Three years of result data, two competing explanations, and a proposed attendance bar
The IQAC of a degree college compiles three years of examination results for its four programmes. The table below is the whole of the data available.
| Programme | 2019 enrolled / passed | 2020 enrolled / passed | 2021 enrolled / passed |
|---|---|---|---|
| B.A. | 420 / 318 | 448 / 302 | 402 / 291 |
| B.Sc. | 180 / 151 | 176 / 140 | 168 / 140 |
| B.Com. | 210 / 178 | 205 / 168 | 214 / 186 |
| B.C.A. | 60 / 52 | 68 / 61 | 72 / 66 |
At a meeting, the principal says the 2020 drop proves that 'teaching quality has fallen' and proposes that attendance below 75 per cent become an automatic bar on appearing in the examination. The IQAC coordinator counters that the fall is 'only a B.A. problem'. You are asked to settle both claims with numbers before the file goes to the governing body.
1. Compute overall and programme-wise pass percentages for the three years, and quantify which programme accounts for the 2019 to 2020 fall.
Read the worked answer
Pass rates
| Programme | 2019 | 2020 | 2021 |
|---|---|---|---|
| B.A. | 318/420 = 75.7% | 302/448 = 67.4% | 291/402 = 72.4% |
| B.Sc. | 151/180 = 83.9% | 140/176 = 79.5% | 140/168 = 83.3% |
| B.Com. | 178/210 = 84.8% | 168/205 = 82.0% | 186/214 = 86.9% |
| B.C.A. | 52/60 = 86.7% | 61/68 = 89.7% | 66/72 = 91.7% |
| Overall | 699/870 = 80.3% | 671/897 = 74.8% | 683/856 = 79.8% |
Decomposition. Hold each programme's 2019 rate fixed and apply it to its 2020 enrolment. Expected overall passes = 0.803 x 897 = about 721; actual = 671, a shortfall of roughly 50.
- B.A.: expected 0.757 x 448 = 339; actual 302, shortfall about 37 (roughly three-quarters of the whole).
- B.Sc.: expected 148; actual 140, shortfall about 8.
- B.Com.: expected 174; actual 168, shortfall about 6.
- B.C.A.: expected 59; actual 61, a gain of about 2.
Conclusion: B.A. dominates the movement, mainly because it carries about half of total enrolment and its rate fell 8.3 points. But the B.Sc. and B.Com. rates also fell, so the fall is not exclusively a B.A. phenomenon; only B.C.A. improved.
2. Evaluate the two competing claims - 'teaching quality has fallen' and 'it is only a B.A. problem' - using only the data you have.
Read the worked answer
Neither claim is supported.
Against 'teaching quality has fallen': pass percentage is an outcome of intake, assessment and administration as much as of teaching. Three confounds are visible in these very numbers. First, enrolment rose from 870 to 897 in the year of the fall, so weaker candidates were probably admitted at the margin. Second, 2020 university examinations across the region were disrupted and often replaced by alternative assessment, so year-on-year comparison is not like-for-like. Third, the same B.A. rate recovered to 72.4 per cent in 2021 without any recorded change in staffing. A 5.5-point aggregate swing is also well within what cohort composition changes can produce.
Against 'only a B.A. problem': B.Sc. fell 4.4 points and B.Com. 2.8 points; both recovered in 2021. The aggregate is dominated by B.A. simply because B.A. is about half of enrolment, so 'B.A.-driven' is not the same as 'B.A.-only'.
What the table cannot answer at all: whether any change is due to teaching. It has no intake-quality measure (cut-off marks), no subject-wise results, no backlog or ATKT counts, no attendance data, and no record of the 2020 assessment scheme. Those four items should be requested before the governing body debates causes.
3. Assess the principal's proposed 75 per cent attendance bar as an instrument for improving pass rates, given these data.
Read the worked answer
The data cannot support the proposal. There is no attendance variable in the table, so the link between attendance and passing has not been estimated here at all. The bar is a plausible hypothesis, not an evidence-based policy - and it is an expensive one, because it converts attendance problems into loss of an academic year.
Where the risk falls. A bar on examination appearance would hit precisely the students the B.A. numbers suggest are already at the margin, and would reduce the number of candidates appearing - which mechanically raises the pass percentage while lowering the number of graduates. That is the classic way an indicator stops measuring what it was meant to measure.
A defensible alternative:
- For one semester, link attendance registers to examination eligibility and results by roll number and report the association programme-wise. If the association is strong after controlling for prior marks, the case for a bar exists.
- Run tutorial or mentoring support for the 60-75 per cent attendance band and evaluate it against a comparison group before making the bar universal.
- If a bar is adopted anyway, pair it with an early-warning letter at 80 per cent, a condonation route for medical or employment reasons, and a data audit before it becomes an eligibility rule.
- Ask for the missing evidence - intake cut-offs, subject-wise results, backlogs and the 2020 assessment scheme - before the file goes forward.