1 Improper Focus or Format of Title and Abstract

For your article to become a part of global scientific discussion, people need to read it. Your article’s title should be devised so that it interests potential readers to look more closely at the abstract and the article. The title should communicate the topic that the article engages and the approach the authors used to investigate it. It should be descriptive and specific. Some journals require that the type of study, for example, “randomized controlled trial” or “observational study,” be included in the title.

Less commonly, some journals encourage titles that also communicate the primary finding. Some authors, especially in economics and qualitative reports, include clever phrases that capture a central notion of the paper.

Check the specific “Instructions to Authors” for your target journal, and note the permissible length of the abstract and whether they require a structured abstract that uses subheadings or an unstructured abstract without subheadings. As most readers will only read the abstract, it is important to craft your abstract so that it communicates the essential information within the word limit.

The abstract must stand alone. It must tell the reader why the topic is important, what the researchers did, what they found out (the most important results and data from the study), and how these findings contribute to knowledge. Do not cite references or use abbreviations. In an unstructured abstract, methods and results can be merged if this improves the narrative flow.

To develop a structured abstract, follow the sequence in Appendix 9

  • Background: Explains the rationale for conducting the study, that is, why is this study question important? The last sentence in the background should state the objective of the abstract/manuscript. If space limitations are severe, and there is only sufficient space for a single sentence of background, the one sentence should be a statement of the objective.

  • Methods: Summarizes how the study was carried out. Describe the study population, and explain the key techniques used to generate the primary results reported in the article. For each result, check that you have included a corresponding method.

  • Results: Presents the main findings of the study as specific quantitative results. Include raw data with percentages, measures of association, and either confidence intervals or p-values.

  • Conclusion: Explains what these results mean, that is, what their broader implications are for science or for public health. They may support specific public health action or specific next steps in research. This is not a summary. Do not repeat results.

2 Confusing the Role of Introduction, Methods, Results, and Discussion

The standard structure that most journals prefer for a quantitative scientific paper typically includes the introduction, methods, results, and discussion (IMRAD). The IMRAD structure is explicitly recommended in the “Uniform Requirements for Manuscripts submitted to Biomedical Journals” (www.icmje.org). The content of each of these sections is ruled by conventions that help readers quickly understand the article. The introduction explains why the research question addressed in the manuscript is important, the methods describe how the study was conducted, the results present the findings, and the discussion interprets the results.

These conventions allow the reader to quickly look for the information they are interested in if they choose to read selectively (a common practice). See Appendix 6 for more clarification about what to include in each section.

Examples of the error

Alternative, better options

✗

Distracting details in the introduction section

✓

Avoid lengthy background on the general topic. Direct reader toward the research question or problem statement.

✗

Replicating the level of detail of the study protocol in the methods section of the manuscript

✓

Focus on key considerations that allow the reader to understand what was done. Do not spell out methods that generated results that are not presented in the manuscript. Add an appendix if there are technical details that a specialist interested in replicating the work would want to know.

✗

Too many details in the results section

✓

Narrow down on a set of results that are key for the conclusion. Supplementary tables and figures can be included in an appendix.

✗

Rambling unstructured discussion

✓

Follow the outline for a discussion (Appendix 6). Identify and succinctly defend your primary claims and their connection to the published scientific literature.

3 Not Writing the Methods Section in Chronological Order

The methods section typically involves explaining a number of interrelated activities. A common error is a description that jump back and forth among various components. This risks confusing the reader.

The first part of the methods section for a public health paper is commonly a brief description of the study site and population to explain the context. Then the method section explains in detail the study activities that were performed in sequential chronological order (Appendix 1). Chronological order is standard in narrative and so easy for readers to understand. Departing from chronological order risks confusion.

In a protocol, the methods are written in future tense as these are planned activities. In a manuscript, the methods section is always in past tense to tell the reader what the researchers did.

Example of the error

Alternative, better option

✗

Field research workers will make a monthly promotional visit to all intervention households to educate the primary caregivers about benefits of drinking safe water and to promote the proper utilization of intervention products. We will randomly select 1800 households in five regions, 360 in each. In each follow-up visit, the team also collected stored water from all households who received NaDCC tablets and tested for residual chlorine by using a digital colorimeter. We will provide a 10-liter storage container and 33 mg chlorine (NaDCC) tablets in 600 households, only a 10-liter container in 600 households, and the remaining 600 households will receive no intervention.

✓

Put research and data collection activities into chronological order, and use past tense.

• We randomly selected 1800 households in five regions, 360 in each.

• We provided a 10-liter storage container and 33 mg chlorine (NaDCC) tablets in 600 households, only a 10-liter container in 600 households, and the remaining 600 households received no intervention.

• Field research workers made a monthly promotional visit to all intervention households to educate the primary caregivers of the child about benefit of drinking safe water and to promote the proper utilization of intervention products.

• In each follow-up visit, the team also collected stored water from all households who received NaDCC tablets and tested for residual chlorine by using a digital colorimeter.

4 Not Emphasizing Steps Taken to Protect Human Subjects

When describing how ethical considerations were addressed by the study team, a writer can misplace the emphasis by first citing that it was approved by a specific human subjects review committee. This sequencing mistakenly implies that the cornerstone of ethical practice is approval by a review committee.

Instead, lead off this section by describing key activities undertaken by the study team to conduct an ethical study. Only the last sentence, somewhat as an afterthought, should confirm that all of these procedures, which the study team developed and carefully and systematically implemented, were reviewed and approved by an appropriate committee. The idea is that the study team acted as moral agents; they neither delegated the ethical conduct of the study to an external group nor simply sought the permission of some institutional authority.

Example of the error

Alternative, better option

✗

Our study protocol was approved by the ethical review committee of Aga Khan University. Before collecting data, we obtained written informed consent from each adult study participant in the household.

✓

We obtained written informed consent from the adult study participants in each household. The study protocol was reviewed and approved by the ethical review committee of Aga Khan University.

5 Listing Interpretations, But Not Defending One in the Discussion

The role of the discussion section is to explain what the results mean. Sometimes, it is tempting to list all the possible interpretations and “let the reader choose” which is the most reasonable. This is an abrogation of the responsibility of the author. As the person who analyzed the data and knows the study, you are in the best situation to explain what the most likely interpretation is and defend it.

It can be useful to mention other important potential interpretations, but the authors should clearly state what they believe the data means and why. For example, the reader who looks at the following text has no idea which of these interpretations is the most plausible.

Example of the error

Alternative, better option

✗

The difference between the commuting rate and the injury rate may be because men are more likely than women to exhibit risky behavior, particularly not waiting for the bus to stop, hanging on side and climbing on the roof, and running to catch the bus. It could also be explained by a different gender mix on buses during the observation period in these high-risk areas than at other times, or perhaps there are fewer males injured by buses, but this is more than compensated by a disproportionate number of males injured from motorcycles.

✓

Our study found a higher injury rate among men than women. Although we cannot rule out bias in our observations, we did find, consistent with other research on risk taking (ref, ref, ref), that males were more likely to exhibit risky behavior, particularly not waiting for the bus to stop, hanging on side and climbing on the roof, and running to catch the bus.

6 Not Fully Explaining Limitations

The objective of a section on limitations is not to list all aspects of the study that could be improved with infinite money and flawless data collection tools in a perfect world. Instead, this section identifies limitations in the inferences that can be drawn from the study. There are four rules for discussing study limitations:

  1. 1.

    State only the most serious limitations. Don’t list every possible problem. Although a thesis advisor may be interested in them, a journal reader is not.

  2. 2.

    Focus solely on limitations to scientific inference. It is not the role of a limitation section to list all the shortcomings in the study, all of the issues that you would manage differently given another opportunity in a perfect world. It is not a place to talk about your limitations as an investigator, the limitations of the funding agency, or the limitations of the institutional environment where the study was conducted. Rather, the limitation section should be strictly focused on the limitations to scientific inference that can be drawn from the study.

  3. 3.

    Explain the limitation. Don’t just label it. Instead of writing “one of our limitations is selection bias,” discuss how you enrolled subjects and how this may result in an unrepresentative study estimate.

  4. 4.

    Discuss how you interpret the data in the light of this potential problem, for example, “it is unlikely that this procedure substantially affected our results because….”

Example of the error

Alternative, better option

✗

Our study was limited by focusing on only one subdistrict, and so the results might not be generalizable.

✓

Our study focused on only a single subdistrict and so is not representative of the whole country, but the level of economic development, the percentage of the population engaged in agriculture, and the seasonal availability of water is fairly typical of the country.

✗

We did not have sufficient funding to test all of the specimens.

✓

We found no differences between children and adults, but this evaluation was based on a small enough subset of specimens that we had limited power to evaluate modest differences.

✗

Our study was limited by its small sample size

✓

The association of illness with date palm sap consumption was unlikely to be due to chance, but because the outbreak only affected 12 people, we had limited statistical power to identify other potentially important exposures with smaller effect sizes.

7 Writing Generic Recommendations

Only make recommendations that your data can support. They should be applicable to the specific context. For example, avoid suggesting interventions in low-income countries that require a level of national income and government capacity equivalent to that of Western Europe (Error 3.11.1).

Generally, recommendations should not simply call for “more research.” Such generic calls appear self-serving and do not guide the field. By contrast, it is very useful to reflect on what was learned through your study and identify for the global scientific community (including funding agencies) the one or two important research questions that should next be addressed. Don’t provide a laundry list of everything you think should be done. Usually, you should make no more than two practical recommendations.

Recommendations have to be carried out by someone or some agency. Useful recommendations give clear statements about who the actor is, what they should do, and when. Within public health and other applied sciences, scientists are often asked to actively assist in translating scientific knowledge to practical advice for nonresearchers. A mechanism to achieve this is through knowledge translation briefs or one-page summaries of key messages and evidence-based recommendations for action derived from the research results. Aimed at the right institutions and interest groups, evidence-based information and recommendations can inform national policy and programs to address important problems.

8 Presenting New Data in the Discussion

The role of the discussion is to tell the reader what the authors believe the results mean. It is a violation of the standard IMRAD (introduction, methods, results, discussion) format to present new data in the discussion section to support an argument you are trying to make. If the data are important enough to be mentioned in the discussion, then these data should be presented in the results.

9 Reporting the Number of Enrolled Subjects in the Methods

For studies of human subjects, the methods section should describe the enrollment criteria for study subjects and how the investigators trained study workers to apply these criteria to the community where they implemented the study. The methods section should present the planned sample size and the reason the investigators chose this targeted sample size.

The first sentence of the results should describe how many people were approached, how many agreed to participate, how many were enrolled, and how many had sufficient information collected to be included in the analysis. This opening description of the path to the study population provides the underlying denominator for the subsequent analysis.

The number of enrolled subjects is a topic for the results section rather than methods because there is often some difference between what was planned and what was ultimately implemented. The sample size section in the methods describe the planned sample size. The first line of the results describes the sample size that was ultimately realized.

Examples of the error

Alternative, better options

✗

In the methods section: “Altogether 330 questionnaires were taken for study.”

✓

In the methods section: “Study workers visited the study hospital each morning, approached each inpatient who met the enrollment criteria, and invited them to join the study.”

In the first sentence of the results section: “Study workers ultimately approached 349 patients meeting the eligibility criteria; 19 refused and 330 completed an interview.

10 Specifying the Contents of a Questionnaire

Journals generally limit the number of words in a manuscript. This both saves on paper for printed journals and also helps to preserve the time and attention of readers so that they can focus on the most important elements of the manuscript. With all scientists having more articles to read than they have time, succinct writing improves the influence of your article. Listing the various content area queried within a questionnaire used in the study is neither a good use of precious space nor of the reader’s attention. Results from a questionnaire that are relevant to the issues raised in the manuscript will be presented in the results. Readers can infer that this reported information was asked about in the questionnaires. Items that were included in the questionnaire, but are not presented in the manuscript, need not be included in the methods.

By contrast, the physical samples that were collected or the standardized observations that the research team made should be specified.

Examples of the error

Alternative, better options

✗

Between February and March 2014, researchers administered a 45-minute exposure questionnaire among case and control households. The questionnaire collected information about household size, education level, occupation, and age as well as potential exposures related to food consumption, jewelry, kitchenware, housing materials, and farming practices.

✓

Between February and March 2014, researchers administered a 45-minute exposure questionnaire among case and control households.

✗

Field workers collected data using a standardized questionnaire, performed spot checks on hand and domestic hygiene, and collected food samples. The questionnaires included questions on household sociodemographic factors, household assets, drinking water source, sanitation facilities, food storage duration, food reheating history, and food serving practices. After the interview was conducted, field workers performed spot checks on food and hand hygiene practices including container types used for cooking and food storage, container cover status, animal presence in the food storage area, feces in the household compound and food storage area, and cleanliness of utensils and mother and child’s hands.

✓

After conducting interviews using standardized questionnaires, field workers performed spot checks on food and hand hygiene practices including container types used for cooking and food storage, container cover status, animal presence in the food storage area, feces in the household compound and food storage area, and cleanliness of utensils and mothers’ and children’s hands.

11 Naïve Theories of Change

The underlying motivation for public health research is to generate knowledge that can be used to improve health (in contrast to pure academic research that generates knowledge that is interesting but is not primarily justified by its impact on the world). The introduction section of a public health manuscript explains why the question addressed by the manuscript is important, and the discussion explains the implications of this knowledge. Authors of public health research are expected to explain where their research should lead. This often involves making recommendations that are outside the set of issues where an individual researcher has been professionally trained. When a narrowly trained researcher asserts how to bring about change, the suggestions risk being naïve and therefore not useful. By beginning with a more realistic model of how changes occur, a model of change that is informed with some understanding of history, political science, economics, and sociology, then you can make your scientific work more impactful.

Invoking naïve theories of change create two problems. First, they are lost opportunities. Your published manuscript presents an opportunity to make a credible suggestion to an interested audience on the way forward and to have an impact on public health. A naïve theory of change does not provide useful guidance. Second, naiveté undermines the credibility of your voice, and so of the work, and even of your reputation. Readers conclude, “The author is a narrow scientist who does not understand the world.”

11.1 Recommending a Massive Increase in Funding

When we evaluate a public health problem in the context of a low-income country and consider how a similar problem is addressed in a high-income country context, it may seem reasonable to ask that local government authorities take the same steps that have been taken in other places to address the problem. The difficulty with this practical sounding advice is that low-income country government authorities do not have access to the same level of funds available to authorities in high-income countries. Failing to appreciate these political and economic constraints is naïve.

Although you are concerned about the specific public health problem that is the focus of your paper, if everyone working on their area of interest always requests the government to provide more money to replicate what high-income countries do, this becomes an impossible agenda for government to fulfil. Indeed, from the perspective of government decision makers, every sector, including transportation, infrastructure, education, economic development, energy, and health, wants more money. While we may passionately believe that allocating more money to the specific problem that is the focus of our research would create a better society, in general, this is not a particularly useful suggestion. The demands on government funds so exceed the available funds that your recommendation is only one among a never-ending chorus of similar requests.

If we cannot make a particular government sector richer, what should we do? As the expert on the topic of the paper you are writing, reflect upon and propose practical suggestions that are cost-effective or, even better, that cost no money or cost less money than is currently being spent to address the issue. Such recommendations are much more likely to be implemented. Identifying practical solutions to problems, or at least pointing out where we can begin to develop practical solutions, is a centrally important way that scientists can improve public health.

11.2 Ignoring Incentives and Barriers

Public health studies commonly assess knowledge among residents of low-income countries at risk for a particular health condition. Quite unsurprisingly, such studies generally find that these populations have imperfect knowledge about the health condition under study including ignorance regarding the exposures that increase risk. Many scientific authors then call for an intervention to improve the knowledge of the population and to tell them what they should do.

Such recommendations are naïve because they assume that ignorance is the primary determinant of unhealthy behavior. However, there is abundant evidence both in everyday life and in the scientific literature that knowledge is rarely the primary determinant of behavior. Are people obese because they do not know that eating excessive calorie-dense food leads to weight gain? Do people who smoke cigarettes believe they are using a healthy natural product? Do impoverished households in Bangladesh not serve their children fish more frequently because they are unaware that fish is nutritious?

In general, it is much more productive to consider both the incentives people have and barriers that people must overcome to achieve health rather than a primary focus on improving knowledge. People consume excessive calories for a variety of reasons, including the pleasure of eating, emotional connections to food, and acquired habits. People smoke cigarettes because of addiction to nicotine and enjoyment of the smoking ritual. Poor Bangladeshi households do not eat much fish because they do not have the money to pay for it.

Occasionally, improving knowledge can help to facilitate behavior change, but most knowledge interventions fail to improve health. Thus, when a scientist recommends improving knowledge of the at-risk population as the primary intervention, it suggests to readers that the author is not well read (Error 2.1) and is unaware of the strong and consistent evidence that such knowledge interventions nearly always fail to change behavior. We are much more likely to contribute toward improving health by examining more thoroughly and critically the likely determinants of behavior and then suggesting prevention efforts directed at these determinants.

Example of the error

Alternative, better option

✗

Half of the duck flock owners reported disposing of dead ducks by throwing them into adjacent water bodies. Duck owners should be taught that the Food and Agricultural Organization recommends burying carcasses on site to control avian influenza transmission.

✓

Biosecurity interventions that cost-effectively improve duck survival and egg production are much more likely to be adopted. We recommend further research to develop and evaluate interventions that simultaneously improve duck raisers’ profitability and biosecurity.

11.3 Assuming Weak States Can Implement

When working on public health problems, we often consider regulatory approaches to constrain unhealthy practices, for example, requiring factories not to discharge toxic pollution into the environment or to have people who provide housing, food, water, education, and health care meet certain standards. Most people expect that some agency within government should maintain and enforce such regulations. Indeed, most people think of such regulations and their enforcement as one of the primary roles of government. Government officials often describe this as being a central part of their role including developing and adopting many such regulations.

However, many governments have limited capacity to enforce such regulations. For example, in rural Bangladesh, Pakistan, Kenya, or Guatemala, do drivers on rural roads routinely obey the speed limits? Do industrial factories routinely treat their emissions so that they meet government standards and do not pollute air or water? In cities, are building codes enforced? Are criminals who commit financial crimes routinely identified tried, convicted, and punished?

Most low-income countries are what political scientists refer to as “weak states.” These states have limited capacity across a range of functions, including limited capacity to enforce regulations. Weak states are unable to enforce regulations both because of a lack of technical capacity as well as pervasive incentives that undermine enforcement. Government agencies in weak states lack bureaucratic autonomy [14]. This means that enforcement actions are strongly influenced by political actors. Indeed, some political scientists describe many governments of low-income countries as “predatory states.” They argue that such governments exploit their position to extract resources from citizens, without providing the basic functions of government. In weak states, it is relatively easy to pass laws, so it may appear that there is substantial progress, but there is very limited capacity to enforce such laws. There is often no discernible difference in a situation before and after a law is passed.

If authors are working in a weak state but then suggest solutions that presuppose a strong state, for example, Singapore or Germany, perceptive readers conclude that the authors do not understand the context they are working in. This does not improve your credibility as a scientist. Moreover, an approach that requires a strong state will not be effective in a weak state, so the suggestion is not a useful. It does not help move toward a healthier situation. Public health problems generally result from multiple determinants that create an unhealthy situation. Suggesting practical low-cost approaches that can actually improve the situation within the constraints of a low-income context within a weak state is difficult, but this is why deep creativity, hard work, wide reading, conversation with colleagues, and iterative efforts are required tasks of effective public health researchers.

Example of the error

Alternative, better option

✗

The government of Bangladesh should adopt the manufacturing standards promoted by the European Union to ensure a healthy environment.

✓

Stakeholders responding to incentives have produced the current equilibrium of low-priced bricks with high externalities that harm the environment and health in Bangladesh. Transitioning to an alternative equilibrium that generates less damage to health and the environment will require new approaches that alter the underlying incentives.

12 An Insufficiently Focused Introduction

In a standard scientific manuscript, the role of the introduction is very specific: The introduction is not a mini review of interesting themes within the broader field of your study question. The introduction is an argument crafted to persuade the reader of the importance of the study question. An introduction often describes the gap in knowledge that the study addresses and why filling this gap is important for public health or to broaden scientific understanding.

After outlining the introduction, review each assertion and ensure that it directly contributes to a coherent argument that supports the claim that this study question is important. Remove any other points.

Sometimes, understanding the study question requires an explanation of the context of the study or how the present analysis fits within other analyses that have already been published. When this kind of explanation is required so that the reader can understand the relevance and framing of the study question, these elements should also be included in the introduction.

13 Failure to Clarify Key Sample Size Assumptions

Estimating a reasonable sample size for a study requires that the researcher predict what the results will be and then apply the laws of probability to calculate the number of observations that would be reasonably expected to demonstrate results of this magnitude with a low probability that observed associations were only due to chance. The most common version of this error is the failure to specify a predicted outcome or the failure to explain why the predicted outcome asserted by the scientist is reasonable.

Scientists do not conduct studies when they already know what the results will be. The argument “I don’t know what the outcome is; that is why I am conducting this study” is not an acceptable reason for the absence of a defensible argument for sample size. If it were an acceptable argument, it would apply to all studies. Estimating a sample size is an exercise similar to developing a budget for an activity. We cannot foresee all expenses, but we make a judgment based on prior experience to estimate the costs. Similarly, when calculating sample size, we make an estimate of what we think we will find and explain why we think so. Perhaps there will be studies from other regions that have looked at this phenomenon or a similar phenomenon. Alternatively, you may argue that unless a problem is of a certain magnitude, then either it is not important enough or we accept that we won’t have sufficient power to see it. A funding agency will look at the sample size estimate and ask if the money they are investing is likely to achieve the study objectives. They do not want to overpay, but they want reassurance that their money will not be wasted because the sample size was too small to reach the objectives.

A common variation on this error occurs when the primary study outcome is prevalence. The scientist predicts that the outcome will be 50% because they read in a statistics textbook that estimates near 50% require the largest sample size and so they want to be maximally conservative. This is unreasonable because calculating sample size requires an estimate both of the outcome variable and a reasonable level of precision. If the estimated prevalence is 50%, then a study that estimates this prevalence +5% may be reasonable. By contrast, if the estimated prevalence is 3 per 10,000, then assuming an estimated prevalence of 50% +5% would generate a proposed sample size that is far too low.

There is no simple statistical rule that allows a scientist to assert a sample size by a mechanical process that bypasses estimating an outcome and making a reasoned argument for this judgment. When writing a manuscript, the methods section should clarify the assumptions that the scientists originally made of the study outcomes.

Examples of the error

Alternative, better options

✗

We calculated a sample size of 400 based on 80% power and 95% confidence.

✓

We assumed, based on studies of indoor air pollution from cooking (Alam NE 2004, Jones FJ 1997), that children living in a village located within one kilometer of a brick kiln would be at 30% increased risk of pneumonia compared with children who live in villages > 5 kilometers from a brick kiln. We assumed an incidence of pneumonia in this community would be 45 per 100 child years of observation (SE Arifeen 2007), so a sample size of 400 would provide 80% power to detect a difference in groups of 30% at 95% confidence.

✗

We assume that 50% of the poultry workers (~380) will experience at least one episode of symptomatic illness during the study period.

✓

An earlier study found that 44% of adults in an urban community in Dhaka developed a symptomatic episode of influenza-like illness between March and September (MA Azziz 2006). We assumed that 44% of poultry workers would experience at least one episode of influenza-like illness during 6 months of observations.

14 A High-Level Outline That Is Not High Level

The objective of a high-level outline is to sketch out the major components of the manuscript that will support the data analysis included in the framing document (see 1.2.2). The phrase “high level” means that the document outlines the major issues for the manuscript, not all of the details or even all of the components that will be included in the manuscript. The narrative should be no longer than 1500 words; 1200 words is even better.

If your narrative outline is longer than 1500 words, respect the time of your co-authors, and edit it to focus on key ideas prior to sharing it. Bullet points are fine. This is not the time for refining grammar and English language scientific prose.

15 Specifying Software Used for Routine Data Analysis

There are innumerable elements that contribute to a study that supports a scientific manuscript, but the manuscript need not, indeed cannot, specify all of these elements. For example, it is not necessary to mention the brand and version of word processing software that was used to craft the study protocol. It is not necessary to specify the email program that the principal investigator used to communicate with co-investigators or the operating system that was used on the data server. Similarly, if the statistical analysis is routine, the name of the software program used for data analysis need not be specified. Routine analysis includes calculations of means, medians, standard deviations, interquartile ranges, prevalence, incidence, odds ratios, prevalence ratios, risk ratios and their accompanying 95% confidence intervals, simple linear regression, multiple linear regression, and multiple logistic regression.

The underlying guiding principle for writing the methods section is that the methods should be presented in sufficient detail so that other investigators could replicate the study. If the statistical calculations are routine, they could be conducted on any available statistical platform. However, if the researchers used a nonstandard approach that perhaps required special programming in R or a module that is available only in a particular software package, but is not widely available, then it is appropriate to specify the software and procedure that was used. If not, don’t squander the readers’ limited attention with this irrelevant detail.

In the interest of improving validity and reproducibility, more and more journals are requiring scientists to make the primary data and their analytical code publicly available. There are several platforms including the Open Science Framework (https://osf.io/) that permit this. When posting the analytical program used to analyze the data, it is important to characterize the analytical software and the version number within the posting though it need not be mentioned in the manuscript.

Examples of the error

Alternative, better options

✗

We performed descriptive statistics using STATA version 14 software.

✓

We performed descriptive statistics using STATA version 14 software.

✗

We conducted all of our analysis using R version 3.1.0.

✓

Our prespecified analysis plan is available (https://osf.io/6u7cn/).

16 Presenting Rationale in the Last Sentence of the Introduction

In a standard public health or biomedical manuscript, the last sentence of the introduction is a succinct statement of the objective of the manuscript. All of the other text in the introduction is basically an argument on why the objective is important. When an author inserts a sentence or two after the statement of objectives and concludes with a rationale, this confuses the reader because the basic narrative form has been violated. It makes it difficult for someone who is scanning a paper to quickly identify the objective.

Examples of the error

Alternative, better options

✗

This study aimed to identify national-level menstrual hygiene management knowledge and practices among adolescent school girls and facilities provided by their schools. We examine the association of menstrual hygiene management knowledge, practice, and school facilities with absence from school during menstruation. Findings from this study can guide Bangladesh government policy on female education and inform future initiatives to increase female student attendance and school performance.

✓

This study aimed to identify national-level menstrual hygiene management knowledge and practices among adolescent school girls and facilities provided by their schools. We examine the association of menstrual hygiene management knowledge, practice, and school facilities with absence from school during menstruation. Findings from this study can guide Bangladesh government policy on female education and inform future initiatives to increase female student attendance and school performance.

The rationale is an important aspect of the introduction (see Error 3.12). It is simply out of place at the end of the introduction.

Articles published in economic journals are an exception to this rule. In an economics journal, the last sentence of the introduction explains the organization of the paper.