Data analysis proj. paper

Running head: INTELLIGENCE AND THE FEELINGS SCALE 1
The Impact of Intelligence on Task Expectancy and an Examination of the Feelings Scale
Samuel Dunham
Valdosta State University

INTELLIGENCE AND THE FEELINGS SCALE 2
Abstract
There is much interest in attitudes and how they impact our opinions of our abilities. However,
there is not much research that examines how perceived intelligence impacts individual task
expectancy. There are also few depression scales for adolescents and it is important to determine
what factors are contained in the National Longitudinal Study for Adolescent Health feelings
scale. A one-way ANOVA provided evidence that there were no practical differences in
perceived expectancy on the basis of intelligence. An EFA determined that there were four
factors contained in the Add Health scale: Sadness, Unfriendability, Fatigue, and an atheoretical
factor that only included the reverse coded items. People may have a tendency to believe in their
ability to achieve a task through hard work regardless of their intelligence level. The factors
extracted from the EFA should be examined in order to increase the reliability estimates of the
factors with lower reliability levels.

The Impact of Intelligence on Task Expectancy and an Examination of the Feelings Scale
In the field of social psychology, one important topic is how people perceive themselves,
and more specifically, how people perceive themselves in comparison to others. There is some
published research that looks to investigate how people’s perceptions of themselves impact their
expectancy in terms of their ability to successfully complete a given task, but not much.
However, there is a great deal of research on the topic of how individuals tend to make
inaccurate attributions about themselves. For example, Zell & Krizan (2014) found that people
tend to only have a moderate knowledge level of their abilities because they have a tendency to
ignore important contextual factors. Williams, Dunning, & Kruger (2013) found that people have
a tendency to overestimate their abilities on intellectual tasks, regardless of whether they are
completing the task correctly or not.
For the purposes of this study, data from the National Longitudinal Study for Adolescent
Health (Add Health) was examined and analyzed in order to test the proposed hypotheses
presented in this paper (Harris & Udry, 2013). The Add Health study was conducted using a
nationally representative sample of adolescent youths in the United States from the seventh to
twelfth grade. The study began during the 1994-1995 academic year and consisted of four waves,
with the most recent wave being done in 2008. All of the data taken for this paper was only taken
from Wave I. The study examined many different facets of adolescent life and recorded a large
amount of data in the four longitudinal waves (Harris & Udry, 2013).
In the present study, the factors of intelligence and hard work expectancy were examined
in order to see if there was a relationship between the two factors. Intelligence was defined as
“an individual’s perceived level of intelligence in comparison to other individuals.” The factor
was measured by participant answers to the following question, “Compared with other people
your age, how intelligent are you?” (Harris & Udry, 2013). Hard Work Expectancy was

operationalized as “what an adolescent believes they could achieve through hard work.” This
factor will be measured by participant responses to the following question, “When you get what
you want, it's usually because you worked hard for it” (Harris & Udry, 2013).
What makes these factors interesting is that intelligence is seen as an internal factor that
holds fairly constant throughout an individual’s life, while a person’s belief in what they can
achieve through hard work is also an internal factor, but one that has a tendency to change
throughout a person’s lifespan. Those changes are normally the result of past experiences, new
information, and/or a combination of many other factors.
It is conceivable to believe that people who view themselves as intelligent are more likely
to believe in their expectancy on a task through hard work. However, the inverse is also a
plausible option that should be considered. The logic of the inverse is that a person who views
themselves as intelligent would have a high expectancy in a given task, but the confidence would
be the result of their intelligence, rather than their hard work. While hard work is a factor that is
developed and probably earned, intelligence can be seen as a factor that comes more “naturally”
for the person possessing it since there is evidence to suggest that interventions do not change
intelligence levels. (Redick et al., 2013).
Another area of interest in psychology concerns the emotions/ feelings that people have,
despite the fact that they have a tendency to change. Our emotions drive us in many ways
throughout our lives and they are impactful. Many emotions/ feelings scales have been
developed for a wide variety of purposes. For instance, Beitchman (1996) developed a feelings
scale designed for children, while Diener et al. (2010) developed a scale for the purposes of
distinguishing between positive emotions, negative emotions, and the differentiation between the
two.

In this study, the feelings scale used in the Add Health study was adapted from the
original Center of Epidemiological Studies Depression scale (CES–DS), and it was examined in
order to determine how many underlying constructs were present in the scale (Radloff, 1977).
With 19 items in the scale and the knowledge that there are a wide variety of emotions, it will be
beneficial to see the number of feelings being represented by the scale. Having too many factors
could make the scale nearly uninterpretable, while having too few factors could prevent those
using the results of the ADD Health study from being able to gain a complete perspective on the
feelings that the participants had experienced.
For the first hypothesis, the expectation is that the more intelligent an adolescent sees
themselves as; the more likely they are to believe in what they can accomplish with hard work.
The logic behind this hypothesis is that people who view themselves as intelligent will believe
that they already have an advantage in completing a given task (Steinmayr & Spinath, 2009; Cho
& Lin, 2011). The hard work that they do will add to that advantage and, in turn, the likelihood
that they can complete the task will increase in a continuum type manner. In other words, the
more hard work an intelligent adolescent engages in, the more likely they are to believe in their
expectancy of a given task and the inverse is hypothesized to also hold true.
H1: Adolescents who perceive themselves to be more intelligent will more strongly agree with
the notion that they can accomplish tasks through hard work than their peers who perceive
themselves as less intelligent.
For the second hypothesis, there is the belief that the feelings scale will consist of four
factors that should be revealed in the analyses. Most of the literature has found a four factor
model in the CES–DS, but this particular study looks to examine the factor structure because the
four factor structure did not hold for all groups (Kim, DeCoster, Huang, & Chiriboga, 2011). The
four sub-factors hypothesized are: sadness, worthlessness, irritability, and fatigue. These factors
were determined by reading and analyzing the content of each of the 19 items in the scale and

then grouping them together on similar themes. After analyzing the questions, the four themes
mentioned earlier were found throughout the questions. More specifically, the expectation is that
Items 2, 3, 6, 8, 10, 11, 13, 15, and 16 will load on the Sadness factor. Items 4, 9, 17, and 19 will
load on Worthlessness. Items 1 and 14 will load on Irritability, and items 5, 7, 12, 18 will load on
the Fatigue factor (for item listing, see Appendix A).
H2a: Four factors will be extracted from the feelings scale and the four factors will be Sadness,
Worthlessness, Irritability, and Fatigue.
It is also hypothesized that the four factors extracted from the model will have a moderate
positive correlation with each of the other factors. The idea is that each of the four factors may
be a sub-factor to a main hierarchical factor, but this possibility can only be determined by
examining the results a Confirmatory Factor Analysis (CFA). As a result, there should be some
similarities in the factors that would allow them to correlate in the way they are hypothesized to
correlate. The major implication is that if this does not hold true, then the logic associated with
the four factor hypothesis would be negatively affected.
H2b: The four factors will all have a moderate and positive correlation with each other.
The two hypotheses presented should serve the purpose of not only guiding future
research, but also help to interpret what has actually been measured through the Add Health
survey.
Method
Participants
The Add Health study participants were a nationally representative group of adolescents
who were between grades 7-12 in the United States during the 1994-1995 academic school year.
There were 6,504 participants included in the study. Overall, 48.4 percent of the respondents
were male, while 51.6 percent were female. Sixty-six percent of the respondents were White,

24.9 percent African-American, 3.6 percent American Indian, 4.2 percent Asian, and 6.5 percent
identified themselves as other. Ninety-three percent of respondents came from homes where
English was the primary language spoken. Additionally, 30.4 percent of the respondents were
middle school students while the other 69.6 percent were high school students at the time Wave
I.
Measures
The primary measure used for the present study was the Add Health study questionnaire,
which was used for the interview. The questionnaire consisted of multiple sub-scales that
examined the following topics: respondents' social, economic, psychological and physical well-
being with contextual data on the family, neighborhood, community, school, friendships, peer
groups, and romantic relationships. For the analyses used in this particular study, only the
feelings scale, one that asked participants about their perceived level of intelligence in relation to
other individuals (measuring Intelligence), while the other asked about how strongly they felt
that they could accomplish a task if they worked hard (measuring Hard Work Expectancy). No
reliability statistics were reported.
Procedures
For Wave I, a stratified, random sample of all high schools in the United States was
taken. To be eligible for the study, a school had to have 11th grade students and have a minimum
student enrollment of 30. Feeder schools, which are schools that send graduates to high school
and include a 7th grade, were also recruited for inclusion in the study. Once the schools were
selected, students from each school were selected to participate in the study. Each student was
interviewed individually by an interviewer, who asked the participant to answer the questions in

the questionnaire. The student would then respond to the questions and was allowed to leave the
interview upon completion.
Results
A between-subjects one-way analysis of variance (ANOVA) was conducted with
Intelligence (moderately below average, slightly below average, about average, slightly above
average, moderately above average, and extremely above average) serving as the independent
variable and Hard Work Expectancy as the dependent variable. There was a significant effect of
intelligence on task expectancy, F (5, 6460) = 15.11, p 2 = .01. For the descriptive
statistics of the ANOVA, see Appendix B. Tukey’s HSD post-hoc analyses revealed that there
were statistical differences between certain groups. The group differences found is as follows:
“Extremely above average” (M = 1.90) was significantly different from every other group except
for “moderately above average” (M = 2.02). “Moderately above average” was statistically
different from “about average” (M = 2.16) and “slightly below average” (M = 2.32). “Slightly
above average” was different from “about average” and “slightly below average,” as well as
“extremely above average.”
An Exploratory Factor Analysis (EFA) was also conducted for the present study.
Maximum Likelihood was the extraction method that was chosen for use and Promax was the
factor rotation method selected. The EFA extracted four factors initially based on the Kaiser
criterion. The total amount of variance explained by those four factors was 51.32 percent. Factor
1 had an eigenvalue of 5.98 and accounted for 31.36 percent of the variance. Factor 2 had an
eigenvalue of 1.58 and accounted for 8.32 percent of the variance. Factor 3 had an eigenvalue of
1.17 and accounted for 6.13 percent of the variance. Finally, Factor 4 had an eigenvalue of 1.05
and accounted for 5.51 percent of the variance. To see the factor loadings for each factor, see

Appendix C. All of the items loaded strongly on only one factor except for item 12, which did
not strongly load on any of the four factors. The goodness of fit test results suggest that there is a
significant statistical difference between the theorized model and the actual model of the data, χ2
(101) = 1206.96, p < .01. The interpretation of this result is that the analyzed data may not match
the proposed model well.
EFA factor extractions of three, five, and six factors were also examined. The three factor
model, χ2 (117) = 1907.91, p < .01, had a worse model fit value than the four factor model. The
Chi-Square values were smaller for the five factored model, χ2 (86) = 660.05, p < .01, and six
factored model, χ2 (72) = 486.61, p < .01, than for the commonly theorized four factored model,
which signifies a better model fit for the five and six factor models.
Only including the 18 items that are recommended to be kept through the EFA (excluding
Item 12), the reliability estimates were taken for each of the factors. Factor 1 had good internal
consistency, α = .84 and Factor 2 had an acceptable reliability estimate, α = .72. Factor 3 has
near acceptable internal reliability, α = .69. Factor 4 has an internal consistency estimate, α = .60.
As shown in Appendix D, all four of the factors were at least moderately and positively
correlated with one another. This provides support for Hypothesis 2b and also indirectly suggests
that the factors may possibly be measuring some overall construct that uses the similarities in the
four factors.
Discussion
The results of the ANOVA provided statistical evidence for hypothesis 1. However, the
results are not practically significant despite the fact that they are statistically significance. The
statistical significance is the result of the high level of power that comes from having 6466
participants to compare for the analysis. With a sample of that size, virtually any differences in

the data set likely would have been found to be statistically significant. For the ANOVA, η2 =
.01, which is an effect size so small that it is virtually non-existent. There may truly be an effect
on perceived ability to accomplish a task through hard work by perceived intelligence, but a
much smaller sample should be used to re-examine this research topic. However, this study may
provide evidence to the notion that adolescents tend to believe that they can accomplish nearly
anything with hard work, regardless of their perceived intelligence level.
The results of the EFA support the idea of the four factors that the feelings scale was
hypothesized to contain. However, the four factors did not fit the hypothesized factors. In the
hypothesis, it was predicted that the four factors extracted from the model would be Sadness,
Worthlessness, Irritability, and Fatigue. In the actual analyses, the four factors were different,
with one of the factors having no relevant theoretical commonality in the item content.
Factor 1 appears to be measuring Sadness as predicted, while Factor 2 appears to be
measuring nothing more than the items being reverse coded. Other than the reverse coding, the
items do not share any content commonality. Factor 3 appears to be measuring “social factors”
(that could impact depression) and Factor 4 does appear to measure Fatigue as hypothesized.
Factor 1 items seemed to touch on the key aspects of hopelessness and sorrow. These aspects
were in line with the idea of sadness, which is why the factor was named accordingly.
Psychometrically, reverse coded items have a tendency to form their own factor regardless of the
specific content in each item, so this development in Factor 2 is not surprising. These four items
may need to be examined and revised at a later time. Factor 3 items appeared to key in on the
unfriendly actions of other people. That is why it was labeled Unfriendability. All of the items on
Factor 4 deal with fatigue-like issues, which is why it was labeled Fatigue.

The acceptability of the reliability estimates varies for each factor. Factor 1 appeared to
have good internal consistency and Factor 2 appeared to have an acceptable level of internal
consistency. However, the reliability estimate for Factor 2 may not say much since this factor
only consists of the reverse coded items. Factor 3 has an internal reliability that is near
acceptability, which can be attributed to the small number of items grouped to the factor (n = 2).
Factor 4 has an internal consistency estimate that is lower than what would be considered
acceptable. However, this likely can be attributed to only having three items for the factor.
EFA factor extractions of three, four, five, and six factors were examined. The three
factor model still grouped all of the reverse coded items into their own group just like the four
factor model did, so it was decided that the three factor model was not appropriate. Though the
Chi-Square values were smaller for the five factored and six factored models than for the four
factored model, which signifies a better model fit, the decision was made to only extract four
factors because that is the most common number of factors extracted from the CES-DS version
of the scale (Radloff, 1977). Additionally, there was not any theoretical evidence to suggest
using a five or six factor model would be appropriate.
Looking at Appendix D, the results are in line with the expected relationships. All of the
factors on the feelings scale were hypothesized to at least have a moderate positive correlation
with one another and they appear to have those relationships. One major limitation to this study
is that an EFA cannot directly test hierarchical factor structures. The high inter-factor
correlations between the factors provide evidence of the possibility of a main hierarchical factor,
but that can only be determined through a CFA.
Overall, the results of the ANOVA do not support hypothesis 1, despite being statistically
significant because of the marginal effect size. The results of the EFA suggest that the scale may

be touching on many different aspects of depression despite only containing 19 questions.
However, it is suggested that more items be added to the Unfriendability and Fatigue factors that
were extracted from the EFA. The reliability estimates suggest that the scale is a fairly reliable
estimate of the amount of depressive feelings that adolescents have experienced, but that
improvements could be made to make the scale even more informative and, in turn, valid.

References
Beitchman, J. (n.d.). Feelings, Attitudes, and Behaviors Scale for Children.
Cho, S. & Lin, C. Y. (2011). Influence of family processes, motivation, and beliefs about
intelligence on creative problem solving of scientifically talented individuals. Roeper
Review, 33, 46-58. doi: 10.1080/02783193.2011.530206
Diener, E., Wirtz, D., Tov, W., Kim-Prieto, C., Choi, D., Oishi, S., & Biswas-Diener, R. (2010).
New well-being measures: Short scales to assess flourishing and positive and negative
feelings. Social Indicators Research, 97, 143-156. doi:10.1007/s11205-009-9493-y
Harris, K.M., & Udry, J.R. (2013). National longitudinal study of adolescent health (Add
Health), 1994-2008 (ICSPR Study No. 21600). Retrieved from Inter-University
Consortium for Political and Social Research.
website: https://www.icpsr.umich.edu/icpsrweb/DSDR/studies/21600.
Kim, G., DeCoster, J., Huang, C., & Chiriboga, D. A. (2011). Race/ Etnicity and the factor
structure of the Center for Epidemiologic Studies Depression Scale: A meta-analysis.
Cultural Diversity and Ethnic Minority Psychology, 17, 381-396. doi: 10.1037/a0025434
Radloff, L.S. (1977). The CES-D Scale: A self-report depression scale for research in the general
population. Applied Psychological Assessment. 1, 385-401.
doi:10.1177/014662167700100306
Redick, T. S., Shipstead, Z., Harrison, T. L., Hicks, K. L., Fried, D. E., Hambrick, D. Z., & ...
Engle, R. W. (2013). No evidence of intelligence improvement after working memory
training: A randomized, placebo-controlled study. Journal Of Experimental Psychology:
General, 142, 359-379. doi:10.1037/a0029082

Steinmayr, R., & Spinath, B. (2009). What explains boys’ stronger confidence in their
intelligence?. Sex Roles, 61, 736-749. doi:10.1007/s11199-009-9675-8
Williams, E. F., Dunning, D., & Kruger, J. (2013). The hobgoblin of consistency: Algorithmic
judgment strategies underlie inflated self-assessments of performance. Journal Of
Personality And Social Psychology, 104, 976-994. doi:10.1037/a0032416
Zell, E., & Krizan, Z. (2014). Do people have insight into their abilities? A metasynthesis.
Perspectives On Psychological Science (Sage Publications Inc.), 9, 111-125.
doi:10.1177/1745691613518075

Appendix A
National Longitudinal Study for Adolescent Health (Add Health) Section 10: Feelings Scale
How often was each of the following true during the last week?
1. You were bothered by things that usually don’t bother you.
2. You didn’t feel like eating, your appetite was poor.
3. You felt that you could not shake off the blues, even with help from your family and your
friends.
4. You felt that you were just as good as other people. (Reverse Coded)
5. You had trouble keeping your mind on what you were doing.
6. You felt depressed.
7. You felt that you were too tired to do things.
8. You felt hopeful about the future. (Reverse Coded)
9. You thought your life had been a failure
10. You felt fearful
11. You were happy (Reverse Coded)
12. You talked less than usual.
13. You felt lonely.
14. People were unfriendly to you.
15. You enjoyed life. (Reverse Coded)
16. You felt sad.
17. You felt that people disliked you.
18. It was hard to get started doing things.
19. You felt life was not worth living.
*A 4-point Likert Scale was used for responses with the options to select “refuse to answer” or
“I don’t know” to answer the question

Appendix B
Table 1
ANOVA Descriptive Statistics
N Mean Std. Deviation
Moderately below average 76 2.29 1.043
Slightly below average 321 2.32 .985
About average 2500 2.16 .868
Slightly above average 1419 2.08 .842
Moderately above average 1735 2.02 .857
Extremely above average 415 1.90 .960
Total 6466 2.10 .879

Appendix C
Table 2
EFA Factor Loadings (Four Factor Model)
Item Factor
1 2 3 4
1 .483** -.031 -.021 .168
2 .352* .043 -.087 .189
3 .827** -.049 -.105 -.018
4 -.039 .535** .066 -.014
5 .224 .023 -.010 .398*
6 .897** -.028 -.078 -.039
7 .101 .009 -.012 .504**
8 -.154 .651** -.036 .055
9 .404** .138 .124 .009
10 .341* -.048 .165 .090
11 .129 .658** -.059 -.018
12 .255 .031 .029 .101
13 .626** -.015 .074 -.002
14 -.042 -.037 .683** .055
15 .085 .677** .001 -.018
16 .725** -.042 .062 -.037
17 .064 .021 .751** -.030
18 -.031 -.001 .060 .596**
19 .405** .113 .145 -.049
Note: *Acceptable Factor Loading ( Loadings <.3), **Strong Factor Loading (Loadings > .4)

Appendix D
Table 3
EFA Factor Correlation Matrix
Factor 1 2 3 4
1 1 - - -
2 .572 1 - -
3 .561 .369 1 -
4 .626 .362 .457 1

Data analysis proj. paper

Recommended

Recommended

More Related Content

What's hot

What's hot (17)

Similar to Data analysis proj. paper

Similar to Data analysis proj. paper (20)

More from sedunham

More from sedunham (18)

Recently uploaded

Recently uploaded (20)

Data analysis proj. paper