How to Run SPSS for Beginners: From Data to Output

When to use SPSS

SPSS is software for quantitative data analysis. You use it when you have survey data in an .sav, .xlsx, or .csv file and need to turn responses into statistical results for testing hypotheses in your thesis. The question “what does running SPSS mean” is therefore not just about opening the software and clicking a menu. You need to choose an analysis that matches your variables, model, and research question.

If your study uses Likert scales, the workflow usually moves from descriptive statistics to reliability testing with Cronbach's Alpha, EFA, and then linear regression or another appropriate test. You can consult the overview of SPSS topics, the guide to Cronbach's Alpha, and the guide to EFA when you need the individual steps.

The form of your hypothesis determines the analysis. A hypothesis that “X affects Y” usually leads to linear regression. A hypothesis that “the mean score of group A differs from group B” requires a t-test or ANOVA. A hypothesis about the structure of observed variables usually requires EFA. For a complex latent-variable model, you may consider SmartPLS or AMOS. AMOS is suited to the CB-SEM approach and has its own way of reading fit indices.

SPSS calculates results from the variables and options you enter. Before clicking OK, write down which question the analysis answers, which variables are independent, which variable is dependent, and which decision criteria you will use.

Preparing data before running SPSS

Open Data View and Variable View together. In Variable View, check the variable name, data type, variable label, coded values, and Measure. Variable names should be short and contain no spaces, such as CL1, CL2, HL1, and HL2. Likert items are usually set to Numeric. Grouping variables such as gender, age, or occupation need Value Labels so the output is easier to read.

Checking variable type and coding

A cell containing the text “Male” can prevent the mean from being calculated as intended if the variable should have been coded as 1 and 2. Conversely, an identification variable coded with numbers does not automatically become a continuous variable. Check the nature of the variable before choosing the analysis.

In Variable View, set Missing if the questionnaire uses a special code for unanswered items, such as 99. If you leave 99 as an ordinary value, SPSS will include it in the mean and distort the result. Then run Analyze > Descriptive Statistics > Frequencies to view the frequency, minimum, maximum, and number of missing values.

Checking reverse-coded items

If the questionnaire contains a statement with the opposite meaning, reverse-code it before running Cronbach's Alpha or EFA. For a Likert scale from 1 to 5, the new value can be created with the formula 6 - old value. Go to Transform > Recode into Different Variables, select the item to reverse, give the new variable a name such as HL3R, and define Old Value and New Value.

Do not overwrite the original variable before creating a data backup. Check the reverse-coded item with Frequencies to confirm that its values remain between 1 and 5. If an item that should move in the same direction as the other items has a negative corrected item-total correlation, check this step before deleting the item.

Checking missing values and unusual responses

A questionnaire with too many unanswered items should be flagged according to the rule stated in your research methods. For the remaining rows, check for one-point responses across all items, unusually short completion times, or values outside the scale. Keep an original data file and a cleaned version, and record the number of removed rows and the reason for removal.

SPSS only needs the columns used in the analysis. Do not place names, phone numbers, or personal identification codes in the output if they do not serve the model. When saving the .sav file, name the version by date, such as data_clean_v2.sav, so you can return to it when your supervisor asks about a variable that was removed.

Steps for running SPSS

Step 1: Import and save the data

For an Excel file, select File > Open > Data, choose the file, and check Read variable names from the first row of data if the first row contains variable names. For a CSV file, check the delimiter and review several of the first rows after importing. Save the file immediately as an .sav file through File > Save As to preserve the format and variable labels.

Step 2: Run descriptive statistics

Go to Analyze > Descriptive Statistics > Frequencies for categorical variables and Analyze > Descriptive Statistics > Descriptives for numerical variables. Select Mean, Standard Deviation, Minimum, and Maximum when you need to describe the scale. This output helps you identify values outside the permitted range before running the next steps.

Step 3: Run Cronbach's Alpha

Go to Analyze > Scale > Reliability Analysis. Move the items measuring the same construct into the Items box and select Alpha as the Model. Click Statistics, select Item, Scale, and Scale if item deleted, then select Continue and OK.

The important output includes Reliability Statistics and Item-Total Statistics. Read Cronbach's Alpha together with Corrected Item-Total Correlation. Do not delete an item only because Alpha if Item Deleted increases slightly. Item removal needs both a statistical and a scale-content justification.

Step 4: Run EFA

Go to Analyze > Dimension Reduction > Factor. Move all items that passed the reliability test into the Variables box. Under Descriptives, select KMO and Bartlett's test. Under Extraction, select the method appropriate for your research design, turn on the Scree plot, and set the eigenvalue criterion according to the procedure you selected.

Under Rotation, select Varimax when you assume that the factors are uncorrelated, or an oblique method such as Promax when the factors may be correlated. Under Options, turn on Sorted by size and select Suppress small coefficients, entering 0.50 if you are using this factor-loading threshold. The output to read includes KMO and Bartlett's Test, Total Variance Explained, and Rotated Component Matrix.

If the rotated matrix shows an item loading highly on two factors, loading weakly, or failing to join its theoretical group, do not delete a large number of items in one round. Record the removed item and the reason, then run EFA again. EFA rarely looks clean on the first run, and controlled reruns are a normal part of data processing.

Step 5: Run linear regression

Go to Analyze > Regression > Linear. Move the dependent variable into the Dependent box and the independent variables into Independent(s). Under Statistics, select Estimates, Model fit, Collinearity diagnostics, and Durbin-Watson. Under Plots, place *ZPRED on X and *ZRESID on Y if you need to inspect the residuals. Select Save when you need to save predicted values or residuals.

Read Model Summary, ANOVA, and Coefficients in that order. R Square shows the proportion of variation in Y explained by the variables in the model. The ANOVA table shows whether the overall model is significant according to its p-value. The Coefficients table shows the coefficient, direction of the effect, p-value, and VIF for each independent variable. For a deeper explanation, see the guide to linear regression.

Step 6: Save syntax and output

Click Paste instead of OK when running important analyses. SPSS will open the Syntax window with the corresponding command. Save the .sps file in the same folder as the .sav file. In Output Viewer, select File > Save As to save the .spv file. Syntax lets you reproduce the same procedure when the data are updated and explain each round of processing.

Reading SPSS output

Read the output in relation to your research question. Do not simply read every table in the order SPSS produces it. The table below is illustrative output, showing columns that commonly appear in regression output. The figures only demonstrate how to read the table and are not results from a real study.

In this example, R Square = 0.426 means that the model explains 42.6% of the variation in the dependent variable in the illustrative sample. CL has a positive Beta and a Sig. value displayed as 0.000. In your thesis, write p < 0.001 rather than saying that p is actually equal to 0. GC has a Sig. value of 0.141, so there is not enough statistical evidence at the 5% significance level to conclude that this variable has an effect in the illustrative model.

Do not interpret one number in isolation. If the model has a high R Square but a large VIF, you need to examine multicollinearity. If a coefficient is significant but has the opposite sign from the hypothesis, check variable coding, reverse-coded items, the construction of composite variables, and the theoretical basis before writing the conclusion.

Evaluation criteria

The thresholds below are commonly used reference points. They do not replace methodological reasoning and attention to the study context. Each row includes a source that you can retrieve when explaining the decision to your supervisor.

An Alpha below 0.7 can occur in exploratory research, but if you use the 0.6 reference point, state the reason clearly. Do not use one acceptable threshold to justify a scale that has several other problems.

If you use SmartPLS, do not place CFI, TLI, or RMSEA in this SPSS guide. Those indices belong to CB-SEM and are not core criteria for PLS-SEM. With SmartPLS, you will read outer loading, CR, AVE, HTMT, VIF, R², and Q² through a separate procedure.

How to write this in your thesis

Write in this order: name the analysis, report the main statistic, compare it with the criterion, and then state its implication for the next step. Do not copy the entire output into Chapter 4. The complete tables can go in the appendix, while the main text should retain only the figures needed for the hypotheses.

How to write this in your thesis: “The reliability test showed that the [construct name] scale contained [number of items] items, with Cronbach's Alpha of [value]. The Corrected Item-Total Correlation values ranged from [lowest value] to [highest value]. Since the indicators met the criterion of [threshold and source], the scale was retained for [EFA or regression].”

For regression, you can use this template: “The regression model had an R Square of [value], and the Sig. value of the ANOVA test was [value]. The [variable name] variable had a Beta coefficient of [value] and a p-value of [value], indicating a [positive or negative] effect on [dependent variable]. Since the p-value was [less than or greater than] 0.05, hypothesis H[ ] was [accepted or not supported] within the study sample.”

Replace the square-bracketed sections with the actual figures from your output. If a hypothesis is not supported, report that result accurately. Do not change the sign of a coefficient or call a variable “influential” simply because it correlates with the dependent variable.

Common errors when running SPSS

Selecting the wrong variable or measurement level

Putting gender into a regression as a continuous numerical variable can make the model difficult to interpret. For a categorical variable with several groups, use appropriate coding and treatment. Check the codebook before running the analysis instead of trying to fix the issue in the output.

Using text data in a calculation

If Likert columns were imported as String, SPSS may not calculate the Mean or include the variable in the analysis. Use Transform > Automatic Recode only when you understand the resulting codes, then check Frequencies. Where possible, it is better to correct the data structure in the original file and import it again.

Deleting items only to reach a threshold

Repeatedly deleting items to increase Alpha, KMO, or loading can strip the scale of its content. Each deletion needs a reason, a recorded version, theoretical review, and transparent reporting. If the rotated matrix remains unreasonable after several rounds, the problem may be in the questionnaire or model rather than in the SPSS run command.

Copying the wrong p-value or decimal

SPSS displays 0.000 when the p-value is very small. This does not mean that the p-value is exactly 0. When entering a table, preserve the decimal marks and distinguish between B, Beta, t, and Sig.. A small error in these columns can reverse the interpretation of the output.

Failing to save syntax and intermediate data

If you save only one final file, it will be difficult to explain which items were removed. Save the original data, cleaned data, syntax, output, and a log of each analysis round. This also lets you rerun the analysis if you discover that a reverse code was entered incorrectly.

Frequently asked questions

What is SPSS and what is it used for?

SPSS is software for entering, cleaning, describing, and analysing quantitative data. In a thesis, SPSS is commonly used to run Cronbach's Alpha, EFA, correlation, regression, t-tests, ANOVA, and descriptive statistics.

SPSS does not choose the model or explain the theoretical meaning for you. You still need to define the variables, hypotheses, assumptions, and reporting approach before running the analysis.

What order should a beginner follow when running SPSS?

Start by checking the data and running descriptive statistics. Then run Cronbach's Alpha, EFA if the model includes scales that need exploration, and finally regression or the test that matches your hypotheses. Save the output at each step and record which items were retained or removed.

Do not start with regression before checking coding, missing values, and scale quality. Output may still appear, but the conclusion will lack a sound basis.

How much SPSS output is enough?

There is no single number that applies to every analysis. Check the relevant statistic at each step, such as Alpha, corrected item-total correlation, KMO, Bartlett, factor loading, R Square, VIF, and p-value.

The thresholds in the table are sourced reference points. Apply them consistently, explain exceptions, and ask your supervisor when a result is close to a threshold.

What should I do when SPSS will not run?

First read the error message and check whether the variables have the correct Numeric type, special missing codes, or out-of-range values. Then check whether the Dependent, Independent, or Items box has been left empty.

If no table appears in the output window, run a simple analysis such as Frequencies to determine whether the problem is in the file or the dialog settings. Save a copy of the file before making changes, then rerun the analysis from syntax if you saved the command.

Should I download SPSS 27 for free from an unknown source?

Use a licensed copy or a license provided by your university. An installation file from an unknown source can put your computer and survey data at risk, and it may make support difficult when the software fails. If you only need practice, check whether your university offers a trial license or computer lab access.

Alternative tools such as JASP or R may suit particular situations, but their interfaces, command-saving methods, and output formats differ from SPSS. Choose a tool that you can explain clearly during your defence.

Open your data file again, check Variable View, and create a copy before cleaning the data. Then run Frequencies, save the syntax, and follow the analyses required by your model in order. If you need to run the analysis on your actual .sav or .csv file, see M4 data analysis.