Lecture 28 May 15, 2012
There 9 question in the midterm, all short answer, and they are application question, only
exception is qquestion 9 which is 30 marks worth. You get a scenario, which you have to
read and then answer the question and expand. Individual questions will all cover the
aspects of course and ask the example and work with the example. 1 to 8 will cover the
course material either you have to apply the concept or you have to use a concept to
answer the question. 2 of the questions were covered in class lecture notes and textbook
for details. It is mainly everything that we read and cover so far.
Following is what is not required to read for the test but you do need to know them if you
are doing quantitative analysis. 165 to 172 different types of index, you do not need to
know the indexes.
185 to 198 covers things such as sampling error confidence interval, and more of
statistical types of thing, which will be not asked in the text.
Today we are talking about the measurement and indexes and scales. Sampling is on
Validity and reliability when we are doing research we are always trying to achieve more
valid results more reliable results. One way we can do that is by creating measures that
are composite. So composite measures are measures that is kind of an umbrella of
different measures, so you are not relying on just one singe measure indicator it is a mix
of several or a lot of indicators into one composite measure. There are 3 types Indexes,
Scales, and Typologies. Indexes and scale are ordinal, there is ranking order attributes are
equal to values. Typologies on the other hand are nominal so that means that the different values or attributes cannot be ranked, they are different from one and other. Indexes and
scales are used more often in quantitative research (more focused on numbers and ranks,
quantifying things which indexes and scales allow us to do, and typologies are more
common in qualitative research (you go in you do not have any idea what you are looking
at but once you are down in the field you really get a feeling of what the data look like.
Example: Looking at the ads of TV to find out what is going on, typology is good to find
pattern what is the certain typology and is used in data organization.
The logic to use composite measure is straightforward because a lot of time when you are
looking at social phenomenon there not a single good indicator for a certain concept.
Example: concept of depression is very complicated it is complex because you can not
say that someone is depressed just because they feel blue which is an indicator of
depression. However there are a variety of measures that indicate depression. All bunch
of indicators such us I feel blue, I feel bad, I am always sad. I cry for no reason whole
bunch of indicators for concept of depression because no one single indicator can
describe the concept of depression. Similarly concepts need variations, there are different
types of variation. Example environmental activism, there are different types of activism,
one single indicator is not the best indicator for all different types of activism. So you
will be writing different indicators to get a field of different types of concept.
Index definition: a composite measure that reflect a concept and each item summarizes
and rank-orders several items (aka observations, indicators) that reflect a concept; each
item presumed to demonstrate the same intensity of a concept (BB).
Example; ”Environmental activism” Tindell looked at the environmental activism. He
looked at the different ways that we can actually conceptualize activism. So environmental activism is this broad big umbrella but there are different indicators that
can suggest activism, four examples of environmental activism are on slide 7. There are
different dimension that can measure environmental activism. Dimensions are useful
because if we are dealing with a broad concept such as environmental activism any one
of the indicators on slide 7 would it suffices. So if you say you want to do a project on
environmental activism and presume that buying a book is an environmental indicator,
that is one dimension, but the other ones also count therefore Tindell come with index of
different ways of someone can be an activist. And four different ways of activism is
weighted in different ways or the same.. Every index on the indicator is presumed to have
the same amount to it. Key to indicator is that every dimension is presumed to have a
relation to it.
Constructing index: The first thing is selecting the items (indicators) each of them are
items or indicators for an index what is really important is the issue of validity does this
item make sense, is this a valid measure of a pre defined concept for instance
environmental activism that we are trying to measure. The measure should always reflect
the concept you study. Second of all is that you want to make indexes that measure only
one concept, for instance environmental activism is one concept that you want to
measure. There are some times really really broad concepts so for instance concepts such
as gender inequality and racial inequality these broad concepts you might need to define
them better and make sure that the indexes that you are creating actually only measure
one aspect of that really big concept. So if you take gender inequality as an example that
could be gender inequality any thing. When you are constructing an index there will be so
many different indicators of gender inequality, indicator of gender inequality in the house hold in the work place the school the media etc. etc. this index can get out the hand what
you will be doing instead to really break down the concept and make an index based on
that specific definition. You will be measuring more than one dimension of it, if you do
not specify and the key is when you measuring that you only measure one dimension.
Second point, when you are constructing an index is you are measuring in effect.
Example: Environmental activism is most likely dependent variable. Likely there is a
question what factors lead to environmental activism and environmental activism could
be depended variable deeper effect, is being measured within index. You write an index
when it is a dependent variable not independent.
Each item has to make sense. Each item in the index have a relationship with one and
other, if this is not the case it means that this is not measuring the concept in question any
way. If you were writing some statistics on it you will find out that way. Second thing
that is important to keep in mind is that if the two variables correlate with one and other
completely, there is really two indexes measure the same thing you will drop them, chose
the better one.
Another point that is important to keep in mind when you are constructing an index is
how valid is your index in terms of internal and external validity. So if your internal
validity is good each one of those items should be able to predict one and other. Example:
if the index was internally valid it would make sense that answering yes to one item mean
answering yes probably to another items.
If you run perfect correlation and you see one it means exactly the same. They should be
close enough External validity mean if there were another index that is talking about the environmental
activism the indicator of one index is able to predict the result of sec