Although focussed on cyclists and not pedestrians, recent research from the University of Skövde concludes: it depends. The research appears to be ongoing. In 2014, they published a study where they researched what subconscious visual cues determine the intent of cyclists in traffic:
Hemeren, Paul E., Mikael Johannesson, Mikael Lebram, Fredrik Eriksson, Kristoffer Ekman, and Peter Veto. "The use of visual cues to determine the intent of cyclists in traffic." In Cognitive Methods in Situation Awareness and Decision Support (CogSIMA), 2014 IEEE International Inter-Disciplinary Conference on, pp. 47-51. IEEE, 2014.. This appears to be based on a lengthy report in Swedish.
Then, they investigated what visual enhancements help motorists to see this. I haven't found a peer-reviewed study for the second part. All I found specifically for the reflectors is press release in Swedish.
Some cyclists were wearing no reflectors at all, others were wearing some reflectors, and others were wearing reflectors at body places deemed optimal according to previous research. Their conclusion is that it is very important where on the body the reflectors are worn. From the press release:
– Det vi har sett i vår forskning är till exempel att reflexvästen ger en falsk säkerhet nattetid, då den inte på något sätt hjälper bilisten att tolka cyklistens rörelseinformation, säger Paul Hemeren, doktor i kognitionsvetenskap vid Högskolan i Skövde. Sätter man däremot en reflexrand bak på hjälmen, som fortsätter i ett lodrätt streck ner på ryggen, skapar man en linje som bryts av när cyklisten vrider på huvudet. Detta visar att cyklisten med stor sannolikhet ska svänga. Förstärker man dessutom kroppens andra leder med reflexer når man ett ännu bättre resultat. Med vår forskning kan vi påvisa hur man som cyklist kan synas ännu bättre i trafiken. Detta har ingen kartlagt tidigare.
Translation:
What we have seen in our research is that for example a reflective vest gives a false security at night, because it doesn't help motorists to judge the movement information by cyclists, says Paul Hemeren, doktor in cognitive sciences at the University College of Skövde. However, if one puts reflectors on the rear of the helmet, that continues straight down over the back, then one creates a line that breaks when the cyclist turns their head. This shows that the cyclist very likely will turn. If one additionally enhances visibility of other body parts with reflectors, one gets an even better result. With our research we can show how cyclists can be even more visible in traffic. Nobody has systematically looked at this before.
The press release also states:
Resultaten av forskningen är väldigt tydliga. Om reflexerna är placerade så att de förstärker cyklistens omedvetna rörelsemönster gör det att testpersonerna kan göra korrekta bedömningar av cyklistens avsikter i upp till 97 procent av fallen– även nattetid. Helt utan reflexer ligger testpersonernas träffsäkerhet strax över 70 procent.
Translation:
The result of the research is very clear. If the reflectors are placed so that they enhance the cyclist's unconscious movements, the test persons are able to make correct judgements of the cyclist's aims in up to 97% of the cases — including at night. Without reflectors the correct rate is slightly above 70%.