Failure rates in introductory programming

About three times the respondent pool of the authors' 2007 survey

CS1 instructors internationally.

A voluntary instructor survey with self-reported institutional failure rates, no verification, and self-selected responding institutions. The weakest design in this subject, included only for the base rate and the comparison. The full registered title is "Failure rates in introductory programming: 12 years later"; Crossref registers the main title separately from the subtitle. Read at abstract depth.

The average CS1 failure rate was 28% in 2017, down from 33% in 2007. The authors' own framing is the useful one: against 42-50% failure rates for US college algebra, an average CS1 failure rate of 28% does not seem alarmingly high. The premise that programming is exceptionally hard to teach is weaker than the field's rhetoric assumes.

Not applicable — descriptive base rates.

This is itself a replication of the authors' own 2007 study with roughly three times as many respondents, and it agrees with Watson & Li's independent harvest of published pass rates.