C8057 (Research Methods 2): Simple Regression

Simple Regression
An Introduction to Regression
In regression analysis we fit a predictive model to our data: we use a model to predict values of the
dependent variable (DV) from one or more independent variables (IVs).1 Simple regression seeks to predict
an outcome from a single predictor whereas multiple regression seeks to predict an outcome from several
predictors. This is an incredibly useful tool because it allows us to go a step beyond the data that we
actually possess. The model that we fit to our data is a linear one and can be imagined by trying to
summarize a data set with a straight line.
With any data set there are a number of lines that could be used to summarize the general trend and so
we need a way to decide which of many possible lines to choose. For the sake of drawing accurate
conclusions we want to fit a model that best describes the data. The simplest way to fit a line is to use
your eye to gauge a line that looks as though it summarizes the data well. However, the eyeball method
is very subjective and so offers no assurance that the model is the best one that could have been chosen.
Instead, we use a mathematical technique called the method of least squares to find the line that best
describes the data collected (See Field, 2005, Chapter 5 for a description).
Some Important Information about Straight Lines
Any straight line can be drawn if you know: (1) the slope (or gradient) of the line, and (2) the point at
which the line crosses the vertical axis of the graph (the intercept of the line). The equation of a straight
line is defined in equation (1), in which Y is the outcome variable that we want to predict and Xi is the ith
subjects score on the predictor variable. b1 is the gradient and b0 is the intercept of of the straight line
fitted to the data. There is a residual term, i, which represents the difference between the score
predicted by the line for subject i and the score that subject i actually obtained. The equation is often
conceptualized without this residual term (so, ignore it if its upsetting you); however, it is worth knowing
that this term represents the fact our model will not fit perfectly the data collected.

Yi = b0 + b1X i + i (1)
A particular line has a specific intercept and gradient. Figure 1 shows a set of lines that have the same
intercept but different gradients, and a set of lines that have the same gradient but different intercepts.
Figure 1 also illustrates another useful point: that the gradient of the line tells us something about the
nature of the relationship being described: a line that has a gradient with a positive value describes a
positive relationship, whereas a line with a negative gradient describes a negative relationship. So, if you
look at the graph in Figure 1 in which the gradients differ but the intercepts are the same, then the
thicker line describes a positive relationship whereas the thinner line describes a negative relationship.
If it is possible to describe a line knowing only the gradient and the intercept of that line, then the model
that we fit to our data in linear regression (a straight line) can also be described mathematically by
equation (1). With regression we strive to find the line that best describes the data collected, then
estimate the gradient and intercept of that line. Having defined these values, we can insert different
values of our predictor variable into the model to estimate the value of the outcome variable.

1
Unfortunately, you will come across people (and SPSS for that matter) referring to regression variables as
dependent and independent variables (as in controlled experiments). However, correlational research by
its nature seldom controls the independent variables to measure the effect on a dependent variable.
Instead, variables are measured simultaneously and without strict control. It is, therefore, inaccurate to
label regression variables in this way. For this reason I label independent variables as predictors, and the
dependent variable as the outcome.

Dr. Andy Field, 2000, 2005 Page 1

C8057 (Research Methods 2): Simple Regression

100 100
80 80

60 60

40 40

20 20

0 0
0 10 20 30 40 0 10 20 30 40

Same intercept, different slopes Same slope, different intercepts

Figure 1: Shows lines with the same gradients but different intercepts, and lines that share the same
intercept but have different gradients

Assessing the Goodness-of-Fit: Sums of Squares, R and R2

Once we have found the line of best fit it is important that we assess how well this line fits the actual
data (we assess the goodness-of-fit of the model). One measure of the adequacy of a model is the sum of
squared differences (think back to lecture 1, or Field, 2005 Chapter 1). Sticking with this theme, there are
several sums of squares that can be calculated to help us gauge the contribution of our model to
predicting the outcome. Imagine that I was interested in predicting record sales (Y) from the amount of
money spent advertising that record (X). One day my boss came in to my office and said Andy, how many
records will we sell if we spend 100,000 on advertising? In the absence of any data probably the best
answer I could give would be the mean number of record sales (say, 200,000) because on average thats
how many records we expect to sell. However, what if he the asks How many records will we sell if we
spend 1 on advertising? Again, in the absence of any accurate information, my best guess would be the
mean. There is a problem: whatever amount of money is spent on advertising I always predict the same
levels of sales. As such, the mean is a fairly useless model of a relationship between two variables.
Using the mean as a model, we can calculate the difference between the observed values, and the values
predicted by the mean. We saw in lecture 1 that we square all of these differences to give us the sum of
squared differences. This sum of squared differences is known as the total sum of squares (denoted SST)
because it is the total amount of error present when the most basic model is applied to the data. Now, if
we fit the more sophisticated model to the data, such as a line of best fit, we can work out the
differences between this new model and the observed data. Even if an optimal model is fitted to the data
there is still some inaccuracy, which is represented by the differences between each observed data point
and the value predicted by the regression line. These differences are squared before they are added up so
that the directions of the differences do not cancel out. The result is known as the sum of squared
residuals (SSR). This value represents the degree of inaccuracy when the best model is fitted to the data.
We can use these two values to calculate how much better the regression line (the line of best fit) is than
just using the mean as a model (i.e. how much better is the best possible model than the worst model?).
This improvement in prediction is the difference between SST and SSR. This difference shows us the
reduction in the inaccuracy of the model resulting from fitting the regression model to the data. This
improvement is the model sum of squares (SSM).
If the value of SSM is large then the regression model is very different from using the mean to predict the
outcome variable. This implies that the regression model has made a big improvement to how well the
outcome variable can be predicted. However, if SSM is small then using the regression model is little
better than using the mean (i.e. the regression model is no better than taking our best guess). A useful
measure arising from these sums of squares is the proportion of improvement due to the model. This is
easily calculated by dividing the sum of squares for the model by the total sum of squares. The resulting
value is called R2 and to express this value as a percentage you should multiply it by 100. So, R2 represents
the amount of variance in the outcome explained by the model (SSM) relative to how much variation there
was to explain in the first place (SST). Therefore, as a percentage, it represents the percentage of the
variation in the outcome that can be explained by the model.

Dr. Andy Field, 2000, 2005 Page 2

C8057 (Research Methods 2): Simple Regression

SS M
R2 = (2)
SS T

Interestingly, this value is the same as if you square the Pearson correlation between variables (see Field,
2005, Chapter 4) and it is interpreted in the same way. Therefore, in simple regression we can take the
square root of this value to obtain the Pearson correlation coefficient. As such, the correlation coefficient
provides us with a good estimate of the overall fit of the regression model, and R2 provides us with a good
gauge of the substantive size of the relationship.
A second use of the sums of squares in assessing the model is through the F-test. The F-test is something
we will cover in greater detail later in the term, but briefly this test is based upon the ratio of the
improvement due to the model (SSM) and the difference between the model and the observed data (SSR).
In fact, rather than using the sums of squares themselves, we take the mean sums of squares (referred to
as the mean squares or MS). The result is the mean squares for the model (MSM) and the residual mean
squares (MSR) see Field 2005 for more detail. At this stage it isnt essential that you understand how the
mean squares are derived (it is explained in Field, 2005, Chapter 8 and in your lectures on ANOVA).
However, it is important that you understand that the F-ratio (equation (3)) is a measure of how much the
model has improved the prediction of the outcome compared to the level of inaccuracy of the model.

MS M
F= (3)
MS R
If a model is good, then we expect the improvement in prediction due to the model to be large (so, MSM
will be large) and the difference between the model and the observed data to be small (so, MSR will be
small). In short, a good model should have a large F-ratio (greater than one at least) because the top half
of equation (3) will be bigger than the bottom. The exact magnitude of this F-ratio can be assessed using
critical values for the corresponding degrees of freedom.

Simple Regression on SPSS

There are some data given by Field 2005 in the file Record1.sav. This data file has 200 rows, each one
representing a different record. There are also two columns, one representing the sales of each record in
the week after release and the other representing the amount (in pounds) spent promoting the record
before release. This is the format for entering regression data: the outcome variable and any predictors
should be entered in different columns, and each row should represent independent values of those
variables. The pattern of the data is shown in Figure 2 and it should be clear that a positive relationship
exists: so, the more money spent advertising the record, the more it is likely to sell. Of course there are
some records that sell well regardless of advertising (top left of scatterplot), but there are none that sell
badly when advertising levels are high (bottom right of scatterplot). The scatterplot also shows the line of
best fit for these data: bearing in mind that the mean would be represented by a flat line at around the
200,000 sales mark, the regression line is noticeable different.
To find out the parameters that describe the regression line, and to see whether this line is a useful
model, we need to run a regression analysis. To do the analysis you need to access the main dialog box by
using the AnalyzeRegressionLinear menu path. Figure 3 shows the resulting dialog box. There is a
space labelled Dependent in which you should place the outcome variable (in this example sales). So,
select sales from the list on the left-hand side, and transfer it by clicking on . There is another space
labelled Independent(s) in which any predictor variable should be placed. In simple regression we use only
one predictor (in this example adverts) and so you should select adverts from the list and click on to
transfer it to the list of predictors. There are a variety of options available, but these will be explored
within the context of multiple regression (see Field, 2005). For the time being just click on to run
the basic analysis.

Dr. Andy Field, 2000, 2005 Page 3

C8057 (Research Methods 2): Simple Regression

Figure 2: Scatterplot showing the Figure 3: Main dialog box for regression
relationship between record sales and the
amount spent promoting the record

Output from SPSS

Overall Fit of the Model
The first table provided by SPSS is a summary of the model that gives the value of R and R2 for the model.
For these data, R is 0.578 and because there is only one predictor, this value represents the simple
correlation between advertising and record sales (you can
Model Summary confirm this by running a correlation). The value of R2 is
Std. Error
0.335, which tells us that advertising expenditure can
Adjusted R of the account for 33.5% of the variation in record sales. There
Model R R Square Square Estimate
might be many factors that can explain this variation, but
a
1 .578 .335 .331 65.9914 our model, which includes only advertising expenditure
a. Predictors: (Constant), Advertising Budget (thousands
explains 33%: 66% of the variation in record sales is
of pounds)
unexplained. Therefore, there must be other variables that
have an influence also.
SPSS Output 1
The next part of the output reports an analysis of variance
(ANOVAsee Field, 2005 Chapter 7). The most important part of the table is the F-ratio, which is
calculated using equation (2), and the associated significance value. For these data, F is 99.59, which is
significant at p < 0.001 (because the value in the column labelled Sig. is less than 0.001). This result tells
us that there is less than a 0.1% chance that an F-ratio this large would happen by chance alone.
Therefore, we can conclude that our regression model results in significantly better prediction of record
sales than if we used the mean value of record sales. In short, the regression model overall predicts
record sales significantly well.

ANOVAb

Sum of Mean
Model Squares df Square F Sig.
a
1 Regression 433687.833 1 433687.833 99.587 .000
Residual 862264.167 198 4354.870
Total 1295952.0 199
a. Predictors: (Constant), Advertising Budget (thousands of pounds)
b. Dependent Variable: Record Sales (thousands)

SPSS Output 2

Dr. Andy Field, 2000, 2005 Page 4

C8057 (Research Methods 2): Simple Regression

Model Parameters
The ANOVA tells us whether the model, overall, results in a significantly good degree of prediction of the
outcome variable. However, the ANOVA doesnt tell us about the individual contribution of variables in
the model (although in this simple case there is only one variable in the model and so we can infer that
this variable is a good predictor). The table in SPSS Output 3 provides details of the model parameters
(the beta values) and the significance of these values. We saw in equation (1) that b0 was the Y intercept
and this value is the value B for the constant. So, from the table, we can say that b0 is 134.14, and this
can be interpreted as meaning that when no money is spent on advertising (when X = 0), the model
predicts that 134,140 records will be sold (remember that our unit of measurement was thousands of
records). We can also read off the value of b1 from the table and this value represents the gradient of the
regression line. It is 9.612 E02, which in unabbreviated form is 0.09612.2 Although this value is the slope
of the regression line, it is more useful to think of this value as representing the change in the outcome
associated with a unit change in the predictor. Therefore, if our predictor variable is increased by 1 unit
(if the advertising budget is increased by 1), then our model predicts that 0.096 extra records will be sold.
Our units of measurement were thousands of pounds and thousands of records sold, so we can say that for
an increase in advertising of 1000 the model predicts 96 (0.096 1000 = 96) extra record sales. As you
might imagine, this investment is pretty bad for the record company: they invest 1000 and get only 96
extra sales! Fortunately, as we already know, advertising accounts for only one-third of record sales!

Coefficientsa

Unstandardized Standardized
Coefficients Coefficients
Model B Std. Error Beta t Sig.
1 (Constant) 134.140 7.537 17.799 .000
Advertising Budget
9.612E-02 .010 .578 9.979 .000
(thousands of pounds)
a. Dependent Variable: Record Sales (thousands)

SPSS Output 3
The values of b represent the change in the outcome resulting from a unit change in the predictor. If the
model was useless at predicting the outcome, then if the value of the predictor changes, what might we
expect the change in the outcome to be? Well, if the model is very bad then we would expect the change
in the outcome to be zero. A regression coefficient of zero means: (a) a unit change in the predictor
variable results in no change in the predicted value of the outcome (the predicted value of the outcome
does not change at all), and (b) the gradient of the regression line is zero, meaning that the regression
line is flat. Hopefully, what should be clear at this stage is that if a variable significantly predicts an
outcome, then it should have a b value significantly different from zero. This hypothesis is tested using a
t-test (see Field, 2005, Chapter 7). The t-statistic tests the null hypothesis that the value of b is zero:
therefore, if it is significant we accept the hypothesis that the b value is significantly different from zero
and that the predictor variable contributes significantly to our ability to estimate values of the outcome.
The values of t can be compared to the values that we would expect to find by chance alone: if t is very
large then it is unlikely to have occurred by chance. SPSS provides the exact probability that the observed
value of t is a chance result, and as a general rule, if this observed significance is less than 0.05, then
social scientists agree that the result reflects a genuine effect. For these two values, the probabilities are
0.000 (zero to 3 decimal places) and so we can say that the probability of these t values occurring by
chance is less than 0.001. Therefore, they reflect genuine effects. We can, therefore, conclude that
advertising budget makes a significant contribution to predicting record sales.

2
You might have noticed that this value is reported by SPSS as 9.612 E02 and many students find this
notation confusing. Well, think of E02 as meaning move the decimal place 2 steps to the left, so 9.612
E02 becomes 0.09612.

Dr. Andy Field, 2000, 2005 Page 5

C8057 (Research Methods 2): Simple Regression

Using the Model

So far, we have discovered that we have a useful model, one that significantly improves our ability to
predict record sales. However, the next stage is often to use that model to make some predictions. The
first stage is to define the model by replacing the values in equation (1) with the values from SPSS
Output 3. In addition, we can replace the X and Y with the variable names:

Record Sales i = b0 + b1Advertising Budgeti

(4)
= 134.14 + (0.09612 Advertising Budgeti )
It is now possible to make a prediction about record sales, by replacing the advertising budget with a
value of interest. For example, imagine a record executive wanted to spend 100,000 on advertising a
new record. Remembering that our units are already in thousands of pounds, we can simply replace the
advertising budget with 100. He would discover that record sales should be around 144,000 for the first
week of sales.

Record Sales i = 134.14 + (0.09612 Advertising Budgeti )

= 134.14 + (0.09612 100 ) (5)
= 143.75

Task 1
Lacourse et al. (2001) conducted a study to see whether suicide risk was related to listening to heavy
metal music. They devised a scale to measure preference for bands falling into the category of heavy
metal. This scale included heavy metal bands (Black Sabbath, Iron Maiden), speed metal bands (Slayer,
Metallica), death/black metal bands (Obituary, Burzum) and gothic bands (Marilyn Manson, Sistsers of
Mercy). They then used this (and other variables) as predictors of suicide risk based on a scale measuring
suicidal ideation etc. devised by Tousignant et al., (1988).
Lacourse, E., Claes, M., & Villeneuve, M. (2001). Heavy Metal Music and Adolescent Suicidal Risk. Journal
of Youth and Adolescence, 30 (3), 321-332. [Available through the Sussex Electronic Library].
Lets imagine we replicated this study. The data file HMSuicide.sav (on the course website) contains the
data from such a replication. There are two variables representing scores on the scales described above:
hm (the extent to which the person listens to heavy metal music) and suicide (the extent to which
someone has suicidal ideation and so on). Using these data carry out a regression analysis to see whether
listening to heavy metal predicts suicide risk.

How much variance does the final model explain?

Your
Answers:

Does listening to heavy metal significantly predict suicide risk (quote relevant statistics)?

Your
Answers:

Dr. Andy Field, 2000, 2005 Page 6

C8057 (Research Methods 2): Simple Regression

What is the nature of the relationship between listening to heavy metal and suicide
risk? (sketch a scatterplot if it helps you to explain).

Your
Answers:

Write out the regression equation for the model

Your
Answers:

As listening to heavy metal increases by 1 unit, how much does suicide risk increase?

Your
Answers:

This handout contains material from:

Field, A. P. (2005). Discovering statistics using SPSS (Second Edition).

London: Sage.
So, is Andy Field, 2005 please consult a copy for more detail.

Dr. Andy Field, 2000, 2005 Page 7