It’s the 93rd Academy Awards and this year the Academy has fielded quite a diverse set of nominees for once. This recaps and explores the woeful track record of BIPOC representation at the #Oscars (1/9)

TL;DR: the data suggest BIPOC actors 1) have to be better to get nominated (higher win rate conditional on nomination), 2) are more likely to be recognised and to win in supporting roles than as leads, and 3) are more disadvantaged if they happen to be female. (2/9)

The detail: since Hattie McDaniel’s win in 1940 (at a segregated ceremony where she wasn’t allowed to sit with the rest of the Gone with the Wind cast and crew), 92% of acting nominees have been white; not even close to representative of the US population. (3/9)

“Twice as good...”: When a BIPOC actor does make the cut they win 25% of the time. If winners were picked at random a nominee would have a 20% chance of winning, so this is higher than expectation, although not conventionally significant (p=0.134 #smallsampleproblems) (4/9)

“... to get half as far”: 61% of BIPOC acting noms are for Supporting roles (p=0.017*). An impressive 30% of Supporting noms for BIPOC actors led to wins vs 19% in Lead categories. For black actors 14% of Lead nominations end in a win vs 33% Supporting (p = 0.06.) (5/9)

This Lead/Supporting imbalance is larger for women: 30% of acting nominations for BIPOC women are in Lead roles vs 50% expectation (p=0.004**). For BIPOC men it’s 46% (vs BIPOC women’s 30% p = 0.07.) (6/9)

And overall BIPOC women are even less likely to be nominated *at all* than BIPOC men (7% of nominations vs 9% of nominations, p=0.07.) (7/9)

Notes on data: I used @Wikipedia entries on nominees, winners, and nominee characteristics. Noms analysis includes this year, winners obv does not. Given low # of BIPOC noms/ winners, subgroup analysis is hard and small changes matter (8/9)

Notes on classifying: racial and ethnic identity is complex and not for me to decide. I looked at self-ID when Wikipedia was unclear. E.g., Spanish actors (Cruz, Banderas etc) are Hispanic and in a minority in the US but do not typically self-ID as BIPOC. (9/9)