You are on page 1of 10

1

Multivariate Analysis of Covariance (MANOVA) in SPSS

In this week's assignment, we are asked to review and discuss MANOVA. A MANOVA

(an expansion of the ANOVA) is where the researcher examines the statistical difference of two

or more IV and its effects on a continuous dependent variable (Warner, 2013). The dataset that I

decided to use for this statistical analysis was the General Social Survey. In this study, the

participants' mean age was 18 years old and older who lived in the United States (Smith et al.,

2013).

Hypotheses

The Ho: There are differences in number of hours usually worked a week (continuous
IV)

and Respondent income (continuous variable) race (categorical DV).

The Ha: There are no differences in number of hours usually worked a week and

Respondent income across Race.


.
Research Design I Would Align with This Study

The research design I would align with this study is quasi-experimental. In this study,

there is no random assignment, there will be a comparison of more than one variable, and there

is no manipulation of any variables (Creswell & Creswell, 2018; Warner, 2013).

Results

The results of the MANOVA yielded there was no statistically significant. A one-way

MANOVA was performed. In viewing the descriptive results, it appears that Blacks (M is 36.67)
2

worked just a little more hours in a work week than Whites (M is 26.60). When examining

respondent income, it appears that Whites' (M is 8.80) income on average was higher than

Blacks' (M is 7.67). Box’s M = 8.128 F(3, 405.704 = 1.548, p < .001 (.202- we fail to reject the

null). We meet the assumption of equal covariances matrices. Pillai’s Trace = .617, F(2, 5.000)

= 4.023, p > .05 (.091); the effect size partial is η2 = .617. In examining Levene’s test, there is a

homogeneity of variance because it is p > .05.

Possible Social Implication

What is the possible social implication of this study? It is essential to put the results

from this study into context. There has been racial disparity called structural racism since the era

of Jim Crow (1877 to 1954) (Yearby, 2018). Structural racism is the unconscious interactions

among organizations, policies, and practices that put up a wall to block opportunities for Blacks

(Yearby, 2018). Even though the Voting Rights Act of 1956 and Civil Rights were put into

place, there still exists structural racism (Yearby, 2018).

Subsequently, as a result, structural racism causes inequalities among Blacks and White

individuals when it comes to wealth, jobs, income, etc. (Yearby, 2018). For example, the

unemployment rate for Blacks is much than it is for Whites during the past 50 years (Pew

Research Center, 2016:26; as cited in Yearby, 2018). One of the biggest reasons is an

individual’s Black name does not sound white enough (Bertrand and Mullainathan, 2004, as

cited in Yearby, 2018).

Hopefully, studies like these will motivate the government to enforce laws and keep the

promise of equality for all (Yearby, 2018).


3

References

Creswell, J. W., & Creswell, J. D. (2018). Research design: Qualitative, quantitative, and

mixed methods approaches (5th ed.). Thousand Oaks, CA: Sage.


Smith, T W., Hout, M, & Marsden, P. V. General Social Survey, 1972-2014 [Cumulative File].
Arbor, MI: Inter-university Consortium for Political and Social Research [distributor],
2016-03-14. https://doi.org/10.3886/ICPSR36319.v2

Yearby, R. (2018). Racial disparities in health status and access to healthcare: the
continuation of inequality in the United States due to structural racism. American

Journal of Economics and Sociology, 77(3-4), 1113-1152.


Warner, R. M. (2013). Applied statistics: From bivariate through multivariate techniques (2nd
ed.). Thousand Oaks, CA: SAGE Publications.
4

Appendix
SPSS OUTPUT
GET
FILE='C:\Users\gpven\Downloads\General Social Survey Dataset C.sav'.
DATASET NAME DataSet1 WINDOW=FRONT.
GLM hrs2 rincome BY race
/METHOD=SSTYPE(3)
/INTERCEPT=INCLUDE
/POSTHOC=race(BONFERRONI)
/PLOT=PROFILE(race) TYPE=LINE ERRORBAR=NO MEANREFERENCE=NO YAXIS=AUTO
/EMMEANS=TABLES(race)
/PRINT=DESCRIPTIVE ETASQ HOMOGENEITY
/CRITERIA=ALPHA(.05)
/DESIGN= race.

General Linear Model

[DataSet1] C:\Users\gpven\Downloads\General Social Survey Dataset C.sav

Warnings
Post hoc tests are not performed for RACE OF RESPONDENT
because there are fewer than three groups.

Between-Subjects Factors
Value Label N
RACE OF RESPONDENT 1 WHITE 5
2 BLACK 3

Descriptive Statistics
RACE OF RESPONDENT Mean Std. Deviation N
NUMBER OF HOURS WHITE 26.60 12.054 5
USUALLY WORK A WEEK BLACK 36.67 15.275 3
Total 30.38 13.298 8
RESPONDENTS INCOME WHITE 8.80 4.604 5
BLACK 7.67 5.132 3
5

Total 8.38 4.470 8

Box's Test of
Equality of
Covariance
Matrices a
Box's M 8.128
F 1.548
df1 3
df2 405.704
Sig. .202

Tests the null


hypothesis that the
observed covariance
matrices of the
dependent variables
are equal across
groups.a
a. Design: Intercept +
race

Multivariate Tests a
Effect Value F Hypothesis df Error df Sig.
Intercept Pillai's Trace .894 21.124b 2.000 5.000 .004
Wilks' Lambda .106 21.124 b
2.000 5.000 .004
Hotelling's Trace 8.449 21.124 b
2.000 5.000 .004
Roy's Largest Root 8.449 21.124b 2.000 5.000 .004
race Pillai's Trace .617 4.023b 2.000 5.000 .091
Wilks' Lambda .383 4.023 b
2.000 5.000 .091
Hotelling's Trace 1.609 4.023 b
2.000 5.000 .091
Roy's Largest Root 1.609 4.023b 2.000 5.000 .091

Multivariate Testsa
Effect Partial Eta Squared
Intercept Pillai's Trace .894
Wilks' Lambda .894
Hotelling's Trace .894
6

Roy's Largest Root .894


race Pillai's Trace .617
Wilks' Lambda .617
Hotelling's Trace .617
Roy's Largest Root .617

a. Design: Intercept + race


b. Exact statistic

Levene's Test of Equality of Error Variances a


Levene Statistic df1 df2 Sig.
NUMBER OF HOURS Based on Mean .086 1 6 .779
USUALLY WORK A WEEK Based on Median .005 1 6 .944
Based on Median and with .005 1 4.543 .945
adjusted df
Based on trimmed mean .078 1 6 .790
RESPONDENTS INCOME Based on Mean .002 1 6 .965
Based on Median .002 1 6 .967
Based on Median and with .002 1 5.701 .967
adjusted df
Based on trimmed mean .001 1 6 .975

Tests the null hypothesis that the error variance of the dependent variable is equal across groups. a
a. Design: Intercept + race

Tests of Between-Subjects Effects


Type III Sum of
Source Dependent Variable Squares df Mean Square F
Corrected Model NUMBER OF HOURS 190.008 a
1 190.008 1.088
USUALLY WORK A WEEK
RESPONDENTS INCOME 2.408b 1 2.408 .105
Intercept NUMBER OF HOURS 7505.008 1 7505.008 42.973
USUALLY WORK A WEEK
RESPONDENTS INCOME 508.408 1 508.408 22.190
race NUMBER OF HOURS 190.008 1 190.008 1.088
USUALLY WORK A WEEK
RESPONDENTS INCOME 2.408 1 2.408 .105
7

Error NUMBER OF HOURS 1047.867 6 174.644


USUALLY WORK A WEEK
RESPONDENTS INCOME 137.467 6 22.911
Total NUMBER OF HOURS 8619.000 8
USUALLY WORK A WEEK
RESPONDENTS INCOME 701.000 8
Corrected Total NUMBER OF HOURS 1237.875 7
USUALLY WORK A WEEK
RESPONDENTS INCOME 139.875 7

Tests of Between-Subjects Effects


Source Dependent Variable Sig. Partial Eta Squared
Corrected Model NUMBER OF HOURS USUALLY WORK .337 .153
A WEEK
RESPONDENTS INCOME .757 .017
Intercept NUMBER OF HOURS USUALLY WORK .001 .877
A WEEK
RESPONDENTS INCOME .003 .787
race NUMBER OF HOURS USUALLY WORK .337 .153
A WEEK
RESPONDENTS INCOME .757 .017
Error NUMBER OF HOURS USUALLY WORK
A WEEK
RESPONDENTS INCOME
Total NUMBER OF HOURS USUALLY WORK
A WEEK
RESPONDENTS INCOME
Corrected Total NUMBER OF HOURS USUALLY WORK
A WEEK
RESPONDENTS INCOME

a. R Squared = .153 (Adjusted R Squared = .012)


b. R Squared = .017 (Adjusted R Squared = -.147)
8

Estimated Marginal Means

RACE OF RESPONDENT
95%
Confidence
Interval
Dependent Variable RACE OF RESPONDENT Mean Std. Error Lower Bound
NUMBER OF HOURS WHITE 26.600 5.910 12.139
USUALLY WORK A WEEK BLACK 36.667 7.630 17.997
RESPONDENTS INCOME WHITE 8.800 2.141 3.562
BLACK 7.667 2.764 .905

RACE OF RESPONDENT
95% Confidence
Interval
Dependent Variable RACE OF RESPONDENT Upper Bound
NUMBER OF HOURS USUALLY WORK A WHITE 41.061
WEEK BLACK 55.336
RESPONDENTS INCOME WHITE 14.038
BLACK 14.429
9

Profile Plots

NUMBER OF HOURS USUALLY WORK A WEEK

RESPONDENTS INCOME
10

You might also like