30.11.

2023

Choosing an IV

How to Decide Which Variables to • If you want to know whether people press a button in response to a
light more quickly when a tone is given as a warning signal, the IV is:
Manipulate and Measure § The presence or absence of the tone.

• If you want to find out whether children are more aggressive after
exposure to violent versus nonviolent television programs, the IV may
be harder to define.
Experimental Psychology 2023-2024 § Not everyone would agree on a particular definition.
Dr. Derya Durusu Emek-Savaş

1 2

Defining Your IV Defining Your IV

• Experimental psychologists should provide operational definitions of • Many psychological questions require complex operational
the IV and DV. definitions:
§ Must specify the operations anyone must go through to set up the — Do people whose mothers are affectionate make more successful
IV in the same way as they did. marriage partners?
— Do students learn more from popular professors?
— Does a worker’s morale affect work output?
• Psychology researchers have more difficulty agreeing on operational
definitions than do physical scientists. — Does anxiety cause depression?
• Before doing an experiment to answer any of these questions, you
need operational definitions for the terms.

3 4

1
 30.11.2023

Defining Your IV Choosing The Range Of Your IV

• Once a concept is operationally defined, the definition carries some • Once you have defined your IV, you have to choose the range
status, and it is sometimes difficult to convince others that a new one of the variable
is needed. — The range is the difference between the highest and lowest level
of the variable you choose.
• As you begin trying to operationally define the terms for your — Large vs. small ranges
experiment, be sure to do a literature search and find out how others — How do we determine what the range should be?
have defined the concepts you wish to investigate.

5 6

Choosing The Range Of Your IV Choosing The Range Of Your IV

Be realistic! Select a Range That Shows Effect
• Try to choose a range that is realistic in that it is similar to • Within realistic limits, you should have a range that is large enough to
the levels found in the situation you will be generalizing to. show an effect of the IV on the DV if such an effect exists.
• Real-world experiments require special attention be paid to choosing
a large enough range.
• Avoid sledgehammer effects — Because the experimenter does not always have complete
— Setting the levels of the IV at such extremes that you are certain control over the levels of the IV.
to find a difference in behavior. — You can choose an approximate level, but the actual level may
vary from trial to trial.

7 8

2
 30.11.2023

Choosing The Range Of Your IV Choosing The Range Of Your IV

Do a Pilot Experiment Do a Pilot Experiment
• During your literature search you may find experiments that use the • Because you don’t need to report the results, you may break some of
same IV you are planning to use that can give you an idea about an the rules here.
appropriate range. § You might use your friends or yourself as participants
§ You can also change the levels of your IV halfway through a trial,
• However, if your experiment is original and nobody else has used an stop the experiment, or do only part of the experiment, depending
IV similar to yours, you may choose to do a pilot experiment on what you learn as you proceed.
— A small-scale version of the experiment you are planning.
— By performing a pilot, you can eliminate any problems before you
proceed.

9 10

Choosing The Range Of Your IV

Choosing a DV
Do a Pilot Experiment
• When doing a pilot, you will sometimes find that what looked good • A measure of behavior
on paper just does not work.
• We could choose an infinite number of behaviors to measure
• A pilot may also help you determine whether the levels of your IV are
what you expected. • In selecting our DV, we must decide what we will measure
§ Levels that seem realistic during the planning stage may seem
unrealistic in the laboratory.
• The pilot becomes the guide for future experiments.

11 12

3
 30.11.2023

Choosing a DV Reliability and Validity

• We need to determine an operational definition with DVs • A measuring instrument is perfectly reliable if we get the same result
when we repeat the measurement several times under comparable
• But we also must know whether the measurement is reliable conditions.
and valid
• Formally determining reliability is particularly important when the DV
is the score from a test instrument
— a test of achievement, aptitude, or personality traits
• The reliability of a standardized test will already have been tested
• If you use a test or questionnaire that you constructed, you have to
determine its reliability yourself

13 14

Reliability Reliability
Test-retest reliability Alternate-form and Split-half
• The same test is simply repeated on the same group at a later time • A second way is the alternative-form method
• The reliability is determined by calculating a correlation coefficient § A second test having items similar to the first is constructed and
given to the same people
using two scores from each test taker.
§ Again, the two scores for each person are correlated
— However, the score on the second test given to the same person
can be contaminated by the previous testing
— The events occurring during the interval between test • A third way of establishing reliability is to use the split-half technique
administrations can also influence scores § A single test is statistically split into halves (such as using odd
versus even questions) and scores for the two halves are
correlated.

15 16

4
 30.11.2023

Choosing a DV
Reliability and Validity
• If your DV is not a test score, you may not have to formally
determine its reliability
• Nevertheless, you should be aware of the necessity of having
a reliable measure

17 18

Validity Validity

• Validity refers to confirming whether we are measuring what • Face validity

we want to measure — The easiest and weakest form of validity
— We need to know whether our measuring instruments measure — On the surface, it looks as if the test measures what it is
in the same units as does a standard measuring device known to be supposed to
valid — Subjective and superficial
— When a test score is used as a DV, it is necessary to formally — Does not include objective measurements
establish the test’s validity and its reliability.

19 20

5
 30.11.2023

Validity Validity

• Content validity • Predictive validity

— A more formal and defensible validation procedure — To determine whether the test successfully predicts some specific
— The subject matter covered by the test is carefully analyzed in criterion
detail for its content — The tests for entrance to college are valuable in partially
— The test is then designed so that it contains a representative predicting the criterion of their college grade point average (GPA)
sample of questions from each content area identified

21 22

Validity Directly Observable DVs

• Concurrent validity • The closer you can come to directly observing a behavior, the less
— Also established by comparing the test score with a criterion, but controversy there will be over your measure
in this case the two measures are taken at the same time • However, if your interest is in determining the workings of the human
mind, you should recognize that all dependent measures are indirect
§ Memory experiments - How much do people remember?
— Spontaneous recall
— Recognition
— Savings

23 24

6
 30.11.2023

Directly Observable DVs Single DV

• DVs, even those that at first appear to be directly observable, • Suppose we want to know whether
may be linked only indirectly to the behavior you are people respond more quickly to a bright
interested in. light than to a dim light when signaled to
push a button.
• We would probably start a clock when a
light goes on and stop the clock when
each participant presses a button.

25 26

Single DV Single DV
• In this experiment, only one characteristic of the response is being • From this diverse set of responses, we chose to measure only one
measured characteristic of the response:
• We could have chosen any number of other characteristics: — time from light onset to button press
§ How people press the button?
§ Does an individual move her finger from the side of the button on one trial • In other words, we selected a single dependent variable
and from directly over the button on the next?
§ On one trial, does she miss the button on the first try? — Any single DV we choose may or may not be the appropriate
§ On another trial, does she hit the button lightly at first and then mash it measure to take
down?

27 28

7
 30.11.2023

Multiple DVs Multiple DVs

• One way of improving the chances that we are choosing appropriate • Many types of research use choice reaction time as a DV
behaviors to measure in our experiment is by using multiple — The time it takes to give one of several responses when one of several
dependent variables stimuli occurs
— In some areas of experimental psychology, it is considered quite — If people want to make as few errors as possible, they must respond
inappropriate to report only one DV rather slowly
— If they are willing to be less accurate, they can respond more quickly

• This speed–accuracy trade-off makes it necessary that both speed

and accuracy be reported as DVs.

29 30

Composite DVs Composite DVs

• Although reporting as many aspects of behavior as possible is • A number of areas in experimental psychology, such as intelligence
generally a good idea, this practice can make interpreting the results testing, use composite DVs
much more difficult — In WAIS, the IQ is a composite of two subscales, a verbal scale
• Suppose we have 4 DVs: and a performance scale.
— One measure shows great improvement across conditions, two stay § The score on each of these subscales is a composite made up of subtests.
the same, and one decreases slightly.
• To say anything about the overall change in behavior, we need a way • The idea behind intelligence testing is that having a single measure
of combining our single DVs into a composite dependent variable that characterizes intelligence in general is useful.
that will give some indication of overall performance

31 32

8
 30.11.2023

Composite DVs Indirect DVs

• A second type of composite DV combines several instances of a single • Sometimes, directly observing the behavior you are
measure interested in is impossible
— These instances are taken at different times or under different — How can we do scientific research in such areas as emotion,
conditions learning, or intelligence?
§ Percent savings used in memory research — We need an indirect variable that changes along with the
internal behavior we are interested in.
• Through this type of composite DV, you can use a single number to
show the effect of a change caused by the IV

33 34

Indirect DVs Indirect DVs

Physiological Measures Physiological Measures
• The most popular types of indirect variables • When we use physiological measures to infer internal states, we are
— Based on the idea that if the behavior is a private event, such as assuming that a unique physiological pattern accurately reflects an
an emotion, the physiology of the body will change along with the internal state.
private event — A polygraph, measures four
— Because modern technology allows us to observe physiological physiological processes
changes, experimenters use these changes to infer what the private § respiratory rate, heart rate, blood
event must have been pressure, and galvanic skin response
— These measures are used to
determine whether a person is telling
the truth

35 36

9
 30.11.2023

Indirect DVs Indirect DVs

Physiological Measures Behavioral Measures
• In the past several decades, one of the fastest-growing areas of • Behavioral measures can also be used as indirect dependent
psychology is brain imaging, in which the activity of the brain can be variables
measured as various tasks are being carried out. — Similar to physiological measures, changes in the way a person
— EEG, ERPs performs a behavioral task can reflect the person’s internal state
— fMRI — Particularly important in some areas of cognitive psychology

37 38

Indirect DVs Indirect DVs

Behavioral Measures Behavioral Measures
• Cognitive researchers are interested in determining what goes on in the • The indirect behavioral measure is only as good as the assumptions
“black box” of the human mind during cognitive tasks that underlie it.
• Because all they really must work with are the inputs to (stimuli) and
outputs from (responses) the box, they have had to devise clever ways of
inferring what must be happening in the box
— Reaction time
§ The more information processed, the longer the response time will be
§ Single measure

39 40

10
 30.11.2023

Indirect DVs
Behavioral Measures
• In general, the more indirect the measure, the more elaborate the
underlying assumptions must be, and the less confident we can be of
our inferences.
— Indirect measures offer a way of investigating experimental
questions for which we have no direct measures.
— As long as we are aware of the assumptions we are making when
using indirect measures, they can be a valuable tool for helping us
get an idea of the nature of otherwise unobservable events.

41

11