Anda di halaman 1dari 6

M.

Arias MR Assignment 1

Research Questions

There were two research questions in this study, developed from the literature review:

RQ1: What impact do gender, self-concept, and value of education (all measured

at time1) have on student self-reported effort (at time 2) when controlling for self-

reported effort at time 1?

H0= Gender, self-concept, and value of education have no impact on self-reported

effort (time 2)

RQ2: What impact do students perceptions of teacher interest, teacher caring,

and teacher competency, beyond the students personal characteristics, have on

self-reported effort?

H0= Students perceptions of teacher interest, teacher caring, and teacher

competency have no impact on have on self-reported effort beyond the students

personal characteristics

Methods

Participants

Participants in this study ranged in age from 13-18, with 15-year-olds

representing the largest grouping at 38% of participants. Students in their first year of Commented [MGP1]: Once it is a continuous variable
(and within realms of normality) report means and SD
high school represented 36% of the sample, with 33.6% sophomores, 18.7% juniors and

8.7% seniors. There was a fairly even split between males (43.3%) and females (54.4%).

Caucasians represented 73.1% of the sample with 7.1% reporting ethnicity as other,

6.4% Hispanic, 4.7% African American, 4.2% Asian American, and 1.6% Native

American.
M. Arias MR Assignment 2

Procedure

The independent variables examined were self-concept, gender, effort, value of

education and student perception of teacher interest, caring and competency. Self-concept

was scored on a scale from 1-4; effort, value of education, teacher interest, caring and

competency were scored from 0-4. Gender was dummy coded with males=1 and

female=0.

Before proceeding with the data analysis, all variables were screened for possible

code and statistical assumption violations and outliers, with IBM SPSS 24. Frequencies

and box-plots were used to check for outliers on all applicable variables. Three outliers

identified for Self-Concept were deleted as invalid values. Outliers for teacher interest

and teacher competency were valid and so were retained for analysis (See Figure 1 for

example). Commented [MGP2]: I admire that you did not want to


remove cases. However, MR requires normal distribution.
Figure 1. Teacher Interest Box Plot You have to make some justification for keeping variables
that are suggesting non-normal. Maybe use 3 SD z-score as a
method for keeping values?

Computing the Mahalanobis distance for each case on the continuous variables Commented [MGP3]: Write the statistics generated for
this test
screened for multivariate outliers, six cases were identified as potential outliers but all

were retained in the analysis. A scatter plot was created from the standardized residuals Commented [MGP4]: Assumptions need to be met. Do
you believe that these cases will jeopardize your conclusion
if removed?
M. Arias MR Assignment 3

and the standardized predicted variables. The Loess line follows the general flow of the

zero line, confirming linearity and the scatter plot confirms homoscedasticity (Figure 2). Commented [MGP5]: There are three points I can easily
identify as possible outliers in the plot youve presented. I
Figure 2. Scatter Plot know you see them as well

Histograms were used to confirm normality for all variables (see Figure 3 for example

histogram). Self-concept was normally distributed with skewness of -.566 (SE=.117) and

kurtosis of .413 (SE=.234). Value of education was normally distributed with skewness

of -.282 (SE=.117) and kurtosis of -.564 (SE=.234). Effort was normally distributed with

skewness of -.831 (SE=.117) and kurtosis of .179 (SE=.234). Teacher interest, caring and

competency were all normally distributed with skewness of -.910, -.467, -.772 (SE=.117)

and kurtosis of .799, -.473, .269 (SE=.234) respectively. Commented [MGP6]: Thank you for not including gender
here
M. Arias MR Assignment 4

Figure 3. Histogram for Self-Concept

While multivariate normality is not directly tested, univariate normality, linearity and Commented [MGP7]: Mahalanobis distance can be used

homoscedasticity were confirmed, so multivariate normality can be assumed.

Two variables, teacher interest and teacher competency, are highly correlated

(r=.812) and could indicate a problem with multicollinearity. The correlations of the Commented [MGP8]: I understand why you would say
this. However, in my experience .85-.90 are more likely to
be problematic
variables are shown in Table 1.

Table 1. Correlations of the Variables in the Analysis (N=431)


Variable 2 3 4 5 6 7 8
1. Effort (2s) .530 -.206 .194 .321 .237 .184 .214
2. Effort (1s) -- -.254 .230 .521 .362 .353 .368
3. Gender -- .117 -.036 -.059 -.047 .002
4. Self-Concept -- .346 .114 .181 .185
5. Value of Education -- .485 .526 .536
6. Teacher Interest -- .771 .818
7. Teacher Caring -- .754
8. Teacher --
Competency
Note: All correlations were statistically significant (p<.05)
M. Arias MR Assignment 5

Results

A three stage hierarchical linear regression analysis was used to predict the level

of the level of student self-reported effort (time 2). In the first block, effort (time 1) was

entered as a co-variate; in the second block, student characteristic (gender, self-concept,

and value of education) were simultaneously entered; in the third block teacher interest,

caring and competency were simultaneously entered. The correlations of the variables are

shown in Table 1. As can be seen, student self-report effort (2s) correlated most strongly

with self-reported effort (time 1) and value of education.

Table 2. Hierarchical Regression Results (N=426)


R2 Structure
Block R2 Change Model b SE-b Beta Pearson r sr2 coefficient
1 .281 .281 Constant .464 .165
Effort 1s .666 .052 .530 .530 .280

2 .296 .015* Constant .440 .200


Effort .580 .063 .462 .530 .168 .974
Gender -.163 .073 -.097 -.206 .012 -.379
Self-Concept .098 .053 .083 .194 .008 .357
Value of Education .044 .045 .048 .321 .002 .590

3 .303 .007 Constant .291 .228


Effort .573 .064 .457 .530 .162 .964
Gender -.161 .073 -.096 -.206 .011 -.375
Self-Concept .108 .053 .091 .194 .01 .353
Value of Education .050 .051 .055 .321 .230 .584
Teacher Interest .167 .093 .140 .237 .008 .431
Teacher Caring -.105 .064 -.113 -.079 .006 .014
Teacher
Competency -.033 .088 -.029 .214 .001 .389
sr2 is the squared semi-partial correlation.
*p<.05.

The results of the hierarchical regression are shown in Table 2. Effort (time 1)

was a significant covariate, F(1, 424) = 165.973, p<.001. When three student
M. Arias MR Assignment 6

characteristic variables were added on the second block, the prediction model was

statistically significant, F(4, 421) = 44.307, p=.242, R2 = .296, adjusted R2 = .290. The Commented [MGP9]: Significant?

final model, which included teacher interest, caring and competency, was statistically

significant F(7, 418) = 25.992, p<.001, R2 = .303, adjusted R2 = .292, however, the

change contributed by this model was not significant, p=.312, determining that these

variables do not predict a significant amount of variance beyond personal characteristics.

Prior effort is the most effective predictor of later effort. All other variables held constant,

mens effort will be less than womens (b= -0.163). Commented [MGP10]: The R2 Change statistics are more
helpful in answering your research questions. I would focus
more on those. And always report a few coefficients so the
reader will be able to understand the others

Commented [MGP11]: Overall, very good job!!