the introduction of the notice system would reduce the “prejudicial effect” as the accused not be unduly “surprised”.
However, as pointed out by the Australian Law Reform Commission, there is much ambiguity in what constitutes “significantly probative” and when the probative effect will “substantially outweigh” the prejudicial effect. Significantly, in Australia, the trier of fact and law are different.
Heavy reliance on the judge’s discretions, added with the ambiguity and wide application of the balancing test, would increase the risk of prejudice in bench trials.
(2) India – Statutory Amendments
Some amendments to s 11, 14 and 15 of The Indian Evidence Act  , upon which the EA was modeled on  , had been made pursuant to a review in 2003.  However, the changes made were very minor. 
(a) Section 11
An explanation, inserted after s 11(2)  , qualified the type of evidence rendered relevant under s 11 such that the degree of relevancy is dependent on “the opinion of the Court”  .
(b) Section 14
Clarifications to illustration (h) of s 14 were made such that the similar fact evidence has to show that “A” either had constructive or actual knowledge of the public notice of the loss of the property. 
(c) Section 15
The changes merely show that the acts made, pursuant to s 15, must be done by the “same person”. 
It would appear that the amendments to the Indian Evidence Act do not have any substantial impact on the law. However, the amendment of s 11 explicitly mentions that the degree of relevancy of facts is subjected to the “opinions of the court”, which suggests an approach more in line with the balancing test.
(3) England and Wales – Criminal Justice Act 2003  (“CJA”)
The admissibility of criminal evidence is presently governed entirely  by the CJA. Part II of CJA addresses the admissibility of “bad character”  evidence, which is defined as a person’s “disposition” for a particular misconduct.
Evidence of the defendant’s bad character can be adduced if it falls under one of the seven “gateways” under s 101(1)  . Significantly, under s 101 (1) (d), “bad character” evidence is admissible if it is “relevant to an important issue between the defendant and the prosecution”  . This includes the propensity to commit a particular type of offence which the accused is charged with or the propensity to lie  . Moreover, past convictions may be admissible to prove the defendant’s propensity to commit the crime he was charged with. 
Regardless, s 101 (3) allows the court to exclude evidence, falling under s 101 (1)(d), upon the application of the defendant if the court views that admission of such evidence would infringe on the fairness of the proceedings. 
The scope for the admissibility of “bad character” evidence has widened under the CJA.  In fact, evidence of bad character has changed from one of “prima facie inadmissibility” to that of “prima facie admissibility”  .
Although, under s 103, “bad character” evidence would be inadmissible if it does not heighten the culpability of the accused, suggesting that the evidence must have some probative force, it appears that the CJA has abandoned the “balancing test”.
Hence, “similar fact evidence” in the form of “bad character” evidence would be inadmissible if it lacks probative value in the establishment of the defendant’s culpability. However, s 101 (1), with the exception of s 101(1)(e), merely requires the evidence to be probative, without the need to outweigh prejudicial effect.
The EA would require more than a mere amendment. Hence, the type of reform advocated under the Indian Evidence Act should not be adopted. However, a radical change from the “categorization approach” to the “balancing approach”, following UEA, would give the court too much discretion. This is worrying as judges may not be entirely indifferent to inadmissible facts when deciding the case.
On the other hand, the approach under the CJA would be contrary to Phyllis Tan  and potentially allows highly prejudicial similar fact evidence to be admissible as long as the prosecution can show some probative value which points towards the guilt of the accused.
Hence, a hybridised model should be adopted where the “categorisation approach” under s 14 and s 15 is kept, but admissibility of such evidence must be subjected to the balancing test.
(1) Changes to the Headings
s 5 to s 16, under the general heading of “Relevancy of facts”, could be further sub-divided. First, s 5 to s 11 should be placed under a sub-heading of “Governing Facts in issues”  . Second, s 12 to s 16 could be placed under “Relevancy of other facts” sub-heading.
This sub-division of the relevancy provisions would clarify the functions of each section and discourage courts from admitting similar fact evidence which does not have a specific connection with the facts in issue via s 11(b).
(2) Incorporating “balancing test”
Most importantly, EA should be amended to include the “balancing test” with guidance drawn from the UEA, such that the probative value of the similar fact evidence should “substantially outweigh” the prejudicial effect. However, the definition of “substantially outweigh” should be left undefined and allowed to develop under the common law. The courts could draw assistance from the Australian case law.
(3) Proving Identity
A provision could be inserted to the EA allowing similar fact evidence to prove that the accused was responsible for the offence. In addition to the “balancing test”, the provision should also reflect the proposition that that such similar fact evidence must satisfy the threshold test of being “strikingly similar” to the offence, almost akin to a “signature” or “special feature” of the accused.
However, it is submitted that the extension of similar fact evidence to prove actus reus should be limited to the situation where the identity of the perpetuator is in doubt.
(1) Similar fact evidence produced by the accused mistakenly
It is proposed that such evidence should subjected to a higher standard of “balancing test” where the probative value “substantially outweighs” prejudicial effect.
(2) Only prior convictions are allowed
Prior acquittals should not be admissible as similar fact evidence as this would be unduly prejudicial to the accused. Furthermore, this could result in the undesirable situation whereby the accused is subjected to a “second round” of trials with regards to the prior charges.
In relation to the type of prior convictions that should be allowed, the EA should incorporate the definitions found in s 103(2)  and s 103(4)  of the CJA.
(3) Clarification of s. 15
s 15 should be extended to include single act or conduct of the accused to rebut the defendant’s defence of “accident”. However, in light of a potential danger of placing too much importance on a single episode, a qualification, such as the “no rationale reason”  test, should be implemented along with the extension of s 15.
5. Additional Safeguards
A system of notice, as seen in UEA, should be included into the EA. This reduces the prejudicial effect as the evidence would not be a “surprise”. Furthermore, this approach would be in line with the CPC, which has already adopted such a procedural reform.
Regarded as one of the most difficult area of the law of evidence, similar fact evidence can serve as a double-edged sword. It has the potential to convict the accused although the facts do not relate to the facts in issue and is highly prejudicial to the defendant.
In light of the rapid development of the rules governing the admissibility of similar fact evidence under the common law and the recent trend of common law countries codifying the test of admissibility into statutes, the EA should be reformed to reflect these developments, instead of relying on case law which can lead to inconsistency and uncertainty.
Further, as information of a person’s past is easily obtained with present technological advancements, safeguards against similar fact evidence should be incorporated so as to ensure a fair trial.