Logistic regression teaching

Binary Logistic Regression
Mark Tranmer
Mark Elliot

2
CONTENTS
.....................................................................................................................1) Introduction 3
.....................................................................................Categorical data and 2 x 2 tables 3
..................................................................................................Odds and Relative Odds 5
.............................................................................................................................Odds 5
...............................................................................................................Relative odds 5
...............................................................................................2) Logistic regression theory 6
......................................................................................................................Introduction 6
......................................................................................................................The Theory 6
...........................................................................................Logistic regression theory 7
.........................................................................................................Dummy variables 9
.......................................................................................................................Exercise 2 10
.....................................................................................3) Logistic regression in SPSS 13 11
..................................................................................4) Summary and further comments 42
.......................................................................................................................Summary 42
..........................................................................................................Further comments 42
........................................................................................................................Reading list 43

3
1) Introduction
Socio-economic variables are very often categorical, rather than interval scale. In
many cases research focuses on models where the dependent variable is
categorical. For example, the dependent variable might be ‘unemployed’ /
‘employed’, and we could be interested in how this variable is related to sex, age,
ethnic group, etc. In this case we could not carry out a multiple linear regression
as many of the assumptions of this technique will not be met, as will be explained
theoretically below. Instead we would carry out a logistic regression analysis.
Hence, logistic regression may be thought of as an approach that is similar to that
of multiple linear regression, but takes into account the fact that the dependent
variable is categorical.
Categorical data and 2 x 2 tables
We can write categorical data in two forms: list form or table form. The important
point to make about this is that whichever way we choose to think about this kind
of data, the information is the same. For example, if we were interested in the
association between unemployment and sex for a sample of 12 people (this is a
smaller sample than we would tend to use in general but it illustrates the point),
we could write the data in list form as:
OBS UNEM FEMALE
1 0 0
2 0 1
3 0 1
4 0 1
5 0 0
6 1 1
7 0
1
8 1 0
9 0 0
10 0 1
11 0 0
12 1 0

4
Or the same data in table form as:
UNEM NOT UNEM TOTAL
MALE 2 4 6
FEMALE 1 5 6
TOTAL 3 9 12
2 x 2 tables are quite a good way to present the information, because the
relationship between the two variables can usually be clearly interpreted. When
we are interested in the association between several variables we can, of course,
still construct a multi-way table. However, it is less easy to interpret the
relationships from the tables when several variables are involved.
Example 1
We will now consider an example from Plewis, I (1997), Chapter 5.
Table 1.
Ethnic Group
Behaviour
Problems
White Black Total
NO 90 [0.83] 30 [0.48] 120 (70%)
YES 19 [0.17] 33 [0.52] 52 (30%)
Total 109 (63%) 63 (37%) 172 (100%)
Table 1 is a cross tabulation of two binary variables for a sample of 172 boys in
reception classes.
• Whether or not the child is perceived by their teacher to have a behaviour
problem (which we will later model as the response).
• Ethnic group (which we will later model as the explanatory variable).

5
We can see that the majority of the sample of boys (70%) are not perceived to
have a behaviour problem and that 63% of them are white. The conditional
probabilities of having a behaviour problem, given ethnic group are shown in
square brackets after each of the cell frequencies. For example the probability of
being perceived to have a behaviour problem for white boys is 0.17, and for black
boys is 0.52.
Odds and Relative Odds
A useful way of using the information in cross tabulations where one dimension
of the table is an outcome of interest (whether 2x2 tables or more complicated
ones), is to calculate odds and relative odds (odds ratios).
Odds
In the above table, the odds of a white boy being seen to have a behaviour
problem are 19/90 = 0.21 or 0.21 to 1. In betting terms that is about 5 : 1 against
– much less than even money.
For black boys, the corresponding odds are 33/30 = 1.1, or 1.1 to 1.
Equivalent to 11 to 10 on, (or a little better than even money.). Note that odds are
not the same as probabilities – they are not restricted to the range 0 to 1.
Relative odds
We can also think of the information in the table in terms of relative odds. The
relative odds of a black boy compared with a white boy being seen as having a
behaviour problem are 1.1 / 0.21 or 5.2 to 1. In other words a black boy is 5.2
times more likely than a white boy to be seen as having a behaviour problem.
Equally, boys perceived to have behaviour problems are 5.2 times more likely to
be black rather than white, compared with boys without perceived behaviour
problems. Relative odds are symmetrical in that sense; like correlation, we do not
think of this measure in terms of a dependent variable and an explanatory
variable. We just think in terms of the association between two variables.
Exercise 1
Suppose we are interested in the relationship between unemployment and ethnic
group for a sample of 18 year olds and we have the following data
Unemployed at 18? Ethnic group
TOTALWhite Black
No 1700 40 1740
Yes 112 8 120
Total 1812 48 1860

6
Calculate the probabilities, odds and relative odds of being unemployed at 18 for
white and black ethnic groups
2) Logistic regression theory
Introduction
When we want to look at a dependence structure, with a dependent variable and a
set of explanatory variables (one or more), we can use the logistic regression
framework.
Multiple linear regression may be used to investigate the relationship between a
continuous (interval scale) dependent variable, such as income, blood pressure or
examination score. However, socio-economic variables are very often categorical,
rather than interval scale. In many cases research focuses on models where the
dependent variable is categorical. For example, the dependent variable might be
‘unemployed’ or ‘not’ (as we saw in Exercise 1) , and we could be interested in
how this variable is related to sex, age, ethnic group, etc. In this case we could not
carry out a multiple linear regression as many of the assumptions of this
technique will not be met, as will be explained theoretically below. Instead we
would carry out a logistic regression.
The Theory
If we wrote the ‘perceived behaviour problems’ table as data in list format, we
would be interested in modelling the variation in the probability of being
perceived to have behaviour problems, and for the table data we are interested in
modelling the variations in the proportions with perceived behaviour problems
amongst black boys compared with white boys. It is important to note that
regardless of whether we consider the analysis in terms of data in a list or a table,
the results will be exactly the same.
Proportions and probabilities are different from continuous variables in a
number of ways. They are bounded by 0 and 1, whereas in theory continuous
variables can take any value between plus or minus infinity. This means that we
cannot assume normality for a proportion, and we must recognise that
proportions have a binomial distribution. Unlike the normal distribution, the
mean and variance of the Binomial distribution are not independent. The mean is
denoted by P and the variance is denoted by P*(1-P)/n, where n is the number of
observations, and P is the probability of the event occurring (e.g. the probability
of being unemployed, or having ‘perceived behavioural problems’) in any one
‘trial’ (for any one individual in this example). If we were considering the data in
‘list’ rather than table form we would assume that the variable had a mean P and
a variance P*(1-P) and this variable would have a Bernoulli distribution.

7
When we have a proportion as a response, we use a logistic or logit
transformation to link the dependent variable to the set of explanatory variables.
The logit link has the form:
Logit (P) = Log [ P / (1-P)]
The term within the square brackets is the odds of an event occurring. In the
example above this would be the odds of a person being perceived to have
behaviour problems.
Using the logit scale changes the scale of a proportion to plus and minus infinity,
and also because Logit (P) = 0, when P=0.5. When we transform our results back
from the logit (log odds) scale to the original probability scale, our predicted
values will always be at least 0 and at most 1.
Logistic regression theory
Let:
Then we can write the model:
In our example Pi is the probability of being perceived as having behaviour
problems, and xi is the boy’s ethnic group. Therefore the parameter β0 gives the
log odds of a white boy being perceived to have behaviour problems (when xi =0)
and β1 shows how these odds differ for black boys (when xi =1).
We can write the model in terms of odds as:
Pi/(1-Pi) = exp(β0 + β1xi)
Or in terms of the probability of the outcome (e.g. perceived behaviour problems)
occurring as:
Pi= exp(β0+β1xi)/(1 + exp(β0+β1xi))
Conversely the probability of the outcome not occurring is

8
1- Pi = 1/(1 + exp(β0+β1xi))
Notice that we have so far not included a residual term in the models, and have
instead expressed the model in terms of population probabilities. But we could
write it as:
pi = Pi + fi = exp(β0+β1xi)/(1 + exp(β0+β1xi)) + fi
Note that in this case, fi is not normally distributed, as it was assumed to be for
linear regression.
Returning to Example 1: perceived behavioural problems by ethnic
group
We will now consider some logistic regression theory and return to Example 1,
session 1.
Ethnic Group
Behaviour
Problems
White Black Total
NO 90 [0.83] 30 [0.48] 120 (70%)
YES 19 [0.17] 33 [0.52] 52 (30%)
Total 109 (63%) 63 (37%) 172 (100%)
We can fit a logistic regression model to the data in Table 1. We get:
Logit P = -1.56 + 1.65EG
Which we can interpret as the log odds of a white boy (EG=0) seen as having a
behaviour problem being equal to –1.56, hence the odds of a white boy having a
behaviour problem are: exp(-1.56) = 0.21
The log odds of a black boy (EG=1) having a perceived behaviour problem are –
1.56 + 1.65 = 0.09. Hence the odds of a black boy having a perceived behaviour
problem are exp(0.09) = 1.1 Alternatively we can say that the odds for black boys
are exp(1.65)=5.21 times as high as they are for white boys. That is, the relative
odds of a teacher perceiving a black boy to have behavioural problems compared
with a black boy are 5.21.
Notice that these results correspond exactly to the results in Table 1. This is
because for Table 1 there is one degree of freedom: we can calculate the degrees
of freedom in the table as (r-1)*(c-1) = 1, where r is the number of rows in the

9
table and c is the number of columns. So if we fit one parameter (ethnic group,
EG) we have used up, or saturated, all the degrees of freedom and hence fitted a
saturated model. This means we have fitted enough terms in the model to explain
everything that is going on in Table 1. Before we fitted the ethnic group term, we
could not have explained everything that is going on in the table, and we would
hence find a deviance (or –2 log likelihood) of 22.8. The deviance is a measure
of how much variation is left having fitted the model (how much is left
unexplained by the model). The deviance follows a chi2 distribution and we would
in general compare the difference in deviance in two models to find out if the
extra terms we added were significant. In the current example, we cannot really
talk in terms of ‘change in deviance’ because once we have fitted EG to the model,
we have fitted a saturated model, and the deviance is 0, but in general, assessing
the change in deviance is a very useful way of assessing whether we need to add
extra terms to our model. In the workshop examples we will see how this works
in much more detail.
Dummy variables
When an explanatory variable is categorical we use dummy variables to contrast
the different categories. For each variable we choose a baseline category and then
contrast all remaining categories with the base line. If an explanatory variable
has k categories, we need k-1 dummy variables to investigate all the differences in
the categories with respect to the dependent variable.
For example suppose the explanatory variable was housing tenure coded like this:
Tenure
1: Owner occupier
2: renting from a private landlord
3: renting from the local authority
We would therefore need to choose a baseline category and create two dummy
variables. For example if we chose owner occupier as the baseline category we
would code the dummy variables like this
Tenure: D1 D2
Owner occupier 0 0
Rented private 1 0
Rented local authority 0 1
For logistic regression SPSS can create dummy variables for us from categorical
explanatory variables, as we will see later.

10
Exercise 2
Write down a statistical model to investigate the relationships in the
following table
Unemployed Ethnic group
TOTALAfro Caribbean Pakistani Indian
No 50 40 45 135
Yes 9 5 4 18
Total 59 45 49 153

11
3) Logistic regression in SPSS 13
These data are taken from the British Election Study 2005 pre-campaign and
post-election panel data. More information:
http://www.essex.ac.uk/bes/
We will consider the propensity to vote (sometimes called ‘turnout’) as the
dependent variable, which has 2 categories. 0=did not turn out to vote, 1 turned
out to vote.
We will consider turnout in relation to three explanatory variables: gender, age
and housing tenure of the respondent. Turnout is known to be lower amongst
young people in western democracies, and may also be associated with tenure
and gender. We will use logistic regression to investigate the extent of the
association between the propensity to turn out to vote, with respect to gender,
age and tenure in the 2005 election data.
But first some exploratory data analysis: we will check the distributions of each of
the variables and do some filtering of the data and re-coding of the variables.
NB: the dataset which we will use here is called turnout1.sav.

13
frequency of turnout from unfiltered data shown below.
We will now filter the dataset so that it only contains those people who either
answered yes or no to “did you vote in the general election 2005?".

15
Frequency of turnout from filtered data:
We will now recode the bq12a variable into another variable called ‘vote2005’
which we will recode as 0=didn’t turn out to vote, 1=did turn out to vote. This will
enable us to model the probability of turning out to vote, which is the response
we require.

20
Question why are some bars much lower than their neighbours?

22
From these results we can see that:
The conditional probability of male turning out to vote are 1346/1837 = 0.733
(which we note that when multiplied by 100 is equal to the row % in this table,
given the way the cross tab is organised).
The conditional probability of female turning out to vote are 1729/2316 = 0.747

24
The odds of a male turning out to vote are:
1346/491 = 2.741
The odds of female turning out to vote are
1729/587 = 2.945
relative odds (female: male) are
(1729/587) / (1346/491) = 1.074
We will now cross tabulate vote2005 and housing tenure.

25
From this table we can see that, according to these data, owner occupier (‘owns’)
are much more likely to turnout to vote than renter (‘rents’). The conditional
probability of owner occupier turning out to vote is 0.805 whereas for renters it is
0.569.
(If time permits, please work out the odds and relative odds for owns and rents).
There are a total of 49 people who describe their housing tenure as neither ‘owns’
or ‘rents’.

26
A scatterplot of the relationship between age and the proportion at each age
turning out to vote shows that there is a much higher chance of turning out to
vote when you are older.
The relationship between age and propensity to turn out to vote
(using the variable vote2005mean).

27
Line of best fit: linear
Line of best fit: quadratic

28
Logistic regression models
We can access the logistic regression procedure in SPSS as follows:
vote2005 is our dependent variable. We begin by adding gender – a categorical
variable (which is coded on the dataset as 1=male, 2=female). We must ensure
that the fact that gender is a categorical variable is declared in our analysis and
we must choose the reference category. We will choose the first category, male, as
the reference category as is shown below.

30
Now Click on Continue and then OK to run the model!
We can see from the table above that we are modeling 4156 cases here (some
cases are deleted from the analysis where information is missing. The SPSS
default for this is listwise. Only cases where all dependent and explanatory
variables are complete are included in the analysis.). The tables below show us
firstly that we have coded our dependent variable in the right direction and
secondly that the categorical variable for gender has reference category of male.
The (1) means that gender (1) in the results refers to female here.

32
Block 1: Method = Enter
Adding one variable to the
We have added one new variable to the model, which has reduced the -2 log
likelihood by 1.018 with 1 degree of freedom. The -2 log likelihood is a measure of
how well the model explains variations in the outcome of interest, in this example
turnout. The -2 log likelihood (sometimes called, deviance) has a chi squared
distribution. The p value for the result of adding gender to the model is given in
the table above and we can see that this is 0.313 which is greater than the
conventional significance level of 0.05. hence we would conclude that the
addition of gender to the model is not statistically significant. In other words this
variable does not explain variations in turnout.

33
We see from the table above that the estimated model is
As we have recoded gender to 0=male, 1=female, this is equivalent to:
We can see that the coefficient of gender is non-significant (sig = 0.313 > 0.05).
The Exp(B) column shows the relative odds (odds ratio) and indicates that
females are 1.074 times as likely to turnout to vote than males. We can request a
confidence interval for this result as shown below.

34
The confidence interval for exp(B) is 0.935 to 1.235 indicates that females are
between 0.935 and 1.235 times as likely to turn out to vote than females. i.e. the
range has a lower limit of ‘slightly less than males’ and upper limit of ‘slightly
more than males’ and therefore includes ‘males and females are equally likely to
turn out to vote (i.e. exp(B)=1). This is not surprising since we have already
concluded that gender has no statistically significant explanatory power in
explaining variations in turnout. We will now add an additional variable to the
model – age in years (which is a continuous variable, rather than a categorical
one).
We will make use of the ‘block’ procedure to add age to the model, so that we can
see both the effect of adding age alone on the -2 log likelihood as well as seeing
how a model which includes both age and gender might explain variations in
turnout.

35
The addition of age to the model has, as a single variable, reduced the -2 log
likelihood by 300.666 on 1 degree of freedom. The model, which now contains 2
parameters, gender and age has collectively reduced -2 log likelihood by 301.684
but we can see it is age that has the explanatory power, and gender is not adding
anything extra.
the model which includes gender and age explains between 7 and 10% of the
variation in turnout.

36
the model is now:
logit(vote2005) = -.779 + .077gender(1)+.037age
The age coefficient is statistically significant. Exp(B) for age is 1.038, which
means for each year different in age, the person is 1.038 times more likely to turn
out to vote, having allowed for gender in the model. Eg. a 21 year old is 1.038
times as likely to turn out to vote than a 20 year old. This might not seem much
of a difference but a 20 year difference leads to a person being 1.038^20 = 2.11
times more likely to turn out to vote. E.g. a 40 year old is 2.11 times more likely to
turn out to vote than a 20 year old, having allowed for gender in the model.
We can add housing tenure to the model, as ‘block 3’ tenure is a categorical
variable, as was gender. By declaring it as categorical we can set up dummy
(indicator) variables and make the first category ‘owns’ the reference category:

37
Click on Continue and then OK to run the model again.

41
We can immediately see that tenure reduces the -2 log likelihood by 174.499
having added 2 new variables (tenure has 3 categories in all so we need 2 dummy
variables). Tenure is statistically significant in this model.

42
The table above shows us that the estimated model is now:
logit (vote2005) = -.326 + .101gender(1) + .034age – 1.053tenure(1) - .
324tenure(2)
in other words
logit (vote2005) = -.326 + .101female + .034age – 1.053rents - .324neither
tenure(1) which contrasts ‘rents’ with ‘owns’ has an exp(B) of 0.349 which means
that a person who rents is only .349 times (i.e. much less) likely to turn out than a
person who owns their own property, having allowed for gender and age. If we
calculate the inverse of exp(B) here, i.e. 1/0.349 = 2.87, we can say that a person
who owns their own home is 2.87 times more likely to vote than someone who
rents, having allowed for gender and age.
4) Summary and further comments
Summary
We have seen how logistic regression analysis may be used to analyse tabular
data where one of the dimensions of the table is an outcome of interest. This
morning, we looked at some examples where we calculated the probabilities, odds
and relative odds from the table, and we have seen how we can also calculate
these (and get the same results) from the model parameter estimates. Some
theory was introduced and we saw how the logistic model framework is a good
way to investigate associations in multi-way tables where one of the dimensions
of the tables is an outcome of interest, with two categories.
We have seen how we can use SPSS to fit logistic regression models to data using
an example based on the 2005 UK election. We covered main effects models and
models with interactions and we went through the output that SPSS gives us,
including the classification table, the deviance, the model coefficients and other
useful measures such as exp(B), which gives the relative odds or odds ratio for a
particular explanatory variable, given the other explanatory variables in the
model.
Further comments

43
The term ‘generalised linear model’ is used to describe a procedure for
transforming the dependent variable so that the ‘right hand side’ of the model
equation can be interpreted as a ‘linear combination’ of the explanatory variables:
In situations where the dependent (y) variable is continuous and can be
reasonably assumed to have a normal distribution we do not transform the y
variable at all and we can simply run a multiple linear regression analysis.
In situations where the dependent variable is dichotomous or 0/1 as we have seen
today the most common procedure is to use logistic regression, using the logit
link as we have done today. Other similar types of modelling include probit
modelling. (See Dobson, McCullagh and Nelder for further details – details of
these references in the reading list).
When the response variable has several categories we can use a model that allows
for several categories in the response variable such as multinomial regression. If
this response variable is ordinal (as opposed to nominal) we can allow for this in
the modelling (see Agresti – reference details in reading list). An alternative is to
recode the response variable into just two categories and do a logistic regression
analysis (or to fit several logistic regression models to different pairs of categories
in the response variable, although this is not as statistically efficient as doing a
true multinomial analysis.
Note also that logistic regression models can also be fitted with multilevel
components in MLwiN and STATA.
Reading list
Field, A. (2005) Discovering statistics using SPSS for Windows: advanced
techniques for the beginner, London: Sage. Chapter 6.
Plewis, I (1997) Statistics in Education, Edward Arnold. (Especially chapter 5)
Dobson, A (2001) An introduction to generalized linear models (second
edition). Chapman and Hall.
McCullagh P and Nelder J.A, (1989) Generalized linear models (second
edition). Chapman and Hall
Agresti, A. (1996) Introduction to categorical data analysis. John Wiley.

Logistic regression teaching

More Related Content

What's hot (19)

Viewers also liked (20)

Similar to Logistic regression teaching (20)

More from Pakistan Gum Industries Pvt. Ltd (20)

Recently uploaded (20)

Logistic regression teaching