AICE-JRF/SRF (Ph.D.): doctoral fellowship and admission stream
What the AICE-JRF/SRF (Ph.D.) paper tests
AICE-JRF/SRF (Ph.D.) is the doctoral admission and fellowship stream. It tests whether you can move from knowing a discipline to designing and defending research in it. Alongside subject mastery, expect items on research methodology, experimental design, sampling, statistical inference and interpretation of data.
Core rules and concepts
- The three canons of experimental design are replication, randomization and local control; each exists to separate treatment effects from extraneous variation.
- Match the design to the question: completely randomized design for homogeneous units, randomized block design when a known gradient exists, factorial arrangements when interactions are of interest.
- Statistical inference is about the data-generating hypothesis, not proof: state H0 and H1, choose the test, check assumptions, then interpret the p-value.
- Sampling items reward distinction between random, stratified and systematic sampling, and between sampling and non-sampling error.
- Fellowship and admission merit come from the same paper; a strong subject score without methodology accuracy still loses rank.
Worked example
A randomized block design has 5 treatments and 4 replications, so there are 20 plots. Total degrees of freedom = 20 - 1 = 19. Treatment df = 5 - 1 = 4; replication df = 4 - 1 = 3. Error df = 19 - 4 - 3 = 12. The shortcut 'treatments × replications minus 1' gives 19 and is the total df, not error df; using it inflates the error term and changes the F-test.
Traps that cost marks
- Treating a p-value as the probability that the null hypothesis is true.
- Confusing Type I error (rejecting a true null) with Type II error (failing to reject a false null).
- Ignoring the assumptions behind a t-test or ANOVA: independence, normality and variance homogeneity.
- Calling repeated measurements on the same plot independent replication.
- Mixing correlation with causation or R² with the significance of a slope.