Using Statistical Regression Methods in Education Research




Data View

The SPSS Data View is the main spreadsheet in which you enter your data. Variables are listed across the top of the spreadsheet (with each column representing a variable) and cases are listed vertically (with each row representing a case or observation).

Degrees of freedom

'Degrees of freedom' is a very difficult concept to explain but crops up everywhere! Basically it is the number of values in the final calculation of a statistic which are free to vary. SPSS will usually tell you the degrees of freedom for your analysis.

Sorry if this explanation is not satisfactory! You will need to turn to a more statistically focused source for a thorough definition. Luckily, for the purposes of this website, it is not necessary to have an in-depth understanding of the concept.

Design weights

Design weights are used in large scale surveys to adjust for the unequal probabilities that different groups of cases are selected for the sample.

We do not really discuss design weights on this site but if you want to know more you can go to our Resources page.

Deviance (-2LL)

This is the log-likelihood multiplied by -2 and is commonly used to explore how well a logistic regression model fits the data. The lower this value is the better your model is at predicting your binary outcome variable.

Multiplying it by -2 is a technical step necessary to convert the log-likelihood into a chi-square distribution, which is useful because it can then be used to ascertain statistical significance. Don't worry if you do not fully understand the technicalities of this.


A variable is dichotomous (or binary) if it has only two categories. Variables such as gender are usually dichotomous (male or female). The occurrence or non-occurrence of an outcome can also be dichotomous (for example, expulsion from school against not being expelled from school).

Dummy variable

Dummy variables are a series of dichotomous variables created in SPSS so that regression analysis can be performed using a categorical (nominal or ordinal) variable with more than two categories.

One category, usually the one which contains the highest number of respondents, is designated as the 'reference' category and does not have a dummy variable. All other categories have a dummy variable created for them. Participants are coded '1' if they belong to the particular category of each dummy variable and '0' if not. Participants who belong to the reference category are coded as '0' for all dummy variables.

The coefficients (B) for each of these new variables tell us how much difference in the outcome is predicted for a member of that category relative to members of the reference group. See MLR module 3.7 for more detail.

Durbin-Watson test

The Durbin-Watson test allows you to check the independent errors assumption of multiple regression by checking for correlation between residuals. It is confusing! Fear not though: SPSS will calculate it for you and all you need to know is how to interpret it. It ranges between 0 and 4 and values less than 1 or greater than 3 are a cause for concern.

Page contact: Feedback to ReStore team Last revised: Thu 28 Jul 2011
Back to top of page