You must read each slide, and complete any questions on the slide, in sequence.
Nonexperimental
A design in which there is no control or manipulation of the variables. This design does not seek to establish cause and effect and instead focuses on describing or summarizing what takes place.
Experimental Design
A design in which the experimenter controls and manipulates the independent variable and makes comparisons between the different levels, allowing the establishment of cause-and-effect relationships between the independent and dependent variables.
Dependent Variable (DV)
The variable measured in association with changes in the independent variable; the outcome or effect.
Independent Variable (IV)
The variable that influences the dependent variable. In experiments the researcher manipulates or controls this variable.
Experimental Group
The group or condition that gets the key treatment in an experiment.
Control Group
Any group or condition that gets a different form of the treatment in an experiment, or receives a baseline form of treatment.
Empty Control Group
Any group that does not receive any form of the treatment and just completes the dependent variable.
Experimental Realism
The degree to which a study participant becomes engrossed in the manipulation and truly influenced by it.
Mundane Realism
The degree to which a study parallels everyday situations in the real world.
Experimental Control
The technique of keeping everything between groups the same, except for the one element you want to test in an experiment.
Reliability
The stability or consistency of a measure.
Validity
The degree to which a tool measures what it claims to measure.
Sensitivity
The range of data a researcher can gather from a particular instrument.
Experimental Hypothesis
A clear and specific prediction of how the independent variable influences the dependent variable.
IRB
A board that reviews the ethical merit of all the human research conducted within an institution.
Descriptive
Describes what is happening.
Inferential
Tests a specific prediction about why something occurs.
Multigroup Design
This activity will allow you to create a design comparing multiple groups in order to test the impact of different body shapes of dolls on girls' body dissatisfaction.
Dr. Melanie Maggard
Dr. Natalie J. Ciarocco, Monmouth University
Dr. David B. Strohmetz, Monmouth University
Dr. Gary W. Lewandowski, Jr., Monmouth University
Something to Think About...
dirk enters/imageBROKER/AGE Fotostock
Scenario: We are constantly exposed to images of thin people in the media, through advertisements, movies, television, magazines, and online. These images, whether real or fictional, demonstrate the value our culture places on thinness. Being thin is equated with being beautiful and popular, attributes that girls, in particular, are taught to start valuing at a very young age. Even the toys girls are given—such as dolls with overly thin or unrealistic body types—communicate cultural pressure to be thin.
Something to Think About...
A Kompatscher/F1online/AGE Fotostock
When girls are taught to compare themselves with others, they might begin to believe that, unlike the celebrities they see on TV or the dolls they play with, they are not thin or pretty enough. How might constant exposure to messages about the “thin ideal” influence young girls? Could dolls with unrealistic body proportions influence how satisfied or dissatisfied girls are with their own bodies? It might be possible that a short amount of time spent playing with these dolls is all that is needed for unrealistic body images to negatively impact how girls view their bodies.
Our Research Question
Jesse Kunerth/iStock/Getty Images
Based on your experiences with or knowledge of dolls that portray thinness, you can develop a research study that examines their impact on body dissatisfaction. But first, you will need a framework to help you explore this topic. Research studies all start with a question, so here is your chance to ask one of your own.
Now that you have a research question (“Does exposure to dolls with unrealistically thin body proportions lead to greater body dissatisfaction in girls?”), you must decide which type of research design will best answer your research question. To narrow things down, consider the following:
Now that you know you have an experimental design that compares exposure to extremely thin dolls, realistic-looking dolls, and no dolls, you can identify your independent and dependent variables.
Because you have an experiment with 1 independent variable and 3 levels (extremely thin dolls vs. realistic-looking dolls vs. no dolls), you have a multigroup design, or experiment with 3 groups. Next we should identify the experimental and control groups.
Next, we need to operationally define the independent variable (IV) of doll type by determining exactly how we will manipulate it. As we do, we’ll want to be sure our study has a high level of experimental and mundane realism.
It looks like the task that is highest in experimental and mundane realism involves girls playing with virtual dolls that they are allowed to dress with various outfits and accessories. We know that our experimental group will receive the treatment (i.e., the extremely thin dolls) we are interested in, so now we need to define our control group, which will have the most experimental control.
You have now established the key comparison between the extremely thin dolls vs. realistic-looking dolls vs. no dolls groups. Next, we need to specify the exact nature of our dependent variable, “body dissatisfaction.” First, consider the following:
We know we want to use a self-report measure to measure body dissatisfaction. Now it is time to determine which type of self-report measure to use. Keep in mind the ideal number and types of questions, reliability and validity, sensitivity, and appropriateness of the measure for young girls.
Now that you have determined how you will collect your data and your intended sample, you must submit your research procedure to the Institutional Review Board (IRB) for ethical approval. The IRB or ethics board will determine whether or not your study meets all ethical guidelines.
IRB
Each IRB has its own protocol which conforms to the national standard when a researcher submits an application for proposed research to be reviewed. In addition to the appropriate paperwork and other information submitted to the IRB, the board would consider the following description during their evaluation of your proposed experiment:
The purpose of this research is to determine whether exposure to unrealistically thin dolls leads to higher body dissatisfaction than does exposure to realistically-proportioned or no dolls. To study this topic, 8- and 9-year-old girls from a local elementary school will be randomly assigned to an extremely thin dolls, realistic-looking dolls, or no dolls group. Those in the thin and realistic dolls groups will be presented with and allowed to dress virtual dolls with outfits and accessories for 5 minutes. Researchers will then measure all participants’ body dissatisfaction using a 2–3 minute self-report measure, the Body Parts Dissatisfaction Scale (BPDS).
Responding to the IRB
wk1003mike/Shutterstock
The IRB reviewed your submission and has one concern. Although the study appears to present minimal risk to participants, there is no mention of how informed consent will be obtained for the children in the study.
You must now determine how to respond to the IRB, keeping in mind the ethics of surveying vulnerable populations such as children.
Now that we have secured the IRB’s approval, we should determine what the entire study will look like. Below are the steps of the study; can you place them in the proper order?
1.
2.
3.
4.
5.
6.
Give participants the Body Parts Dissatisfaction Scale.
Obtain informed consent and verbal assent.
Allow participants in the extremely thin doll and realistic-looking doll groups to dress virtual dolls for 5 minutes.
Debrief the participants.
Give participants instructions for how to dress the virtual dolls if they are in the extremely thin doll or realistic-looking doll groups.
Randomly assign participants to experimental, control, and empty control groups.
Collecting Data
Image Source/Photodisc/Getty Images
Now that you have a sense of how to conduct this study, it is time to see what data from this study might look like.
If you were to run a full version of this study, you would want to have at least 30 participants in each of your three groups, for a total of 90 participants. Because you have a between-subjects design, each participant will only be in one group.
Example Data Set
This is an example of what your data set would look like. The top row shows the variable names; the other rows display the data for the first 10 participants.
In the “Group” column, a 1 = Thin Dolls Group, a 2 = Realistic Dolls Group, and a 3 = No Dolls Group. The Body Parts Dissatisfaction Scale score was recorded under BPDS and represents the number of body parts the participant would like to change to be either smaller or bigger.
Participant
Group
BPDS
101
1
6
102
1
6
103
2
2
104
3
1
105
2
1
106
1
5
107
3
2
108
1
4
109
2
2
110
3
2
Selecting the Proper Tool
Ines Koleva/E+/Getty Images
Now that you have collected your data, you must decide the best way to summarize your findings. The decisions you made about how to collect your data dictate the statistics you can use with your data now. First, you need to consider if your study is descriptive or inferential.
The following is an example of output for another three-group design. This study was about how hours slept at night (less than 7 hours, 7–9 hours, and more than 9 hours) influence self-reported happiness.
To report these numbers in a results section, put the numbers in as follows:
F(#,#) = #.##, p = .##, eta2 = .##.
Click on the table below to learn more about each element of the output.
ANOVA
Sum of Squares
df
Mean Square
F
Sig.
Between Grops
27.899
2
13.949
30.145
.000
Within Groups
44.424
96
.463
Total
72.323
98
This is the df or degrees of freedom. An ANOVA has two dfs, one for the main effect (between-groups) and one for the error (within-groups).
This is the F statistic. It represents the size of the difference between group means compared to the size of the difference within groups
This is the p level or the significance level. It represents the probability or likelihood that the results happened by chance. The lower the p level, the less likely the result happened by chance.
The F score and p level will only tell you whether there is a significant difference. To determine which means are different, and the nature or direction of those differences, you need to look at the means via a post-hoc test.
Measures of Association
Eta
Eta Squared
happiness * Condition
.621
.386
The eta squared (eta2 ) is the effect size. It tells us the proportion of change in the dependent variable that is associated with being in the different groups of the independent variable.
The eta squared (eta2 ) is the effect size. It tells us the proportion of change in the dependent variable that is associated with being in the different groups of the independent variable.
Tutorial: Evaluating Output
Alex Brylov/Shutterstock
Click on the table below to continue learning about each element of the output.
Multiple Comparisons
(I) Condition
(J) Condition
Mean Difference (I-J)
Std. Error
Sig.
95% Confidence Interval
Lower Bound
Upper Bound
Less than 7 hours sleep
7-9 hours sleep
-1.06061*
.16747
.000
-1.4593
-.6619
More than 9 hours sleep
.12121
.16747
.750
-.2775
.5199
7-9 hours sleep
Less than 7 hours sleep
1.06061*
.16747
.000
.6619
1.4593
More than 9 hours sleep
1.18182*
.16747
.000
.7831
1.5805
More than 9 hours sleep
Less than 7 hours sleep
-.12121
.16747
.750
-.5199
.2775
7-9 hours sleep
-1.18182*
.16747
.000
-1.5805
-.7831
*. The mean difference is significant at the 0.05 level.
This is the difference between the means for those who sleep less than 7 hours
and between 7 and 9 hours.
This is the difference between the means for those who sleep less than 7
hours and more than 9 hours.
This is the difference between the means for those who sleep between 7 and 9
hours and more than 9 hours.
The post-hoc test tells us which comparisons between the means were
significant. The p level tells us the significance level of that comparison.
Those who sleep between 7 and 9 hours are happier than those who sleep less
than 7 hours and those who sleep more than 9 hours.
Those who sleep less than 7 hours and more than 9 hours were not different
but were equally happy.
Those who sleep less than 7 hours and more than 9 hours were not different but
were equally happy.
Descriptive Statistics
Condition
Mean
Std. Deviation
N
Less than 7 hours sleep
3.0303
.80951
33
7-9 hours sleep
4.0909
.72300
33
More than 9 hours sleep
2.9091
.45851
33
Total
3.3434
.85906
99
This is the average or mean (M) happiness rating for those who sleep less
than 7 hours.
This is the standard deviation (SD) of happiness for those who sleep less
than 7 hours.
This is the average or mean (M) happiness rating for those who sleep between
7 and 9 hours.
This is the standard deviation (SD) of happiness for those who sleep between
7 and 9 hours.
This is the average or mean (M) happiness rating for those who sleep more
than 9 hours.
This is the standard deviation (SD) of happiness for those who sleep more
than 9 hours.
In this case the means tell us that those who got 7–9 hours of sleep were
happier than those who got less than 7 hours of sleep and those that got more than 9 hours of sleep. The
results from the post-hoc test support the finding that this group (those who got 7–9 hours of sleep) was
statistically different from the other two groups.
Your Turn: Evaluating Output
Below is the output from your study:
ANOVA
Sum of Squares
df
Mean Square
F
Sig.
Between Groups
256.822
2
128.411
303.306
.000
Within Groups
36.833
87
.423
Total
293.656
89
Measures of Association
Eta
Eta Squared
BPDS * condition
.935
.875
Multiple Comparisons
(I) Condition
(J) Condition
Mean Difference (I-J)
Std. Error
Sig.
95% Confidence Interval
Lower Bound
Upper Bound
Thin
doll group
Realistic doll group
3.600*
.168
.000
3.20
4.00
Control group
3.567
.168
.000
3.17
3.97
Realistic doll group
Thin doll group
-3.600*
.168
.000
-4.00
-3.20
Control group
-.033
.168
.979
-.43
.37
Control group
Thin doll group
-3.567*
.168
.000
-3.97
-3.17
Realistic doll group
.033
.168
.979
-.37
.43
Descriptives
N
Mean
Std. Deviation
Std. Error
Thin doll group
30
5.07
.868
.159
Realistic doll group
30
1.47
.507
.093
Control group
30
1.50
.509
.093
Total
90
2.68
1.816
.191
Your Turn: Evaluating Output
Alex Brylov/Shutterstock
Based on the results of your statistical analyses, match the correct number in the “Answer” column to the term requested under “Prompt”:
Prompt
Answer
F for the ANOVA test
df for the main effect of group (between-groups)
df for error (within-groups)
p for the ANOVA test
p for the difference between thin doll and realistic doll groups
p for the difference between thin doll and no doll group
p for the difference between realistic doll and no doll groups
eta2
Activity: Graphing Results
BPDS
Descriptives
N
Mean
Std. Deviation
Std. Error
Thin doll group
30
5.07
.868
.159
Realistic doll group
30
1.47
.507
.093
Control group
30
1.50
.509
.093
Total
90
2.68
1.816
.191
Drag the bars of each graph to the correct Mean value.
Dolls & Body Dissatisfaction
Doll Group
Your Turn: Results
gguy/Shutterstock
Now that you have worked with your data, you must determine the best way to express your findings in written form. You must be sure that how you describe your findings accurately represents the data.
Now that you have determined how to express your findings in a scientifically responsible way, you also need to be able to talk about what your findings mean in everyday terms so that the world can benefit from your science.