Eleven R&D Management Papers Compared
Eleven papers read for what kind of evidence each one is, not for what it says about R&D. The set runs from a formal model with no data in it to a single insider case study with no method section — and the difference determines what any sentence you write from them is allowed to claim.
Eleven papers, read for how they were built
The papers are referred to as R1 to R11, or by subject. This page is about them as research: what kind of thing each one is, what evidence it rests on, and what a sentence citing it is entitled to claim. What they actually say about selecting, valuing, accelerating, terminating and reviewing R&D projects is elsewhere in this library, condensed on the R&D project management quick reference.
Type, venue and method of each
THE ELEVEN, AS RESEARCH OBJECTS
| ID · year | Type and venue | Method and evidential basis |
|---|---|---|
| R1 · 1991 | Theoretical economics paper in the annual-meeting proceedings issue of a general economics review. Proceedings papers of this kind are short, invited and not fully refereed | A formal microeconomic model presented in outline, with definitions, a metric on information structures and four value equations. No empirical data of any kind — not one number in the paper is a measurement |
| R2 · 2000 | Methods article in a practitioner journal for research and technology management | Derivation of a valuation model plus two worked examples — one adapted from a textbook case, one borrowed from prior authors. No original data |
| R3 · 1984 | Review and tutorial article in a practitioner journal, condensed from the author's book | A survey of a financial toolkit, plus two regressions computed on one division's projects from a single year. The paper makes no generalisation claim for the fitted coefficients |
| R4 · 1989 | Review and synthesis article in a practitioner journal | A causal theory of the time–cost curve, plus three elasticities recomputed from prior studies into one comparable metric, plus a hypothetical example. No original data at all |
| R5 · 2002 | Short empirical article in an applied-management department of a practitioner journal | Questionnaire study: 375 distributed, 205 returned, 135 valid, covering 217 projects across 17 industries in one metropolitan area, 1997 to 1999. Stepwise discriminant analysis run separately at three evaluation points |
| R6 · 1986 | Practitioner-journal report of a large multi-firm academic research programme, one of a series drawing on the same database | 211 projects in 21 companies across 4 lines of business, instrumented by 8 questionnaires totalling over a thousand questions. Data are retrospective by design; the database had been built over nine years |
| R7 · 1983 | Practitioner-journal report of a firm-level survey, adapted from a longer academic marketing article published the previous year | 103 firms, 69 percent effective response, drawn from a standing panel of industrial firms known to be active in new products. Pearson correlations and analysis of variance on cross-sectional data |
| R8 · 2003 | Mixed-methods study plus a normative maturity model, in a practitioner journal | 63 survey responses collected at two executive training events, plus 27 interviews at 13 multinational companies between 1997 and 2001. The paper states the sample is too small for in-depth statistical evaluation |
| R9 · 1994 | Methods and tools exposition in a practitioner journal — a consultant's synthesis illustrated with anonymised client displays | No method section. The evidential basis offered is an experience base of hundreds of projects for dozens of companies over a quarter-century, plus four cited references |
| R10 · 2006 | Regulatory and legal practice review in a specialist law quarterly, written by two case handlers of the authority in a personal capacity | Descriptive and doctrinal synthesis of ten years of that authority's own case handling, with 57 footnotes and no tables or figures. Roughly 65 individual assessments since 2001 form the caseload described |
| R11 · 2006 | Single-organisation case study in a practitioner journal, by an author formerly in the studied organisation's technology office and a serving executive of its parent group | No stated method. No data collection, sampling, interview protocol, analysis or observation period is described, although organisational performance counts are reported |
Nine of the eleven are practitioner-facing. Two are academic: R1 and R10. Practitioner journals in this set publish frameworks and cases; academic venues publish models and doctrine; the two literatures rarely cite each other.
The spread, from a formal model to a single case
Laid out by how much of the world each paper touches, the set spans almost the full evidential range available to a management discipline — and the spread is wider than most reading lists a project manager will encounter.
- No data at all (R1, R4)
- Worked demonstrations (R2)
- One division or one organisation (R3, R11)
- One authority's caseload (R10)
- Multi-firm surveys (R5, R6, R7, R8)
At one end R1 contains no measurement whatsoever; its propositions are conditional results of a stated model, and treating them as findings would be a category error. At the other end R6 instruments 211 projects with eight questionnaires and reports null results as carefully as positive ones. In between sits R10, which is neither empirical nor theoretical — it reports what a decision-maker actually did, which is a third kind of evidence entirely.
What each kind of paper can and cannot evidence
This is the part that matters when you write. The kind of paper sets the strongest verb you are allowed to use about it.
EVIDENTIAL RANGE BY KIND OF PAPER
| Kind | Can evidence | Cannot evidence | Verb to use |
|---|---|---|---|
| Formal model (R1) | That a conclusion follows from stated assumptions; that a common way of framing a question is ill-posed; structural results such as the failure of concavity or the non-existence of a pure-strategy equilibrium | That anything happens in the world. No magnitude, no frequency, no rate, no effect size — and the paper contains no numbers to misquote | "argues", "shows that under its assumptions" |
| Proposed tool or model (R2, R4, R5's method) | That a method is internally coherent, computable from data a manager plausibly has, and able to handle a case a rival method cannot | That applying it improves outcomes. None of these papers compares users against non-users, or reports a result from a decision made with the tool | "proposes", "offers" |
| Multi-firm empirical study (R5, R6, R7, R8) | Associations within a defined sample; distributions and dispersions; relative magnitudes between groups; null results, which are often the most valuable output | Causation, on cross-sectional retrospective self-report data; population estimates where the sample is self-selected or the success rate is set by design | "found, in a sample of…" |
| Synthesis of existing tools (R3, R9) | A map of the option space; the rationale practitioners give for each option; a common metric that makes previously incomparable studies comparable | The efficacy of any tool in the map. A synthesised metric is only as good as the studies feeding it, and R4's headline range rests on three point estimates | "collects", "sets out" |
| Single case study (R11) | A mechanism in enough detail to be copied and tested; an existence proof that a configuration is workable; the sequence in which moves were made | Effect size, attribution or transferability. With insider authorship, no control and no counterfactual, the causal claim is assertion | "describes", "reports" |
| Regulatory analysis (R10) | What the rules were, how a decision-maker applied them, which test actually decided outcomes, and which was assumed rather than performed | Whether the rules were effective, efficient or good policy — no outcome is measured. And nothing about the current legal position, since the framework expired | "records", "documents" |
What they disagree about
A striking feature of this set is how often a paper's central move is to contradict something the reader is assumed to believe. That makes the set unusually useful for a literature review, because disagreement is what a review is supposed to map.
Against received practice
- R7 disputes a widely quoted new-product failure rate of 50 to 90 percent, and attributes it to speculation and studies of questionable merit rather than to data.
- R4 argues acceleration is convex, not proportional — the same increment of time costs progressively more the closer you already are to the minimum.
- R5 argues termination should not rest on managerial judgement or on single indicators, because indicators change meaning as a project advances.
- R2 argues the most widely accepted option-pricing model is the wrong instrument for R&D, because R&D cannot supply the parameters it needs.
- R6 finds that clear, widely agreed goals at initiation do not predict success — only the trajectory of goal clarity does.
- R10 records that starting work before aid approval does not by itself defeat the incentive-effect test.
Against each other
- R3 says appraise R&D by discounted cash flow and speak the finance function's language. R2, sixteen years later in the same venue type, says discounted cash flow is structurally wrong for R&D.
- R1 says both of them are asking the wrong question, because projects differ in what they would let you learn, not merely in risk and return.
- R6 and R7 report success rates that look comparable and are not: one counts a project successful only if it succeeded both technically and commercially, the other measures commercial profitability on fully developed products.
- R11 reports that a move toward long-term basic research made its centre worse, not better — the opposite of the standard prescription for a lab captured by its divisions.
Two of these are genuinely instructive as method. R7's dispute is a definitional argument, not a data argument: its lesson is to interrogate the denominator before accepting any rate. And the R2 versus R3 disagreement is not resolvable by evidence, because neither paper tests its recommendation against the other's — it is a disagreement about what an R&D project is. Leave both standing when you write, and say why they cannot be reconciled from these sources. That is what critiquing a journal article asks you to do.
None of them declares a paradigm
Not one of the eleven states an ontological or epistemological position. No paper says whether it takes reality to be single and measurable or socially constructed, or on what basis it holds its claims to be knowledge. Several do not even say how their data were gathered.
That count is consistent with the ten works profiled earlier in this library and takes the running total to twenty-one out of twenty-one. The tension between that record and a teaching source which treats the philosophical layer as fundamental is worked through on declaring a research paradigm — or not. The short version: publication convention in these venues does not require the declaration, and your examiner probably does. Do not resolve the conflict by copying the papers.
Using a set like this in a literature review
What to do with eleven mixed-kind papers
- Sort by kind before you sort by topic. A model, a survey and a case study cannot be summarised in the same sentence.
- Attach the sample to every figure you quote — the firm count, the project count, the industry, the years, the country-level scope where it is stated.
- Match the verb to the kind. "Proposes" for a tool, "found, in a sample of" for a survey, "describes" for a case, "records" for a regulatory review.
- Keep the contradictions visible and explain what kind of disagreement each one is — definitional, evidential, or a difference of framing.
- Say what the set does not cover. This one has nothing systematic on people management, intellectual property or collaborative innovation, and nothing published after 2006.
- Write the sentence that says the set was assembled as a reading list and is not a systematic search. See stating limitations and contribution.
What to carry forward
- The kind of paper sets the strongest verb you may use about it. Match them deliberately.
- Nine of these eleven are practitioner-facing; two are academic. Frameworks are proposed in this set far more often than they are tested.
- R1 contains no data and R11 contains no method. Neither is worthless — each simply carries a different kind of claim.
- Multi-firm surveys here evidence association within a stated sample, not causation and not population rates.
- Where papers contradict each other, name the kind of disagreement rather than picking a winner.
- Twenty-one examined works, twenty-one with no declared paradigm — and this set is a reading list, not a survey of the field.
Frequently asked questions
Does it matter that most of these papers are not peer-reviewed?
It matters for what you can claim from them, not for whether you may cite them. Practitioner venues in this set publish frameworks, cases and professionally edited empirical reports, which are legitimate sources for what practitioners propose and observe. What they mostly do not carry is independent verification, so a framework from one of them is a proposal with a worked illustration, not a tested instrument.
How should I cite a paper that has no method section?
State that it has none. For a case study with insider authorship, cite the mechanism it describes and say explicitly that no data collection, sampling or analysis is reported, so its causal claims are assertions. For a consultant's synthesis offering an experience base rather than a method, cite it for the taxonomy of tools and chase the primary sources for any number.
Can a paper with no data be worth citing?
Yes, for a different purpose. A formal model can establish that a conclusion follows from stated assumptions, or that a widely used framing is ill-posed — which is precisely what the 1991 economics paper in this set does to the aggregate view of R&D spending. Quote its propositions as conditional results of its model, and never as findings.
Two of these papers give different new-product success rates. Which is right?
Neither, in the sense the question implies. They measure different quantities on different denominators: one requires both technical and commercial success at project level across four lines of business, the other measures commercial profitability at firm level on fully developed products ready for commercialisation. Report both with their definitions attached rather than choosing between them.
What does the twenty-one out of twenty-one paradigm count mean?
It means that across every published work this library has examined — five in one batch, five in another and these eleven — not one declares an ontology or epistemology. It is an observation about twenty-one works, not a statistic about published research. It is useful mainly as evidence that publication convention and thesis convention differ on this point.
Is this a good set of papers for a literature review on R&D project management?
It is a usable starting set with a wide evidential range and several live disagreements, which is more than many reading lists offer. It is not a systematic or representative survey, its coverage is uneven by construction, and nothing in it postdates 2006. Treat it as a seed for a proper search rather than as the search itself.
References and source attribution
- Eleven copyrighted journal articles on R&D project management, supplied as a reading set assembled by a student for a literature review and profiled for this library. Front matter, abstracts, framework sections, tables and figures were read; article bodies were not reproduced, and all content here is paraphrase. The set comprises: a theoretical economics paper in an annual-meeting proceedings issue (1991); practitioner articles on option value (2000), financial appraisal (1984), acceleration cost (1989), termination decisions (2002), project success and failure (1986), new-product success rates (1983), post-project reviews (2003) and decision-analysis tools (1994); a regulatory practice review in a specialist law quarterly (2006); and a single-organisation case study (2006).
- The five journal articles profiled earlier in this library for their research designs, and the five research works profiled before those. Together with the eleven examined here they make up the twenty-one works referred to on this page. None was supplied in full and all are described from profiles rather than reproduced.
- Figures quoted within these eleven papers from other sources — the disputed new-product failure rate, a benchmarking study of 79 R&D organisations, three acceleration elasticities recomputed from prior studies, and a consulting review of complex-project cost overruns — were not supplied to this library and were not independently examined.
- Supplied teaching source for this library (research methods and research process materials). Used here for the criteria by which a source's evidential weight is assessed, and for page conventions and voice.
Suggested questions for Ask KEVOS
- Sort a set of papers I am reading by kind, and tell me the strongest verb I may use about each.
- Draft the limitation paragraph explaining that my reading set was not assembled by systematic search.
- How do I write up two papers that contradict each other without picking a winner?
- What should I check first in a practitioner-journal article that has no method section?
- Explain the difference between what a formal model and a single case study can each evidence.
- Which claims in an R&D management framework need re-testing before I apply them in my organisation?