What Is a Scientific Research Paper? A Clear Process and Example
What Is a Scientific Research Paper
A scientific research paper presents a problem that must be answered with data, a testable method, and evidence-based reasoning. A complete paper usually moves from the research question, literature review, model or hypotheses, and data collection method to the analysis results, conclusion, and implications.
If you are writing a quantitative thesis, think of the research paper as a chain of decisions that someone can trace backwards: why you chose the topic, why you selected the variables, why you used a questionnaire, why you selected the sample size, and why the hypotheses were accepted or rejected. SPSS or SmartPLS only performs the calculations. A research paper has value when the question, data, and analysis fit together.
You can read the overview in What is scientific research, then compare it with research subject to identify who or what you are studying.
In quantitative research, data are usually collected through a questionnaire containing items and coded into a .sav or .csv file. You can measure attitudes, perceptions, intentions, or behavior with a Likert scale. Likert introduced a technique for measuring attitudes with a multi-level rating scale (Likert, 1932). From that data, you run the appropriate analyses to answer the research question instead of simply placing a table of numbers in Chapter 4.
Why a Scientific Research Paper Matters in Quantitative Research
A research paper helps you turn a broad topic into a measurable problem. For example, “the effect of service quality on bank customers’ loyalty” is still too broad. You need to identify the dimensions of service quality, the items used to measure loyalty, the target respondents, and the relationships that will be tested.
In quantitative research, the paper usually needs to answer four groups of questions. First, what phenomenon is being studied. Second, how are the related concepts measured. Third, do the data support the relationships in the model. Fourth, what do the results mean in the research context.
A practical process can follow this order:
- Select the problem and define the boundaries of the research field.
- Read the literature to identify a gap or an unanswered question.
- Build the model, hypotheses, and scales.
- Design the questionnaire, pilot the wording, and collect the data.
- Clean the data before running SPSS or SmartPLS.
- Test reliability, scale validity, and the model.
- Write the results, discussion, limitations, and implications.
This order helps you avoid the common mistake of running the data first and then trying to find a suitable question for the numbers already on the screen. If the file is already open, go back and check the variable names, scales, and hypotheses before clicking Analyze.
What Makes a Scientific Research Paper Acceptable
There is no single number that determines whether a research paper is acceptable. The committee usually considers the fit between the question, research design, data quality, analysis, and explanation. The thresholds below are common checkpoints in quantitative research, not permission to remove items mechanically.
| Check | Reference level | Source and scope |
|---|---|---|
| Cronbach's Alpha | 0.7 or above | (Nunnally, 1978), scale reliability |
| Cronbach's Alpha for an exploratory scale | 0.6 or above | (Hair et al., 2010), context must be explained |
| Corrected Item-Total Correlation | 0.3 or above | (Nunnally and Bernstein, 1994) |
| KMO | 0.5 or above | (Kaiser, 1974), condition for considering EFA |
| Bartlett's Test | p-value below 0.05 | (Kaiser, 1974), correlation matrix suitable for factor analysis |
| Eigenvalue | Above 1 | (Kaiser, 1960), reference rule for factor extraction |
| Total Variance Explained | 50% or above | (Hair et al., 2010) |
| Factor loading in EFA | 0.5 or above | (Hair et al., 2010) |
| Outer loading in PLS-SEM | 0.7 or above | (Chin, 1998) or (Hair et al., 2022) |
| Composite Reliability | 0.7 or above | (Fornell and Larcker, 1981) |
| AVE | 0.5 or above | (Fornell and Larcker, 1981) |
| HTMT | Below 0.85, or 0.90 for closely related concepts | (Henseler et al., 2015) |
| VIF | Below 5 | (Hair et al., 2019) |
| R² in PLS-SEM | 0.25 weak, 0.50 moderate, 0.75 substantial | (Hair et al., 2011) |
| Q² | Greater than 0 | (Stone, 1974) or (Hair et al., 2022) |
State clearly which threshold you use at each stage. The same phrase, “acceptable validity,” can refer to different checks. KMO assesses whether the data are suitable for EFA, Cronbach's Alpha assesses internal consistency, and AVE and HTMT are used when evaluating a PLS-SEM measurement model. Do not put CFI, TLI, or RMSEA in a SmartPLS paper simply because you saw them in SEM literature. Those indices belong to CB-SEM according to (Hu and Bentler, 1999).
Sample size also needs a reason. For the number of items, the rule of 5 to 10 observations per item is cited from (Hair et al., 2010). For regression, one reference approach is a minimum sample size of 50 plus 8 times the number of independent variables, according to (Tabachnick and Fidell, 2013). Present the sample-size decision together with the collection context instead of writing only the final number.
How to Read a Scientific Research Paper from the Output
When you open SPSS output, read it according to the question each table answers. Descriptive Statistics shows the characteristics of the sample and variables. Reliability Statistics shows Cronbach's Alpha for the scale. Item-Total Statistics helps you identify items that reduce reliability. KMO and Bartlett's Test, Total Variance Explained, and Rotated Component Matrix support decisions about whether to run EFA, how many factors to retain, and which factor each item converges on.
The table below is illustrative output, not the result of a real study. The table names and columns remain close to the way SPSS displays them so you can compare them with your own file.
| SPSS output table | Column or row to read | Illustrative output | Interpretation |
|---|---|---|---|
| Descriptive Statistics | N, Mean, Std. Deviation | N = 214; Mean = 3.82; SD = 0.71 | There are 214 valid observations for the variable being examined |
| Reliability Statistics | Cronbach's Alpha; N of Items | 0.842; 5 | The scale has good internal consistency under the selected threshold |
| Item-Total Statistics | Corrected Item-Total Correlation | 0.476 to 0.713 | All illustrative items exceed 0.3 |
| KMO and Bartlett's Test | KMO; Sig. | 0.781; 0.000 | The data are suitable for considering EFA under the selected criterion |
| Total Variance Explained | Cumulative % | 61.438 | The retained factors explain approximately 61.438% of the variance |
| Rotated Component Matrix | Factor loading | 0.624 to 0.831 | The illustrative items have factor loadings of 0.5 or above |
The value 0.000 in SPSS's Sig. column should not be written as p = 0.000 in your thesis. You can write p < 0.001 because a p-value is not actually zero in the usual sense. For the Coefficients table, read B, Beta, t, and Sig. together. Sig. supports the decision about statistical significance, while the sign and size of the coefficient show the direction and strength of the relationship in the model.
If you run SmartPLS, the familiar tables are Outer Loadings, Construct Reliability and Validity, Discriminant Validity, Path Coefficients, and R Square. Separate the process into two stages: assess the measurement model first, then assess the structural model. This two-step approach is presented in (Anderson and Gerbing, 1988). When you run bootstrapping, the current PLS-SEM procedure commonly uses 5,000 bootstrap subsamples according to (Hair et al., 2022).
What to Do When a Scientific Research Paper Does Not Meet the Criteria
If the results do not meet the criteria, work from the original data toward the model. First, check missing values, reverse-coded items, out-of-range values, and rows with too many missing responses. A reverse-coded Likert item entered in the wrong direction can reduce Cronbach's Alpha, but deleting the item immediately would hide the cause.
Next, inspect Item-Total Statistics and Rotated Component Matrix. If an item has a low Corrected Item-Total Correlation or loads on multiple factors, reread the item, check the theoretical basis, and only then consider removing it. Record every run, the reason for each deletion, and the number of remaining items. The first pass of EFA seldom produces a tidy table. Removing items still needs an argument, rather than repeated runs until a preferred number appears.
If the scale fails because of the questionnaire design, report this clearly as a limitation of the data. You can rerun the analysis after a reasonable adjustment if you still have time and a sound basis for collecting more data. Once the data have been collected and the questionnaire cannot be changed, keep the result, report it transparently, and propose improvements for future research. Adding rows of data or changing responses to increase Alpha creates a serious risk when you are asked about the original data.
For SmartPLS, check outer loading, CR, AVE, HTMT, and VIF in the correct order. One failing index is not enough to conclude that the entire model is unusable. You need to identify the affected construct, the item causing the problem, and whether removing the item would strip important content from the concept.
Distinguishing a Scientific Research Paper from a Report and a Proposal
A proposal is the design prepared before the research is conducted. It states the problem, objectives, questions, model, planned method, and data collection plan. A proposal may not yet contain actual analysis results.
A report usually describes the activities, results, or situation of an organization during a particular period. It may contain tables of numbers without necessarily building hypotheses or testing relationships between concepts.
A scientific research paper needs to show readers how evidence was produced and tested. A good quantitative paper usually has clearly measured variables, an explained sampling method, a reproducible analysis process, and conclusions that do not go beyond the data. You can also review the content in the Scientific Research category to distinguish related document types.
A research paper also differs from a summary of opinions. Citing many sources does not turn a document into research if it has no question, method, or way to synthesize evidence. At the same time, research does not need a large number of tables. Each table should answer a specific question in the model.
Common Errors
Choosing a topic that is too broad. “Factors affecting consumer behavior” does not tell the reader which consumers, which behavior, or which context. Limit the population, location, period, and dependent variable.
Using terms without defining the variables. State clearly which source defines “satisfaction,” “purchase intention,” or “service quality,” and which items measure each variable.
Collecting data before finalizing the model. The file may then lack a necessary variable or contain questions that do not serve the hypotheses. Before distributing the questionnaire, create a mapping table connecting each hypothesis, construct, item, and questionnaire item.
Treating the p-value as the entire result. The p-value only supports the statistical decision. You must also read the direction of the effect, the coefficient, the model's explanatory power, and the meaning in context.
Mixing criteria between SPSS and SmartPLS. Cronbach's Alpha and EFA commonly appear in an SPSS workflow, while outer loading, HTMT, R², and Q² belong to the PLS-SEM group of indices. Describe the software and model you actually used.
Writing conclusions that are stronger than the data. If the research surveys one group at one point in time, be careful when using the word “causes” for an observed relationship. Your wording should match the design and the scope of the sample.
Frequently asked questions
What sections does a scientific research paper contain?
A quantitative paper usually contains an introduction, theoretical background and model, research methods, results, discussion, conclusion, limitations, and references. Chapter names may differ by institution, but the logic from question to evidence needs to remain intact.
What is an easy-to-understand example of a scientific research paper?
You could study the factors affecting students' intention to use an e-wallet. The model could include perceived usefulness, perceived ease of use, trust, and usage intention, measured with a Likert questionnaire and analyzed with SPSS or SmartPLS.
How large should the sample be for a quantitative scientific research paper?
Sample size depends on the number of items, the number of independent variables, the model, and the sampling method. You can refer to the rule of 5 to 10 observations per item (Hair et al., 2010) or the 50 + 8m formula for regression (Tabachnick and Fidell, 2013), then explain how you applied it to your topic.
What should I do if Cronbach's Alpha is low?
Check coding, reverse-coded items, missing values, and Corrected Item-Total Correlation first. Then review the theoretical content of the item, consider removing it only with a clear basis, and record the decision. For an exploratory scale, Alpha from 0.6 can be referenced according to (Hair et al., 2010), but you still need to explain the context.
Should I use SPSS or SmartPLS for a research paper?
SPSS is suitable for descriptive statistics, Cronbach's Alpha, EFA, correlations, and regression. SmartPLS is suitable when the model includes latent variables, a measurement model, and a structural model under PLS-SEM. Choose according to the model and hypotheses, not because one program produces more tables.
Can I put illustrative output into my thesis?
No. Illustrative output only helps you understand the column positions and interpretation. In your thesis, every number must come from the real data file, be verifiable, and remain consistent across the methods chapter, results chapter, and appendices.
Open your file now and create a table containing the research question, variables, hypotheses, output tables to read, and corresponding criteria. Then run each step on the real data. If you need help running SPSS or SmartPLS and interpreting the results from your own .sav or .csv file, see M4 Analysis at DoThesis.