Recommendations for future research are the specific, evidence-based suggestions you make at the end of a thesis, dissertation or paper about what should be investigated next. To write them well, read back through your Discussion chapter, turn every limitation and unanswered question into one actionable line, and support each line with a sentence of justification. Ten focused recommendations beat a page of vague statements, and the whole section takes 45 to 60 minutes once you have read the work through.
One note before we start. This guide is about the Recommendations section of a research study. If you searched for a recommendation letter for a job or a university application, that is a different document with a different structure, and nothing in the method below applies to it.
Table of Contents
- 1What You Need
- 2Step-by-Step: How to Write Recommendations for Future Research
- 3Step 1: Identify the Main Research Gap
- 4Step 2: Connect the Gap to Your Findings
- 5Step 3: Write a Specific Recommendation for Future Research
- 6Step 4: Use Future Tense and Research Language
- 7Step 5: Prioritize and Group the Recommendations
- 8Step 6: Review the Recommendations Against the Study
- 9Common Mistakes
- 10Frequently Asked Questions
- 11How many recommendations for future research should I include?
- 12What is the difference between recommendations for future research and recommendations for practice?
- 13Can I use the limitations of my study to write future research recommendations?
- 14Should future research recommendations use words such as should or must?
- 15What can I recommend if my study did not find a significant result?
- 16How specific should a future research recommendation be?
- 17Conclusion
What You Need
You cannot write this section from memory. Everything you need already exists in the study you have already written, and the drafting goes much faster once it is in front of you in one place.
- Your aims and objectives. The original research questions are what every recommendation has to stay connected to.
- Your methodology. Design, sampling approach, sample size, measures and analysis technique. This is where most limitations come from.
- Your key results. The findings that mattered, including the ones that surprised you or came out non-significant.
- Your Discussion chapter. The single most important document. Recommendations must be traceable to something you argued here, not something new.
- Your limitations list. Every constraint you admitted, from recruitment difficulty to a single-site design.
- Unresolved questions and practical recommendations. Anything your data raised but could not answer, and any implication you noted for policy or practice.
Bring a blank page, not a blank document with headings. The first draft of this section almost always looks like a list before it looks like prose.
Step-by-Step: How to Write Recommendations for Future Research
Step 1: Identify the Main Research Gap
A research gap is something that is not yet known, not fully described, inconsistent across studies, or hard to apply in a real setting. A topic your study did not examine is not automatically a gap, and treating it as one is the most common reason a supervisor sends a draft back.
The test is simple. Ask whether a competent researcher could reasonably disagree about what is still unknown. If the answer is yes, you have a gap. If the answer is that nobody has looked at that topic at all, you have a curiosity, and a curiosity belongs in a sentence of context, not in your recommendations list.
Phrase each gap in one of four ways: what remains unknown, what remains incomplete, what is inconsistent, or what is difficult to apply. That phrasing is what turns a limitation into a researchable question rather than an apology.
Step 2: Connect the Gap to Your Findings
Every recommendation must attach to something concrete in your own study. The anchor is usually a result, a limitation, an unexpected finding, a difference between groups, a measurement problem, or a condition specific to your setting.
Map each one directly. This is the exercise that stops recommendations drifting into territory your data never covered.
| Limitation in your study | Matching recommendation |
|---|---|
| Single-site design, one regional hospital | Replicate the study across multiple sites to test whether the relationship between workload and retention holds in other settings |
| Cross-sectional design, data collected at one time point | Follow the same participants longitudinally to establish whether the relationship is causal and whether it changes over time |
| Self-reported measures of motivation | Repeat the analysis using validated objective or observer-rated measures to test whether self-report bias explains the finding |
| Small sample drawn from one cohort | Replicate with a larger, demographically broader sample drawn from two or more institutions |
| Quantitative design with no qualitative explanation | Add a qualitative strand, such as interviews with a subset of participants, to explain why the measured effect occurred |
Students who pair each stated limitation with a matching recommendation tend to receive fewer revision requests, because the link is visible without explanation. If you cannot point to the anchor sentence, the recommendation is not ready to write.
Step 3: Write a Specific Recommendation for Future Research
The formula is short: something + someone + needs to do something, stated in the future tense and scoped to a population, setting or method. If your sentence has no actor and no action, it is a sentence about the world, not a recommendation.
Here is the difference in practice.
Weak: More research is needed in this area.
Strong: Future studies should investigate whether the relationship between clinical placement length and final exam performance persists when cohort size is controlled for, since the present study could not separate the two effects.
Two more pairs to work from.
Weak: Further qualitative research would be beneficial.
Strong: A qualitative interview study with 20 first-year students in the same department would help explain why the reported drop in attendance clusters in the six weeks before assessment weeks.
Weak: The sample was small, so more studies are required.
Strong: A multi-site replication with a sample large enough to detect a difference of the size observed here is needed before any confidence interval is placed on the effect.
Notice what the strong versions share. Each names a population or setting, names a method, and states what the follow-up would settle. None of them claims the follow-up will prove anything.
Step 4: Use Future Tense and Research Language
The grammar carries the honesty. Research has not happened yet, so the verb has to be an open one, and the sentence has to leave room for a null result.
Verbs that fit: investigate, examine, test, compare, replicate, explore, evaluate, establish, extend, refine. Each one leaves open whether the answer will be yes.
Verbs to avoid: prove, demonstrate, confirm, show, will establish, will improve, will lead to. A future study can fail to replicate your finding, and a sentence that assumes success is a claim your evidence cannot support.
Compare the two. “Future research should test whether the intervention reduces readmission within 30 days” is defensible. “Future research will demonstrate that the intervention reduces readmission” is a promise you have no standing to make, and examiners read it as overreach.
Cautious framing also means naming the reason behind the suggestion. “A replication is recommended because the present design cannot separate the effect of setting from the effect of staffing” tells the reader why this gap matters. “Further research is recommended” tells them nothing.
Step 5: Prioritize and Group the Recommendations
Three to six recommendations is typical for a dissertation Chapter Five, and a journal article usually justifies one to three. Examiners read hundreds of these, so an unranked list of twelve reads as a list of twelve because the writer had not decided what mattered.
Order them by strength of evidence in your own study. A limitation that directly qualifies your main finding comes first, because the recommendation is the reader’s route back to trusting that finding. A peripheral curiosity comes last, if it survives at all.
Group by whichever logic suits your study, and pick one: by research gap, by population, by method, by theory, or by priority. Mixing two schemes in one list is the usual reason a section reads as disorganised.
Then tier by audience. Recommendations aimed at other researchers ask for replication, extension or comparison. Recommendations aimed at policy makers ask for a change in decision-making. Recommendations aimed at practitioners ask for a change in day-to-day work, and those belong under implications for practice, not here. Keeping the two apart is a mark of a clear thinker.
Delete anything that restates a suggestion already made in different words. Two sentences recommending longitudinal follow-up in slightly different phrasing look like padding, and supervisors notice.
Step 6: Review the Recommendations Against the Study
Read the finished section against the study, not against your memory of it. Seven checks catch nearly every problem.
- Traceability. Can you point to the Discussion paragraph each recommendation came from? If not, cut it or move the argument into the Discussion first.
- Alignment. Does each recommendation serve one of the original research questions?
- Specificity. Does it name a population, a setting, a method or a measure?
- Feasibility. Could a doctoral student in your field do this in three years with ordinary resources?
- Consistency. Does it match your conclusion rather than contradict it?
- Ethics. Would it need approval you cannot obtain, or involve a population you could not recruit?
- Non-duplication. Is it distinct from the implications for practice and the conclusion?
The traceability check is the one that catches most marks lost. If a recommendation introduces literature, a method or a claim that appeared nowhere in the Discussion, it is new material, and new material does not belong in this section.
Common Mistakes
1. Repeating the conclusion. The conclusion states what you found. The recommendation states what should be done next. If a sentence would still make sense in a paper with no study behind it, move it.
2. Suggesting topics the study never touched. “Research should examine the broader socioeconomic causes of attainment” may be a fine research question and an unearned recommendation. Anchor it first, or drop it.
3. Confusing limitations with recommendations. A limitation is a statement about your study. Its paired recommendation is a statement about a future one. Writing both as limitations leaves the reader with problems and no direction.
4. Proposing untestable work. “More research is needed to improve the situation” cannot be designed, funded or completed. Add a population, a method and an outcome and it becomes researchable.
5. Making claims about impact. Statements about what a finding will change in policy or practice are implications, and they belong in a separate section with a separate audience.
6. Writing them in the imperative. “Researchers must adopt better measurement” is a command to an audience you do not control, and it reads as a claim about the field. “Researchers could adopt” is the register examiners expect.
7. Citing a source inside a forward-looking recommendation. A recommendation points forward, so it normally carries no citation. If you name a specific framework or instrument, cite it, otherwise leave the reference out.
8. Producing generic passages from an AI writing tool. This is the fastest-growing version of the vague-recommendation problem. Generated text defaults to field-wide generalities because that is what a language model has most of. It is easy for supervisors to spot, and it costs you a revision round. Write from your limitations list instead.
Two situations trip people up repeatedly. When results are non-significant, the honest recommendation is a replication with a different design or a larger sample, plus a statement that the current study cannot distinguish a true null from an underpowered test. When findings conflict with earlier work, a comparative replication across settings is a stronger recommendation than any attempt to explain the conflict speculatively.
By discipline, the mechanics stay identical. In health and nursing, the missing piece is usually a different setting, cohort or follow-up period. In education, it is often a subject or a school type outside your sample. In business, it is frequently a second sector or a longitudinal design behind a single cross-sectional survey. In environmental science, seasonality and site conditions tend to dominate. In social science, population and measurement carry most of the weight. In each case the pattern is the same: name the constraint, propose the design that removes it.
Frequently Asked Questions
How many recommendations for future research should I include?
Three to six is the usual range for a dissertation Chapter Five, and one to three for a journal article. Cover every limitation that materially qualifies your findings, then stop. Examiners reward prioritised suggestions, not volume, and padding an unranked list to twelve points reads as a writer who has not decided what matters most.
What is the difference between recommendations for future research and recommendations for practice?
A recommendation for future research asks what should be investigated next, and it is addressed mainly to other researchers. A recommendation for practice asks what should change in day-to-day work, and it is addressed to practitioners and organisations. Mixing the two is one of the most common faults in Chapter Five, so keep them in separate sections.
Can I use the limitations of my study to write future research recommendations?
Yes, and most examiners expect it. The limitations section tells you exactly where your design stopped short, and each limitation pairs naturally with the design that would push past it. Map them one to one in a simple table, and you have the backbone of the section before you have written a sentence.
Should future research recommendations use words such as should or must?
Use should sparingly and avoid must entirely. Should reads as mild direction when a recommendation is genuinely well supported, while must implies an obligation and a claim about the field that you cannot back with your own data. In practice, investigate, examine, test, compare and replicate carry the meaning without the overreach.
What can I recommend if my study did not find a significant result?
A non-significant result still supports recommendations, as long as you describe it accurately. State plainly that the study cannot distinguish a true null from an underpowered test, then propose the design that would separate them: a larger sample, a different measure, or a design with better control of confounding. Never imply the next study will find an effect.
How specific should a future research recommendation be?
Specific enough that a researcher could design the study tomorrow. Name the population or setting, the method or measure, and the question being settled. The check is whether a reader could tell whether the follow-up succeeded. If the sentence would apply to any study in your field, it is not specific enough.
Conclusion
Start from your clearest limitation, not from a blank page. Read the limitations you wrote in the Discussion, pick the one that most qualifies your main finding, and turn it into a single sentence naming who should investigate what, where, and with which method.
Work outward from there, one recommendation at a time, keeping each one traceable to something you actually argued and phrased in language that leaves the outcome open. Three well-anchored suggestions will always hold up better than a page of generalities, and they will still read as yours when your supervisor gets to them.
