A cotton trial can produce a perfectly formatted spreadsheet and still support the wrong conclusion. An entry with heavier harvested bags may have occupied a larger plot. A field with more insects may simply have been sampled more intensively. A demonstration attended by forty people may document forty visits, not forty farmers who adopted a practice. These are useful problems to practise before a research interview because they expose how a candidate thinks when the numbers look persuasive.
This workbook is for applicants considering SKRAU's Young Professional engagements at Agricultural Research Station, Sri Ganganagar. It develops the practical reasoning behind field observation, crop-improvement trials, surveillance and communication. Every numerical exercise below is an original hypothetical teaching dataset, not an SKRAU trial result, a previous interview question or an official assessment scheme. Work through the calculations before reading each explanation; then practise giving the answer without looking at the page.
For the recruitment record and application particulars, use SarkariSapna's SKRAU seven-post recruitment page. The preparation here complements that record rather than treating the advertisement as a complete scientific syllabus.
📌 First establish which appointment you are preparing for
SKRAU's signed Letter 588, dated 22 October 2026, lists five YP-I and two YP-II positions. YP-I requires B.Sc. Agriculture with at least 60%; YP-II requires M.Sc. Agriculture in Genetics and Plant Breeding with at least 60%. Fixed monthly remuneration is ₹30,000 and ₹42,000 respectively. The walk-in is on 31 October 2026 at 10:15 AM at ARS Sri Ganganagar. The notice specifies ages 21–45 with Rajasthan-rule relaxation, without stating an age-reference date. Initial engagement runs up to March 2027; continuation depends on performance and funding.
Carry the plain-paper application, bio-data, original qualification/experience documents, one attested copy set and two recent passport photographs. No interview TA/DA is payable. Research experience and computer knowledge are desirable. A written-test pattern, interview marks distribution and detailed question syllabus are not supplied. Check the university announcement and its signed notification before travel.
The project table separates BG-II hirsutum improvement, Desi cotton improvement, pest/disease monitoring and dissemination. Use that separation to select examples from your own experience. A dissertation in breeding and a field-surveillance internship can both be valuable, but they demonstrate different abilities. Do not imply that competence in one makes you formally eligible for every advertised role.
🧭 Build a preparation file around an observable skill
| Preparation route | Bring an example of | Be ready to explain |
|---|
| Crop-improvement research | A plot layout, trait sheet or thesis table | Experimental unit, checks, randomisation, measurement and uncertainty |
| Field observation | A legible raw record and its cleaned version | Identity, units, dates, corrections and missing observations |
| Pest/disease monitoring | A sampling sheet with effort recorded | Why counts, incidence and severity answer different questions |
| Dissemination | A short explanation and a follow-up record | What was communicated, what was understood and what was actually verified |
These are suggested evidence types, not additional documents prescribed by the university. A candidate without previous employment can prepare a clearly labelled practice example. A candidate with experience should use material they are permitted to share, remove personal identifiers and avoid presenting a supervisor's analysis as their own. The strongest example is usually small enough to explain accurately, including what went wrong and how the correction was documented.
🌱 Case 1: The larger harvest bag does not necessarily identify the better entry
Three teaching plots were harvested. Entry A produced 2.70 kg from a net harvested area of 12 m²; B produced 3.00 kg from 15 m²; C produced 2.64 kg from 12 m². Rank their seed-cotton yields after making the areas comparable.
Calculation: plot weight ÷ net harvested area × 10,000 gives kg/ha. A is 2,250 kg/ha; B is 2,000 kg/ha; C is 2,200 kg/ha. Raw bag weight would place B first; the area-standardised estimate places it last. The area must belong to the portion actually harvested, not automatically to the gross plot including borders, paths or unharvested rows.
📊 Practice dataset: area-standardised seed-cotton yield
| Teaching entry | Yield kg per hectare |
|---|
| Entry A | 2250 |
| Entry B | 2000 |
| Entry C | 2200 |
The chart establishes arithmetic rankings in this exercise only. It does not establish statistical superiority, repeatability across sites or a variety recommendation. With one observation per entry, an interviewer can reasonably ask whether the difference reflects genetics, plot conditions or measurement error. A sound answer gives the ranking and immediately defines its limited scope.
Follow-up: an assistant later discovers that C's recorded area should be 11 m². Do not quietly change the final chart. Check the field map and harvest note, document the authorised correction, recalculate C as 2,400 kg/ha, and regenerate any table derived from that record. One corrected denominator changes the apparent leader. This is why a result needs traceable inputs.
🧵 Case 2: Seed-cotton yield and lint yield can rank entries differently
Continue with the original, uncorrected Case 1 areas. Suppose the teaching ginning-outturn values are A 34%, B 40% and C 36%. For this exercise, ginning outturn means lint mass as a percentage of the corresponding seed-cotton mass, measured on a consistent basis. Multiply seed-cotton yield by that fraction.
| Entry | Seed cotton kg/ha | Ginning outturn | Calculated lint kg/ha |
|---|
| A | 2250 | 34% | 765 |
| B | 2000 | 40% | 800 |
| C | 2200 | 36% | 792 |
B now leads on calculated lint yield even though it had the lowest seed-cotton yield. Neither measurement is wrong. They describe different outputs. The next question is which output the research objective requires, together with fibre properties and other relevant observations. A candidate who simply announces the highest ginning percentage has skipped the yield component; one who considers only seed cotton has skipped the conversion to lint.
ICAR–CICR's crop-improvement programme discusses selection using yield and fibre traits and subsequent evaluation. That research context supports learning to read more than one trait; it does not supply a universal index or weights for this SKRAU engagement. Ask what was measured, how, and for which objective before combining traits into a single score.
📐 Case 3: Twenty plants are not automatically twenty replications
Imagine two cotton entries, each planted in one plot. An observer records boll weight on twenty plants in each plot and calls the experiment “twenty replications per entry”. Explain the problem in ordinary language.
The entry was assigned to a plot. Plants measured inside that plot are subsamples sharing the plot environment. Their measurements can describe variation within the plot, but they do not create twenty independently assigned plots. A plot-level fertility difference remains entangled with the entry difference. More observations within the same plot cannot, by themselves, repair that design limitation.
A clearer practice design uses multiple plots per entry, with treatment allocation determined before planting. In a complete block arrangement, each block contains every entry, with allocation randomised within the block. Blocking addresses an anticipated source of field variation; it does not give permission to place the preferred genotype in the preferred position. See NIST's explanation of blocking and the university-hosted randomized complete block teaching resource.
Oral answer: “I would first identify the unit to which the entry was independently assigned. The twenty plants improve the description of one plot, but they are not twenty independently allocated plots. I would retain the plant observations and their plot identifiers, and use an analysis that respects the design.” This answer avoids both throwing away useful data and overstating what those data can prove.
🔎 Case 4: A higher mean is not the same statement as a demonstrated effect
In a three-block teaching trial, entry A yields 2,100, 2,250 and 2,400 kg/ha. A check yields 2,050, 2,200 and 2,250 kg/ha. A's mean is 2,250; the check's mean is approximately 2,166.67. The numerical advantage is 83.33 kg/ha, or about 3.85% relative to the check.
| Block | A kg/ha | Check kg/ha | A minus check kg/ha |
|---|
| 1 | 2100 | 2050 | 50 |
| 2 | 2250 | 2200 | 50 |
| 3 | 2400 | 2250 | 150 |
The defensible first sentence is “A has the higher observed mean in this dataset.” Whether the evidence supports a statistical claim requires the appropriate model, error estimate, assumptions and comparison procedure. Whether the difference matters in practice is a further question. Neither question is answered by placing an asterisk next to the larger mean.
If asked to proceed, explain what you would inspect: the allocation plan; whether every recorded value belongs to the correct plot; losses or unusual field events; units; and residual behaviour under the proposed analysis. Do not invent a p-value from the percentage difference. Also avoid saying that a small difference must be insignificant; evidence depends on variability and design, not size alone.
🌦️ Case 5: Equal average yields can conceal different responses
Suppose entry A yields 2,800 kg/ha in teaching environment 1 and 1,800 in environment 2. Entry B yields 2,400 and 2,200 respectively. Both simple means are 2,300 kg/ha. A is ahead by 400 in the first environment and behind by 400 in the second.
Calling them identical because their averages match loses the response pattern. Calling B “climate resilient” from these four figures also exceeds the evidence. The environments have not been characterised, replication is not supplied, and no specific stress response was measured. The sensible description is that rankings change between these two hypothetical environments and that the average alone hides this.
Prepare to name the information you would request: site and season, soil and management, weather or defined stress exposure, trial design, checks and uncertainty. For a breeding discussion, this is more useful than reciting “genotype by environment interaction” without explaining what in the data prompted the concern. Distinguish a pattern worth investigating from a mechanism already established.
🐛 Case 6: More insects counted, lower density observed
On visit 1 an observer counts 84 insects across 28 leaves. On visit 2 the count is 120 across 60 leaves. The raw count increased by 36, approximately 42.86%. The average declined from 3 to 2 insects per leaf, a 33.33% decrease. Both calculations are correct; they answer different questions.
📊 Practice dataset: insects per sampled leaf
| Visit | Insects per leaf |
|---|
| Visit 1 | 3 |
| Visit 2 | 2 |
A useful field note would report the numerator and the denominator, not simply “pest increased” or “pest declined”. Before treating the density change as biological, ask whether leaf position, plant selection, observation time, pest life stage and counting method were comparable. Standardising by leaf count fixes unequal effort in this example; it does not remove every other source of difference.
ICAR–CICR's technical bulletin directory provides cotton IPM and resistance-monitoring resources. Choose a source that matches the organism and question. A trap catch, a count per leaf and damaged bolls are different observations. None should be converted into a pesticide decision using a threshold copied without its sampling method and current local guidance.
🍃 Case 7: Incidence is not severity
In a hypothetical assessment, 12 of 60 plants meet a specified disease-symptom criterion. Incidence is 20%. The remaining information does not tell us how much tissue is affected on the twelve plants. One small lesion and extensive affected tissue can both count as “present” under a binary incidence rule.
Now consider a separate classroom severity-rating exercise. Twenty assessed units receive ratings 0, 1, 2 or 3. Their frequencies are 8, 6, 4 and 2. The weighted score is (8 × 0) + (6 × 1) + (4 × 2) + (2 × 3) = 20. Dividing by 20 units × maximum rating 3 and multiplying by 100 gives a normalised rating index of 33.33%.
This example does not prescribe a cotton disease scale. The index is only interpretable if the rating definitions and the sampling unit are specified. In particular, an ordinal rating index must not automatically be relabelled as the exact percentage of leaf area damaged. The exercise's incidence under “rating above zero” would be 12/20 = 60%, demonstrating how incidence and a normalised severity index can differ without either calculation being an error.
| Report statement | What must accompany it |
|---|
| 20% incidence | Criterion for an affected unit; numerator 12 and denominator 60 |
| 33.33% normalised rating index | Rating definitions, frequency counts, maximum score and formula |
| Actual damaged leaf area | A method measuring or estimating area; an ordinal label alone is insufficient |
🪤 Case 8: A zero, an empty cell and a broken trap
A monitoring file contains four observations: a functioning trap with no catch; a missing visit; a damaged trap recovered after an unknown period; and a valid count of nine. Entering 0, 0, 0 and 9 makes the average 2.25 but erases why three records differ. It implies that three valid observations found no insects, which the scenario does not support.
Keep the confirmed zero as a numeric observation. Record the missed visit and damaged trap as missing counts with separate reason fields. The two valid counts average 4.5, but even this number should not be published as a complete four-occasion series. State that only two observations were valid. A transparent incomplete dataset is more informative than a complete-looking invented one.
For trap comparisons, also record exposure. In another hypothetical pair, 30 catches across two traps operating for three nights equal 5 catches per trap-night. Forty catches across two traps for five nights equal 4. Standardising exposure changes the comparison, but it does not guarantee equivalent catchability, trap condition or weather. Describe the rate as an observation under the stated method rather than a direct census of the field population.
🧪 Case 9: Why a field-control failure does not prove resistance
A farmer reports that insects remained after a treatment. An applicant immediately labels the population resistant. What information has been skipped? Identity and life stage of the pest, timing, coverage, product and application records, weather, subsequent immigration and the observation method can all affect interpretation. A resistance claim needs appropriate evidence under an established investigation, not merely confidence in a field impression.
A useful interview response separates recording from diagnosis: document the report and its context; preserve the relevant observation and sample information; bring it to the responsible scientist; and follow the approved sampling or assay procedure. Do not improvise a bioassay concentration, disposal method or recommendation during a practice discussion. The job-preparation objective is to demonstrate disciplined escalation and interpretation.
When reading a resistance-monitoring publication, make a small extraction note: organism, life stage, response measured, comparison population or baseline, exposure conditions, replication and limitations. This makes the reading operational. Memorising a technical term without identifying its denominator or control rarely helps when the interviewer changes the scenario.
💻 Case 10: A spreadsheet merge that multiplies the harvest
Your plot register has one row for plot P07. Its harvest table has three rows because cotton was picked on three dates. A third table has four plant measurements for P07. Joining both detailed tables only on plot ID creates twelve combinations. If you sum harvest weight after this join, each picking can be counted four times.
The error arises from joining two repeated-record tables without respecting their level of detail. First decide the question. For total plot harvest, sum the three pickings into one plot-level value after checking uniqueness of each picking record. Summarise the plant observations separately if the intended analysis needs plot means, then join the two plot-level summaries. Alternatively, keep separate linked tables and analyse them at their proper levels.
| Table | Proposed practice key | One row represents |
|---|
| Plot register | Trial ID + plot ID | One allocated field plot |
| Harvest | Trial ID + plot ID + picking ID | One recorded picking |
| Plant observations | Trial ID + plot ID + plant ID + observation date | One plant observation on a date |
| Corrections | Correction ID | One documented change with reason and authorisation |
Before and after a merge, compare row counts and totals whose behaviour you understand. If a file unexpectedly grows from three harvest rows to twelve, investigate before drawing the graph. Keep identifiers as identifiers: a plot named 001 is not an invitation to let the spreadsheet convert it to 1 if that breaks the field label convention.
📁 Turn the ten cases into a usable field-data dictionary
A data dictionary should allow another person to understand a column without guessing. “Yield” is insufficient when it could mean a picking weight, total plot seed cotton, lint yield or an extrapolated hectare figure. Include the variable's meaning, unit, allowed values, missing-value treatment and the point at which it was derived.
| Suggested field | Example | Validation question |
|---|
| Trial and plot identifier | TRAIN26-P07 | Does it match the allocation map and bag label? |
| Observation date | 2026-10-23 | Is the date explicit and in a consistent format? |
| Net harvested area | 12 m² | Does it describe the actual harvested portion? |
| Picking weight | 2.70 kg | Are unit, tare treatment and picking identity recorded? |
| Sampling effort | 28 leaves | Is the denominator available for a count? |
| Missing reason | Visit missed | Is absence of data distinguished from observed zero? |
| Method version | Training protocol v1 | Can a method change be identified later? |
| Correction reference | COR-003 | Can the change be traced to evidence? |
These are practice conventions, not a required SKRAU file template. Avoid filling an unknown field with a plausible value just to make validation pass. In the field, a useful habit is to review the sheet before leaving the location: missing identifiers and units are easier to resolve beside the labelled plot than after several days of travel.
📢 Case 11: Attendance, understanding and adoption need separate evidence
A practice outreach register contains 48 attendance entries. Deduplication identifies 36 people. Twenty-four of those people complete a follow-up check; fifteen can correctly explain the demonstrated procedure; nine have a verified record of using it under the defined follow-up method.
There are several valid ratios: 24/36 = 66.67% follow-up coverage; 15/24 = 62.5% correct explanations among those followed up; 9/24 = 37.5% verified use among those followed up. Nine of all thirty-six attendees is 25%, but non-followed-up people have unknown outcomes. None of these numbers proves that the programme caused an increase over what would otherwise have happened.
Do not put these overlapping stages into a pie chart: a person can belong to attendance, follow-up, understanding and use simultaneously. A table with explicit denominators is better. A pie is appropriate for a mutually exclusive partition, such as the teaching follow-up status below.
📊 Practice dataset: follow-up status of 36 unique attendees
| Status | People |
|---|
| Follow-up completed | 24 |
| Follow-up not completed | 12 |
A credible outreach note might read: “We recorded 36 unique attendees and completed follow-up for 24. Nine had verified use under our defined follow-up method. Outcomes for twelve attendees remain unverified.” That sentence is less dramatic than claiming universal adoption, but it gives a supervisor information that can guide the next visit.
🗣️ Case 12: Explain one finding to a farmer and to a research supervisor
Take Case 1's original yields. For a farmer-facing practice explanation, say: “These small teaching plots had different harvested areas, so we first compared them on the same area basis. A gave the highest seed-cotton figure here. We would need suitable repeated local evidence before advising a change.” Avoid speaking as if a classroom dataset were a local demonstration result.
For the technical note, include the raw weights and areas, conversion, missing design information and the limited inference. Explain that the chart contains point values without uncertainty estimates. The same evidence can support different levels of detail for different audiences, but it must not become a different claim when the audience changes.
Practise answering the follow-up “Would you recommend A?” A precise answer is conditional: the exercise identifies A's observed seed-cotton yield advantage, but supplies neither adequate comparative trial evidence nor the farmer's agronomic and economic context. Also remember Case 2, in which a different output changes the ordering. Clarifying the decision objective is part of communication, not an evasion.
🎙️ Prepare a research interview answer that survives a second question
A rehearsed introduction often collapses when the panel asks which part of the work the candidate personally performed. Prepare one experience account in five connected parts: the question, your assigned responsibility, the method you used, the observation you can support, and a limitation. Give concrete examples instead of claiming proficiency in every software package or experimental technique.
For example, a field assistant could explain that they checked plot labels, maintained picking records, identified a duplicate row and obtained confirmation before correcting the compiled file. A postgraduate could explain why their thesis used a particular comparison, what assumptions mattered, and what they would redesign. Neither should claim the final research conclusion solely because they participated in one stage.
| Question to rehearse | Evidence a strong answer supplies | Avoid |
|---|
| What did you personally do? | A specific task and traceable output | Taking credit for the whole project's work |
| Why this denominator? | Sampling unit, effort and intended comparison | Using whichever ratio looks strongest |
| What would change your conclusion? | A limitation or additional observation that matters | Claiming the result is beyond revision |
| What if the data are incomplete? | Missingness recorded, scope restricted, follow-up planned | Substituting zeros or invented values |
| Which project interests you? | A connection between the named work and a real skill | Treating interest as qualification equivalence |
🧬 A focused reading route for the breeding applicant
Start with the research question rather than a list of definitions. If the question concerns improvement, identify the target trait, material, comparison and environment. If it concerns selection, explain what observations justify retaining an entry for further evaluation. If it concerns broad suitability, explain what additional environments or seasons would make the claim more persuasive.
Read one section of the CICR crop-improvement account slowly. Extract the material studied, trait measurements, comparison and next step. Keep the institute's reported experiment separate from your own work and from SKRAU's project. Being able to explain a real research paragraph accurately is more useful than memorising a large catalogue of variety names without their context.
Then apply the same reading discipline to your thesis. Can you reconstruct the main table from the raw data? Can you state which measurements were made by you and which came from a laboratory? Can you distinguish a numerically larger mean from a supported statistical claim? If an answer is uncertain, revisit the original method rather than constructing a polished but inaccurate explanation.
🗓️ A six-session preparation sequence with tangible outputs
Use the following as an adjustable sequence, not as an official timetable or a promise of interview success. A session is complete when its output can be checked. Someone with strong field experience may spend longer on statistics; someone with a recent dissertation may need more time explaining observations without specialist shorthand.
| Session | Work | Output to check |
|---|
| 1 | Read the recruitment notice and match qualifications | Correct role, degree evidence and a list of unresolved questions |
| 2 | Work Cases 1–5 without viewing solutions | Unit-labelled calculations and restrained interpretations |
| 3 | Work Cases 6–9 and design a monitoring sheet | Explicit denominators, missingness and method notes |
| 4 | Rebuild Case 10 using a small spreadsheet | No duplicate harvest totals after merging |
| 5 | Practise Cases 11–12 and discuss one real experience | A technical explanation and an accessible explanation |
| 6 | Check sources, documents, route and latest university updates | A travel-ready folder and a concise account of your own work |
For self-review, ask another person to interrupt politely with “What does that number measure?”, “Where did that assumption come from?” and “What would you check next?”. Record which questions reveal a gap. The purpose is not to memorise the model sentences here; it is to make your own reasoning reproducible when the numbers or field conditions change.
✅ Final check before the journey
Revisit the official announcement for changes and keep the linked recruitment details available. Check document names and qualification evidence against the notice, not against an interview-preparation checklist. Where a condition is unclear, obtain the recruiting authority's clarification rather than treating an informal interpretation as permission.
Bring a preparation example you can defend down to its units and original records. In a research conversation, recognising a missing denominator, an unsupported comparison or an uncertain observation is a substantive skill. A graph is useful when it makes that reasoning clearer; it should never make the evidence appear stronger than it is.
