What Is Quantitative Research? How to Apply It and Read the Results
What Is Quantitative Research
Quantitative research uses numerical data to describe a phenomenon, measure concepts, and test relationships between variables. You usually collect the data through a questionnaire with coded response options, then enter it into a .sav or .csv file for analysis in SPSS, SmartPLS, AMOS, JASP, or R.
A quantitative study usually follows this sequence: define the research question, build the model and hypotheses, design the scale, collect the data, clean the data, run the analysis, and interpret the results. For example, if you want to test whether service quality affects customer satisfaction, service quality is the independent variable, customer satisfaction is the dependent variable, and the observed variables are measured through statements in the questionnaire.
Quantitative results may include frequencies, means, Cronbach's Alpha, KMO, factor loading, regression coefficients, p-values, R², or path coefficients. Each statistic answers a different question. You can also read the overview at scientific research and what is scientific research if you are still choosing a research direction.
Why Quantitative Research Matters in Quantitative Research
Quantitative research lets you turn a research idea into variables that can be observed and tested. Instead of saying only that “students care about online learning,” you can measure their level of agreement with items such as “the platform is easy to use,” “the content is relevant,” and “I intend to continue using it.” You code the responses, for example on a 1 to 5 Likert scale, so that you can calculate and compare them.
In a thesis, this method usually helps you answer four groups of questions. First, what are the characteristics of the survey sample. Second, does the scale have adequate reliability and validity. Third, do the independent variables explain the dependent variable. Fourth, are the research hypotheses supported or rejected based on the coefficients and p-values.
You need to distinguish the method from the tool. The questionnaire is the data collection instrument. SPSS is commonly used for descriptive statistics, reliability testing, EFA, correlation, and regression. SmartPLS is suitable for PLS-SEM, where you assess both the measurement model and the structural model. AMOS is commonly used for CB-SEM, while JASP and R offer advantages in software cost and reproducibility but require you to learn an interface or write code.
Quantitative methods do not make a conclusion correct by themselves. If the questionnaire measures the wrong concept, the sample is unsuitable, or you remove items only to make the output look better, the final numbers still carry risk. Record the data version, the reason for removing each item, and the order of your analysis steps.
How Much Quantitative Research Is Acceptable
There is no single number that determines whether an entire quantitative study is acceptable. The threshold depends on the type of analysis, the research objective, the scale context, and your supervisor's guidance. The table below summarises thresholds commonly used in theses and gives you a source to cite when you need to explain them.
| Check | Common threshold | Canonical source |
|---|---|---|
| Cronbach's Alpha | 0.70 or above | (Nunnally, 1978) |
| Alpha for a new or exploratory scale | May be 0.60 or above | (Hair et al., 2010) |
| Corrected Item-Total Correlation | 0.30 or above | (Nunnally and Bernstein, 1994) |
| KMO | 0.50 or above | (Kaiser, 1974) |
| Bartlett's Test | Sig. below 0.05 | (Kaiser, 1974) |
| Eigenvalue for factor extraction | Above 1 | (Kaiser, 1960) |
| Total Variance Explained | 50% or above | (Hair et al., 2010) |
| Factor loading in EFA | 0.50 or above | (Hair et al., 2010) |
| Outer loading in PLS-SEM | 0.70 or above | (Chin, 1998) |
| Composite Reliability | 0.70 or above | (Fornell and Larcker, 1981) |
| AVE | 0.50 or above | (Fornell and Larcker, 1981) |
| HTMT | Below 0.85, or 0.90 for closely related concepts | (Henseler et al., 2015) |
| VIF in PLS-SEM | Below 5 | (Hair et al., 2019) |
| R² in PLS-SEM | 0.75 substantial, 0.50 moderate, 0.25 weak | (Hair et al., 2011) |
| Q² | Above 0 | (Stone, 1974) |
Do not separate these thresholds from their context. For example, an Alpha that is too high can sometimes indicate that the items repeat the same content. A low factor loading is also not an automatic reason to remove an item if that item has theoretical meaning and the decision has not been considered in the model.
If you run CB-SEM, you can use CFI, TLI, RMSEA, and SRMR according to the criteria in (Hu and Bentler, 1999). These indices belong to CB-SEM. Do not present them in a SmartPLS guide as though they were the main criteria for PLS-SEM.
How to Read Quantitative Research in the Output
Suppose you are running a linear regression in SPSS and want to test the effect of service quality on satisfaction. In the Coefficients table, read the B or Beta coefficient, the t value, and the Sig. column. A positive coefficient indicates a positive relationship in the model. Sig. is the p-value used for the statistical decision, but you still need to examine the sign of the coefficient and its theoretical meaning.
The following is illustrative output, not the result of a real study. The column names follow the way SPSS usually displays them, so you can compare them with the file currently open on your screen.
| Model | Unstandardized Coefficients B | Std. Error | Standardized Coefficients Beta | t | Sig. |
|---|---|---|---|---|---|
| (Constant) | 1.024 | 0.286 | 3.58 | 0.001 | |
| CLDV | 0.548 | 0.071 | 0.612 | 7.72 | 0.000 |
In this example, CLDV has a Beta of 0.612 and a Sig. of 0.000 as displayed after SPSS rounding. In your thesis, write p < 0.001 rather than interpreting the result as p being exactly 0. The positive coefficient indicates that CLDV has a positive relationship with satisfaction in the illustrative sample. Your final conclusion must still be based on your model, sample size, assumption checks, and variable coding.
If you run EFA, open KMO and Bartlett's Test, Total Variance Explained, and Rotated Component Matrix in sequence. KMO indicates the overall suitability of the data for factor analysis. Rotated Component Matrix shows which factor each item loads on most strongly. Check the factor loadings, cross-loadings, and theoretical content instead of selecting only the column with the largest number.
If you run Cronbach's Alpha through Analyze > Scale > Reliability Analysis, the Reliability Statistics table reports the overall Alpha. The Item-Total Statistics table reports Corrected Item-Total Correlation and Cronbach's Alpha if Item Deleted. The latter is supporting information, not the only reason to remove an item.
What to Do When Quantitative Research Does Not Meet the Thresholds
Work from the original data toward the model. First, check whether a variable was entered with the wrong code, whether reverse coding has not been handled, or whether it contains too many missing values. A Likert item entered in reverse, such as 1 instead of 5 and 5 instead of 1, can reduce its correlation with the other items.
Next, examine the descriptive statistics and unusual cases. Do not delete respondents in bulk just to increase Alpha or R². You need a clear rule, such as a questionnaire with many unanswered items or a response pattern in which one respondent selects the same option throughout the questionnaire. Record the rule before rerunning the analysis.
If Corrected Item-Total Correlation is low, reread the item and check its coding before considering removal. If an item genuinely does not fit the concept, you can remove it and rerun the analysis, but report the number of processing rounds. If EFA does not produce the groups expected by your model, examine factor loadings, cross-loadings, and the extraction method, then compare the result with the theoretical foundation.
When regression assumptions are not met, check the residual plots, VIF, Durbin-Watson, and influential cases. For regression, the sample-size rule n of at least 50 + 8m is commonly cited from (Tabachnick and Fidell, 2013), while a Durbin-Watson value between 1 and 3 is commonly used when assessing residual autocorrelation according to (Field, 2013). These thresholds do not replace inspection of the actual data.
If SmartPLS reports a high VIF, check whether the predictor variables are too similar in content and whether the model includes unnecessary paths. In PLS-SEM, the assessment should move from the measurement model to the structural model, in line with the guidance of (Hair et al., 2022). Another software package may produce a cleaner table, but it cannot repair a theoretical model or questionnaire that was designed incorrectly.
Distinguishing Quantitative Research from Qualitative Research
Quantitative research prioritises numerical data, measured variables, survey samples, and hypothesis testing. Qualitative research usually focuses on meaning, experience, and how participants interpret a phenomenon. You can read what is qualitative research to identify the differences in objectives and data types.
In a quantitative study, research questions often take the form “How does X affect Y?” or “How large is the difference between groups?” You need to turn concepts into scales and decide on the planned analysis in advance. With a qualitative approach, questions are usually more open and the data-processing procedure is different. DoThesis focuses on a quantitative workflow using questionnaires, .sav files, .csv files, SPSS, and SmartPLS.
These two approaches should not be distinguished only by whether a questionnaire is used. A questionnaire made up of open-ended questions does not by itself constitute a complete quantitative study. Conversely, a quantitative study can use secondary data if the data is coded into numerical variables and fits the analysis model.
Common Mistakes
The first mistake is using “quantitative” to mean any study with many participants. The number of respondents is only one part of the design. You also need clearly measured variables, a sampling approach, a data-cleaning procedure, and an appropriate analysis.
The second mistake is copying a threshold from an online article without checking the source. When you state that Alpha is 0.70 or above, you can cite (Nunnally, 1978). When you state that Corrected Item-Total Correlation is 0.30 or above, the appropriate source is (Nunnally and Bernstein, 1994). Each threshold needs to be attached to the correct statistic.
The third mistake is running the analysis many times without saving the output. Create folders for the original data, cleaned data, syntax or operation log, output from each round, and a summary table of removed items. This lets you answer when your supervisor asks why an item appears in chapter 3 but disappears in chapter 4.
The fourth mistake is making an overly strong causal conclusion from a regression table. If your design is a one-time survey, report the result within the scope of the model and the data collected. A statistically significant coefficient should not become a claim that X certainly causes Y in every context.
Read more: scientific research, what is scientific research, what is qualitative research, research field, research subject.
Frequently asked questions
What is quantitative research, and do I have to use SPSS?
Quantitative research uses numerical data to measure variables and test research questions or hypotheses. SPSS is a common choice, but you can also use SmartPLS, AMOS, JASP, or R depending on the model and your institution's requirements.
How large does the sample need to be for quantitative research?
The sample size depends on the number of items, the number of predictor variables, the analysis method, and the sampling approach. For items, the threshold of 5 to 10 observations per item is commonly cited from (Hair et al., 2010). For regression, you can refer to the 50 + 8m rule from (Tabachnick and Fidell, 2013), then compare it with your design and your supervisor's requirements.
What should I do if my quantitative research has a low Alpha?
First, check reverse-coded items, data-entry errors, and the content of the items. Then examine Corrected Item-Total Correlation, read the Cronbach's Alpha if Item Deleted column, and consider the theory before removing an item. Alpha from 0.60 may be considered for a new scale or exploratory study according to (Hair et al., 2010), but you should explain the context.
Does quantitative research require EFA?
EFA is commonly used when you need to explore the factor structure or check a scale in an adapted context. If the model and scale already have a clear foundation, the appropriate test depends on the research design and your supervisor's guidance. Do not run EFA only because a sample thesis included it.
How are quantitative and qualitative research different?
Quantitative research uses numerical data, scales, and statistical tests to answer questions about levels, relationships, or differences. Qualitative research focuses on the meaning and interpretation of non-numerical data. You can also read research subject and research field while defining the scope of your topic.
How to write this in your thesis
Open your data file again, record the independent variable, dependent variable, scale type, and planned analysis, then compare each output table with a sourced threshold instead of looking only for a convenient number. If you need to run the analysis on your own .sav or .csv file, M4 analysis by DoThesis is the module for this step.