You are on page 1of 11

MOD4: Research Techniques variable called the dependent

Experimental Methods variable (DV).

John Stuart Mill  An experiment must have at least


To rule out the "z" variables, he two values, or levels, of the
proposed that an effect should be environment. The point is that at
present when the cause is present and least two conditions must be
that when the cause is absent, the compared with each other to
effect should also be absent. determine if the independent
variable produces a change in a
In other words, the only way to infer behavior or outcome.
causality is through comparing two E.g.,:
controlled situations: one in which the IV- Temperature
cause is present and one in which the Levels:
cause is absent.  Hot- 18°
Cause Effect  Room temperature- 30°
 Cold- 23°
Manipulated Measured
DV- Abstract Reasoning
Independent Dependent
Variable Variable Note: It is important to give the IV
-In true -Something you levels an operational definition
experiments, the can measure, based on the literature.
IV should be not just observe.
something you (can be Advantages of Experiment
can induce. measured  The control of extraneous variables
through self- (variables that may contaminate
reports/ the data, thus altering the results
standardized of the study)
tests..)
 In the ideal experiment, no factors
- Measure: Ratio (variables) except the one being
(absolute zero) studied are permitted to influence
the outcome; in the jargon of
experimental psychology, we say
What is an Experiment? that these other factors are
 Experimental designs are those controlled.
where the experimenter
manipulates one variable called the
independent variable (IV) to see
what effect this has upon another
 Another advantage of experiments they will cause changes in behavior.
is economy (productive and Failure of an independent variable to
efficient) control behavior, often called null
 The experimenter controls the results, can have more than one
situation by creating the conditions interpretation.
of interest thus obtaining data  To avoid this, experimenters
quickly and efficiently. must be careful to produce a
strong manipulation of the
independent variable.
Why Experiments are Conducted
 Critical experiments try to pit -Failure to do so is a common cause of
against each other two theories null results. Because there is no way
that make different predictions. to determine if the manipulation failed
One outcome favors theory A; the or the null results are correct.
other, theory B.

 Thus, in principle, the experiment 2. Dependent Variable


will determine which theory to - A variable thought to be affected by
reject and which to keep. changes in an independent variable.
You can think of this variable as an
 Less often, researchers perform an outcome.
experiment in the absence of a
compelling theory just to see what -The dependent variable is the
happens; we can call this a what-if response measure of an experiment
experiment. that is dependent on the subject’s
response to our manipulation of the
Variables environment.
Most hypotheses can be expressed in
terms of two variables: a proposed - One criterion for a good dependent
cause and a proposed outcome. variable is stability (replicability; kahit
ulitin sa ibang group yung experiment,
Two Important Variables in Research same pa rin ang results)
1. Independent Variable
- A variable thought to be the cause of Floor Effect
some effect. This term is usually used Is when most of your subjects score
in experimental research to describe a near the bottom. There is very little
variable that the experimenter has variance because the floor of your test
manipulated. is too high. (DV is too easy [if test]; low
stability)
- Independent variables are selected
because an experimenter believes Ceiling Effect
Is the opposite, all of your subjects the other variables that we are
score near the top. There is very little interested in but we may have failed to
variance because the ceiling of your take these into account when
test is too low. (DV is too hard [if test]; designing our study.
low stability) Confounding Variable
A confounding variable is a specific
Intervening Variable type of extraneous variable that is
- Intervening variables are abstract related to both of the main variables
concepts that link independent that we are interested in.
variables to dependent variables.
Confounding Variable (Third
E.g.,: Variable)

IV DV

IV DV
Mediating Spurious Correlation
Variable
What first looks like a causal
Ito usually silbi ng conceptual relationship between IV and DV is
framework (quanti), a model of your ultimately spurious. The confounding
variables’ relationship. While variable is the hidden explanation.
theoretical framework (quali), shows
the theory behind your paper. Control Variable
- Is a potential independent variable
- An intervening variable (sometimes that is held constant during an
called a mediating variable) is a experiment because it is controlled by
hypothetical variable used to explain the experimenter.
causal links between other variables.
- In practice, an experimenter tries to
- Sometimes the cause and effect control as many salient variables as
relationship between independent and possible, hoping that the effect of
dependent variables is not obvious uncontrolled factors will be small
without the discovery of an intervening relative to the effect of the independent
variable. variable.

Extraneous Variables More than One Independent Variable


- Extraneous variables are those
variables that might have an impact on
1. Efficiency – one experiment with Possible DVs:
more than one IVs than separate Retention of meaningful information
experiments for each. after reading text,
Time needed to read a fixed number of
2. Experimental control is often better. words,
Number of errors in recognizing single
3. The results generalized across letters,
several IVs are more valuable. Speed in transcribing or retyping text,
Heart rate during reading,
4. Allows us to study interactions, the Muscular tension during reading
relationships among independent
variables  Despite this, the typical experiment
uses only one, or at the most, two
 When the effects produced by one dependent variables
independent variable are different simultaneously.
at each level of a second
independent variable, we have an  The reason why more dependent
interaction. variables are not used is probably
because it is statistically difficult to
 Remember, an interaction between analyze several dependent
two independent variables variables at once.
indicates that effects produced by
one independent variable (E.g., Experimental Designs
belief in God) are not the same at Essential characteristics of a true
each level of a second experiment:
independent variable (E.g., source 1. Manipulation or Intervention
of the passage- ancient scroll, the 2. Control
bible). 3.Random Assignment- equal
chance to be assigned on either
More than one Dependent Variable experimental or control group. Note:
 It indicates how well or poorly the Random assignment is different from
subject is performing. It permits the random sampling. Sa random
experimenter to score behavior. sampling, lahat ng nasa population
The experimenter must decide has an equal chance to be chosen
which aspects of behavior are as a sample.
relevant to the experiment at hand.
Between-Subjects Design
E.g.,: Let us say we wish to study the - Between- group design is an
legibility of the texts that you are now experiment that has two or more
reading. groups of subjects each being tested
by a different testing factor -CONS: More people are needed
simultaneously. Differences between the participants
may affect the result.
- Assign only some subjects to each
level. (Tested simultaneously) Participant/Subject Variables
These extraneous variables are
E.g.,: You have a total of 45 related to individual characteristics of
participants to test if room temperature each participant that may impact how
is causally linked to abstract he or she responds. These factors can
reasoning. include background differences, mood,
anxiety, intelligence, awareness and
For Between-subjects design, you other characteristics that are unique to
divide your participants into groups each person.
and expose each group to a different CONTROL: after the participants have
level of temperature. been recruited, they should be
Expose 15 to randomly assigned to their groups.
Hot
Temperature
Expose 15 to
Equivalent Groups
45 Participants Room - This method of assigning subjects to
Temperature experimental conditions has no bias,
Expose 15 to since it ignores all characteristics of
Cold
Temperature the subjects; we expect that the
groups created would be equivalent on
- There is no chance that one any and all relevant dimensions.
treatment will continue to contaminate
the other, because each person Baseline
receives only one treatment. - A baseline is a measurement of the
variable of interest at the beginning of
Pros and Cons of Between-Subjects treatment or a study that is used to
Design compare to later measurements in
- Drawback: the between-subjects order to judge the effectiveness of the
design must deal with differences treatment or conditions.
among people, and this decreases its  E.g., Pre-test & Post Test (In
efficiency. (Can be addressed by some references, this is not
random assignment) considered a true
experiment)
-PRO: Avoids order effects as people
participate in one condition only. Matching
- A technique where one member of
each pair would be randomly assigned
to one group and the other member to
the second group. (This is used to
balance the experimental and control E.g.,
groups)
 E.g., You assign two of your
respondents who have high
IQs into separate groups

- One difficulty with matching is that an


experimenter cannot match subjects
on every possible characteristic.

Other Cons:
Within-Subjects Design
- A within-subject design is a type of
1. Order Effects - The order in which
experimental design in which all
the participant receives a treatment
participants are exposed to every
(first, second, etc.) will affect how the
treatment or condition.
participant behaves.
E.g., You exposed all 45 participants
2. Practice Effects - after doing the
to all levels of temperature (hot, room
dependent-measure task several
temp. and cold)
times, a participant’s performance may
improve. This improvement might be
Pros and Cons of Within-Subjects
incorrectly attributed to having
Design
received a treatment.
- PROS: Participant variables are 3. Fatigue Effects- decreased
reduced. Fewer people are needed. performance on the dependent
measure due to being tired or less
- CONS: There may be order effects. enthusiastic as the experiment
(Can be addressed by continues. This decrease in
counterbalancing) performance might be incorrectly
attributed to a treatment.
CONTROL: Should counterbalances
the order of the conditions. 4. Carryover Effects- the effects of a
treatment administered earlier in the
Counterbalancing – removes experiment persist so long that they
confounding variables from an are present even while participants are
experiment by giving slightly different receiving additional treatments.
treatments to different participant
groups.
5. Sensitization- - after getting several  Control condition (within-
different treatments and performing the subjects design).
dependentvariable task several times,
participants may realize (become - Sometimes the best baseline is no
sensitive to) what the hypothesis is. treatment, but often the best baseline
 This is why some requires some activity.
experiments perform
deception, to avoid bias from - A frequent example occurs in
the participants. memory research, where a group of
subjects is required to learn two
different lists of words. The
experimenter is interested in how
learning one list interferes with
Mixed Design learning the other.
- Experiments need not be exclusively
of within-subjects or between-subjects - A control condition with no treatment
design. would have a control group learn list A,
then sit around for the time it took the
- It is often convenient and prudent to experimental group to learn list B, and
have some independent variables then be tested again on list A.
treated as between-subjects and
others as within- subjects in the same - A proper baseline condition would
experiment. occupy the control group during the
time the experimental group was
learning list B; perhaps the
Control and Experimental Group experimenter would have them do
 Independent variables must be arithmetic or some other “busy work”
varied (or manipulated) by the that would prevent rehearsal.
experimenter. This implies that
each and every independent Control Group
variable must vary either in amount - The control group is composed of
(quantitative variation) or in kind participants who do not receive the
(qualitative variation) within the experimental treatment. When
experiment. conducting an experiment, these
people are randomly selected to be in
Control Conditions this group.
- Many experiments contain, in
addition to independent variables, - This group serves as a benchmark,
some: allowing researchers to compare the
 Control group (between- experimental group to the control
subjects design) group to see what sort of impact
changes to the independent variable experimenters cannot communicate
produced. what they do not know.

Pitfalls in Experimental Research A. Single Blind: The researcher know


the participants’ groups.
1. Demand Characteristics
Many times, research participants B. Double Blind: This procedure is
spontaneously form hypotheses or termed a double-blind experiment
assumptions about the experimenter’s because neither the experimenter nor
purpose in conducting the experiment the research participant knows which
and then behave or respond in a way subjects are in which treatment
that will satisfy this purpose. conditions.

To the extent that the behavior of First, since the participants do not
research participants is controlled by know which group they are in, their
demand characteristics instead of by beliefs about the treatment are less
independent variables, experiments likely to influence the outcome.
are invalid and cannot be generalized Second, since researchers are
beyond the test situation. unaware of which subjects are
receiving the real treatment, they are
2. Hawthorne Effect less likely to reveal accidentally subtle
The alteration of behavior by the clues that might influence the outcome
subjects of a study due to their of the research.
awareness of being observed.
C. Automation Experiments
3. Experimenter Effect Experimenter effects can be eliminated
- Influences the outcome accidentally or greatly reduced by having
by providing participants with slight computers or other equipment conduct
cues as to the experimenter’s the experiment so that the subject is
expectations. untouched by human hands.

The gender, race, and ethnicity of the Quasi-Experimental Designs


experimenter are also potential  Quasi-experimental research
experimenter effects. involves the manipulation of an
independent variable without the
Eliminating Experimenter Effect random assignment of participants
The best way to eliminate this kind of to conditions or orders of
experimenter effect is to hide the conditions.
experimental condition from the
experimenter on the premise that  A quasi-experiment is simply
defined as not a true experiment.
Since the main component of a as natural disasters or unnatural
true experiment is randomly disasters, are both physically and
assigned groups, this means a morally difficult to implement.
quasi-experiment does not have
randomly assigned groups.  The technique here is similar to the
 Random assignment ex post facto examination in
helps ensure that there is correlational research, except that
no pre-existing condition two or more levels of the variable
that will influence the of interest are examined rather
variables and mess up the than correlated.
results.
Advantage of Quasi-Experiments
 Although the independent variable - The advantages of quasi-
is manipulated, participants are not experiments are obvious: They use
randomly assigned to conditions or naturally occurring independent
orders of conditions. variables, most of which have a high
degree of intrinsic interest and
 In terms of internal validity, quasi- important practical implications. (E.g.,
experiments are generally Sex, Age, Ethnicity)
somewhere between correlational
studies and true experiments. - The typical quasi-experiment has a
 Has lower validity than subject variable as an independent
true experiment. variable. E.g.,, age, sex, ethnic group.

 For one reason or another, many - Socially caused subject attribute can
variables cannot be manipulated be social class, region of residence.
directly. One deterrent to
manipulation of variables in - Disease-and-illness related subject
experiments is the ethical attribute can be mental illness, effects
considerations all scientists must of disasters.
have.
 Independent Variable
cannot be induced (E.g.,  An experimenter cannot
Depression) manipulate a subject variable while
holding other factors constant; she
 A second barrier to manipulation is or he can only select subjects who
Mother Nature. Some variables, already have the characteristic in
such as the sex of our subjects, some varying degree and then
cannot be varied by the compare them based on the
experimenter other variables, such behavior of interest.
education program on elementary
 When subject variables are school students’ attitudes toward
investigated, we cannot safely illegal drugs. The researcher could
attribute differences in behavior to measure the attitudes of students at a
this variable, as we can with true particular elementary school during
experimental variables. Such one week, implement the antidrug
designs, then, essentially produce program during the next week, and
correlations between variables. We finally, measure their attitudes again
can say that the variables are the following week.
related, but we cannot say that one
variable produces or causes the - It is unlike a within-subjects
effect in the other variable. experiment, however, in that the order
of conditions is not counterbalanced
Quasi-Experiments: Non Equivalent because it typically is not possible for a
Groups participant to be tested in the
- A nonequivalent groups design is a treatment condition first and then in an
between-subjects design in which “untreated” control condition.
participants have not been randomly
assigned to conditions. - If the average posttest score is better
than the average pretest score, then it
E.g., a researcher who wants to makes sense to conclude that the
evaluate a new method of teaching treatment might be responsible for the
fractions to third graders. One way improvement.
would be to conduct a study with a
treatment group consisting of one - If the posttest takes too long, it’s
class of third-grade students and a possible that the results won't be as
control group consisting of another valid because of history and
class of third-grade students. maturation. (You can’t be sure that it’s
the treatment that caused the result
Quasi-Experiments: Pre-test- Posttest because of the possibility of other
Design events.)
- In a pretest-posttest design, the
dependent variable is measured once Regression to the Mean
before the treatment is implemented - This is a statistical phenomenon
and once after it is implemented. stating that data that is extremely
higher or lower than the mean will
likely be closer to the mean if it is
Posttes measured a second time.
Pretest Intervention
t
E.g., a researcher who is interested in E.g., A bowler with a long-term
the effectiveness of an antidrug average of 150 who suddenly bowls a
220 will almost certainly score lower in
the next game. His score will “regress” E.g., Common cold If one were to
toward his mean score of 150. measure symptom severity in 100
common cold sufferers today, give
- It occurs whenever the sample or them a bowl of chicken soup every
subjects are chosen on the basis of day, and then measure their symptom
extreme pretest scores. Regression severity again in a week, they would
toward the mean is expected in these probably be much improved.
cases where there is non-random
sampling or assignment in Is the study a randomized controlled
experiments. trial?
Yes No
- It essentially occurs on the posttest Intervention No intervention
where the measures on the average Control Group Descriptive Non-
regress toward the mean on average. Random experimental
Sample
- The net effect of regression toward No control group or
the mean is that the lower scores (or randomization
measurements) on the pretest tend to Quasi-Experimental
be higher on the posttest, and the
higher scores (or measures) on the
pretest tend to be lower on the
posttest.

- It is an important phenomenon to
take note of in conducting experiments
because it affects the internal validity
of the experimental design.

- What happens here is that one can


end up concluding that the significant
difference or effect is due to the
treatment when in fact it is due to
chance and by this phenomenon.

Spontaneous Remission
- This is the tendency for many
medical and psychological problems to
improve over time without any form of
treatment.

You might also like