Which tones are hardest, why, and where to start.
Only a single-subject case study exists (30 hours of pinyin/tone training, then read-aloud rated by native speakers): T2, the more marked tone, had a higher error rate than T4. With n = 1 this cannot serve as a prior.
Source: Assessment of Non-Native Learner Chinese Tone. http://has.hcu.ac.th/jspui/bitstream/123456789/2440/1/
Group averages tell you where to look, not what you actually do. Two speakers of the same first language can differ a lot. The only way to know is to record yourself and compare.
Take the free two-minute checkup →No signup, no download. Your audio is scored and discarded.