Sie sind auf Seite 1von 21

Final Report

KEDGE Business School - From Data to Decision Making


Group 5: Brenda Silva, Dorothe Scheich, Alara Ozhan, Samet Turkkan, Yifan Li

1. Research Objectives

Understand what factors determine pay levels


Influence of employee characteristics on salary
Find discrepancies in the pay policy
Is the pay policy consistent?

2. Measurement and Scaling


Dependent variables

current salary (scale, ratio)


beginning salary (scale, ratio)

Independent variables (employee characteristics)

gender (nominal)
m - male
f female

employment category (ordinal)


1 - Clerical
2 - Custodial
3 Manager

minority status (nominal)


0 - No
1 Yes

educational level in years (scale, ratio)


Time in the firm in months (scale, ratio)
Previous experience in months (scale, ratio)

3. Sample and Statistical Methods used


Sample

currently employed people (but within this group we have random sampling)
we do not know from which industries or companies
Size: n = 474 employees
Male = 258 (54%)
Female = 216 (46%)
No Minority = 370 (78%)
Minority = 104 (22%)
Clerical = 363 (77%)
Custodial = 27 (5%)
Managers = 84 (18%)
Educational level = from 8 to 21 years
Previous experience = from 0 to 476 months
Time hired = from 63 to 98 months
Conclusion:
Gender representation is good; wide and good range of educational level, previous
experience and time hired
Minority representation is insufficient; not sufficient number of higher positions
(custodial, manager).

Statistical methods

Univariate analysis
Central tendencies
Dispersion
Bivariate Analysis
- Independent T-test
- One way ANOVA
- Pearson Correlation
- Simple regression
- Chi-square test
Multivariate analysis
Multiple regression
Two way ANOVA

4. Data Analysis and Interpretation/Results


for all analyses, we used = 0,05

Univariate analysis

Variables

Type of
Variable

Central Tendency

Standard
Deviation

Variance

Range

Current salary

Scale

Mean: 34419,57

17075,661

291578214,500

119250

Beginning salary

Scale

Mean: 17016,09

7870,638

61946944,960

70980

Gender

Nominal

Mode: m = Male

Minority status

Nominal

Mode: 0 = No

Employment category

Ordinal

Median: 1 = Clerical

Time hired

Scale

Mean: 81,11 months

10,061

101,223

35

Previous experience

Scale

Mean: 95,86 months

104,586

10938,281

476

Educational level

Scale

Mean: 13,49 years

2,885

8,322

13

Bivariate analysis
Independent T-Test
H0: 1 = 2
H1: 1 2
a) Gender and current salary
Group Statistics
Gender
Current Salary

Mean

Std. Deviation

Std. Error Mean

Male

258

41441,78

19499,214

1213,968

Female

216

26031,92

7558,021

514,258

Independent Samples Test


Levene's Test for
Equality of

t-test for Equality of Means

Variances
Sig.

95% Confidence Interval of

(2-

Current Salary

Equal

df

the Difference

taile

Mean

Std. Error

d)

Difference

Difference

Sig.

Lower

Upper

119,669

,000

10,945

472

,000

15409,862

1407,906

12643,322

18176,401

11,688

344,

,000

15409,862

1318,401

12816,728

18002,996

variances
assumed
Equal
variances not

262

assumed

Result: t(344,26) = 11,688; p = 0,00


The means of salary of male (M = 41441,78) and female (M = 36031,92) are
significant (p = 0,00 < 0,05 H0 rejected), so they are different, given that equal
variances assumption is not validated.

b) Minority status and current salary


Group Statistics
Std. Error
Minority Classification
Current Salary

Mean

Std. Deviation

Mean

No

370

36023,31

18044,096

938,068

Yes

104

28713,94

11421,638

1119,984

Independent Samples Test


Levene's Test for
Equality of
Variances

t-test for Equality of Means


95% Confidence Interval of

Sig.
F
Current Salary

Equal

28,487

df

(2-

Mean

Std. Error

tailed)

Difference

Difference

the Difference

Sig.

Lower

Upper

,000

3,915

472

,000

7309,369

1867,111

3640,491

10978,246

5,003

262,

,000

7309,369

1460,936

4432,707

10186,030

variances
assumed
Equal
variances not
assumed

188

Result: t(262,188) = 5,003; p = 0,00


The means of salary of employees with (M = 28731,94) and without (M = 36023,31)
minority status are significant (p = 0,00 < 0,05 H0 rejected), so they are different.
The result is slightly biased, because the sizes of the subgroups are not the same
and variances (standard deviations) differ.

One way ANOVA


H0: 1 = 2 = 3
H1: 1 2 3
a) Employment category and current salary

Descriptives
Current Salary
95% Confidence Interval for
Mean
Std.
N
Clerical

Mean

Deviation

Std. Error

Lower

Upper

Bound

Bound

Minimum

Maximum

363

27838,54

7567,995

397,217

27057,40

28619,68

15750

80000

Custodial

27

30938,89

2114,616

406,958

30102,37

31775,40

24300

35250

Manager

84

63977,80

18244,776

1990,668

60018,44

67937,16

34410

135000

474

34419,57

17075,661

784,311

32878,40

35960,73

15750

135000

Total

Test of Homogeneity of Variances


Current Salary
Levene Statistic
59,733

df1

df2
2

Sig.
471

,000

The variances are not the same, so we have a bias.


ANOVA
Current Salary
Sum of Squares
Between Groups

89438483930,000

df

Mean Square
2

44719241960,000

F
434,481

Sig.
,000

Within Groups
Total

48478011510,000

471

137916495400,000

473

102925714,500

Result: F(2,471) = 434,481; p = 0,00


There is a significant difference (p = 0,00 < 0,05 H0 rejected) between the means
of the subgroup so we need to run a post hoc test to see where the differences are.

Post Hoc: Multiple Comparisons


Dependent Variable: Current Salary
Tamhane
Mean
(J) Employment

95% Confidence Interval

Difference

(I) Employment Category

Category

Clerical

Custodial

-3100,349*

568,679

Manager

-36139,258*

Custodial

(I-J)

Lower Bound

Upper Bound

,000

-4483,07

-1717,63

2029,912

,000

-41078,30

-31200,21

3100,349*

568,679

,000

1717,63

4483,07

-33038,909*

2031,840

,000

-37982,78

-28095,04

Clerical

36139,258*

2029,912

,000

31200,21

41078,30

Custodial

33038,909*

2031,840

,000

28095,04

37982,78

Clerical
Manager

Manager

Std. Error

Sig.

*. The mean difference is significant at the 0.05 level.

The means of salary of all subgroups/employment categories are significantly different (p


= 0,00 < 0,05 H0 rejected).
Mean salary Clerical: 27828,54
Mean salary Custodial: 30938,89
Mean salary Manager: 63977,80

Bivariate Correlation (Pearson)


H0: variables are not linearly correlated
H1: variables are linearly correlated

a) Beginning salary and previous experience


Descriptive Statistics

Mean
Beginning Salary

Std. Deviation

17016,09

7870,638

474

95,86

104,586

474

Previous Experience (months)

Correlations
Previous
Experience
Beginning Salary
Beginning Salary

Pearson Correlation

(months)

,045

Sig. (2-tailed)

Previous Experience (months)

,327

474

474

Pearson Correlation

,045

Sig. (2-tailed)

,327

474

474

They seem to be slightly positively correlated (r = 0,045), but the significance is not
strong enough (p = 0,327 > 0,05 H0 accepted), so they are not linearly correlated.

b) Current salary and time hired in the company


Descriptive Statistics
Mean
Current Salary
Months since Hire

Std. Deviation

34419,57

17075,661

474

81,11

10,061

474

Correlations
Current Salary
Current Salary

Pearson Correlation

Months since Hire


1

Sig. (2-tailed)

Months since Hire

,084
,067

474

474

Pearson Correlation

,084

Sig. (2-tailed)

,067

474

474

They seem to be slightly positively correlated (r = 0,084), but the significance is not
strong enough (p = 0,067 > 0,05 H0 accepted), so they are not linearly correlated.

Simple regression
H0: no prediction possible of dependent variable by independent variable
H1: prediction possible of dependent variable by independent variable

a) Previous experience on beginning salary


Coefficientsa
Standardized
Unstandardized Coefficients
Model
1

B
(Constant)
Previous Experience

Coefficients

Std. Error

16690,478

490,646

3,397

3,460

Collinearity Statistics

Beta

Sig.

34,017

,000

,982

,327

,045

Tolerance

1,000

VIF

1,000

(months)

Model Summary
Std. Error of the
Model

R Square
,045a

Adjusted R Square

,002

Estimate

,000

7870,942

ANOVAa
Model
1

Sum of Squares
Regression

df

Mean Square

59692531,850

59692531,850

Residual

29241212430,000

472

61951721,260

Total

29300904970,000

473

Result: b = 3,397; t(472) = 0,982; r = 0,327; R = 0,002

Sig.
,964

,327b

The previous experience is not significant for predicting the beginning salary (r =
0,327 > 0,05 H0 accepted).
The linear regression model predicts the outcomes very insufficiently (R = 0,002).

b) Time hired in the company on current salary


Coefficientsa
Standardized
Unstandardized Coefficients
Model
1

B
(Constant)
Months since
Hire

Std. Error

22843,324

6362,214

142,723

77,844

Coefficients
Beta

Collinearity Statistics
t

,084

Sig.

3,590

,000

1,833

,067

Tolerance

1,000

VIF

1,000

Model Summary
Std. Error of the
Model

R Square
,084a

Adjusted R Square

,007

Estimate

,005

17033,194

ANOVAa
Model
1

Sum of Squares
Regression
Residual

df

Mean Square

F
3,362

975277786,300

975277786,300

136941217600,00

472

290129698,400

Sig.

0
Total

137916495400,00

473

Result: b = 142,723 ; t(472) = 1,833; r = 0,067; R = 0,007


The time hired in the firm is not significant for predicting the current salary (r = 0,067
> 0,05 H0 accepted).
The linear regression model predicts the outcomes very insufficiently (R = 0,007).

,067b

Chi-square test
H0: variables are statistically not associated (independent) with each other
frequency observed = frequency expected
H1: variables are statistically associated with each other
frequency observed frequency expected

a) Gender and employment category


Employment Category * Gender Crosstabulation
Count
Gender
Female
Employment Category

Clerical

Male

Total

206

157

363

Custodial

27

27

Manager

10

74

84

Total

216

258

474

Chi-Square Tests
Asymptotic Significance (2Value

df

sided)

79,277a

,000

Likelihood Ratio

95,463

,000

N of Valid Cases

474

Pearson Chi-Square

Result: Chi(2, N = 474) = 79,277; p = 0,00


The observed and expected frequencies are significantly different (p = 0,00 < 0,05
H0 rejected), so gender and employment category are statistically associated with
each other.
Women hold less higher positions than men.

b) Minority status and employment category


Employment Category * Minority Classification Crosstabulation
Count
Minority Classification
No
Employment Category

Clerical

Yes

Total

276

87

363

Custodial

14

13

27

Manager

80

84

370

104

474

Total

Chi-Square Tests
Asymptotic
Significance (2Value
Pearson Chi-Square
Likelihood Ratio
Linear-by-Linear Association

df

sided)

26,172a

,000

29,436

,000

9,778

,002

N of Valid Cases

474

Result: Chi(2, N = 474) = 26,172; p = 0,00


The observed and expected frequencies are significantly different (p = 0,00 < 0,05
H0 rejected), so minority status and employment category are statistically associated
with each other.
Minorities hold less higher positions than employees with no minority status.

Multivariate analysis
Multilinear regression
H0: no prediction possible of dependent variable by independent variable(s)
H1: prediction possible of dependent variable by independent variable(s)
a) Employee characteristics on current salary
Coefficients

Standardiz
ed
Coefficients

Unstandardized
Coefficients
Model
1

Std. Error

-27034,413

5490,338

3939,449

209,982

Months since Hire

97,519

Previous Experience
(months)

(Constant)
Educational Level (years)

Minority Classification

Beta

Collinearity Statistics
t

Sig.

Tolerance

-4,924

,000

,666

18,761

,000

,924

1,082

58,037

,057

1,680

,094

,995

1,006

13,940

5,793

,085

2,406

,016

,924

1,083

-4290,574

1429,928

-,104

-3,001

,003

,966

1,035

Model Summary
Model

,674

R Square
a

VIF

Adjusted R Square

,455

,450

ANOVA

Std. Error of the Estimate


12664,187

Model
1

Sum of Squares

df

Mean Square

Regression

62697514160,000

15674378540,000

Residual

75218981280,000

469

160381623,200

137916495400,000

473

Total

Sig.

97,732

,000

Results:
Educational level: b = 3939,449; t(469) = 18,761; r = 0,00; R = 0,455
Time hired: b = 97,519; t(469) = 1,680; r = 0,094; R = 0,455
Previous experience: b = 13,940; t(469) = 2,406; r = 0,016; R = 0,455
Minority status: b = -4290,574; t(469) = -3,001; r = 0,003; R = 0,455
The time hired is not significant for predicting the current salary (r = 0,094 > 0,05 H0
accepted).
The educational level, the previous experience and the minority status are significant for
predicting the current salary (r = 0,00 and r = 0,016 and r = 0,003 < 0,05 H0 rejected),
with the educational level being the strongest predictor (highest t-value)
The linear regression model predicts the outcomes fairly well (R = 0,455).

b) Employee characteristics on beginning salary


Coefficients
Unstandardized
Coefficients
Model
1

Std.
Error

(Constant)

6266,18
6

2522,92
0

Educational Level (years)

1857,33
5

96,491

Months since Hire

-37,420

Previous Experience
(months)
Minority Classification

Standardize
d
Coefficients
Beta

Collinearity Statistics
t

Sig.

Tolerance

VIF

-2,484

,013

,681

19,249

,000

,924

1,082

26,669

-,048

-1,403

,161

,995

1,006

17,406

2,662

,231

6,538

,000

,924

1,083

1866,41
5

657,080

-,098

-2,840

,005

,966

1,035

Model Summary
Model

,677

R Square

Adjusted R Square

,458

Std. Error of the Estimate

,453

5819,446

ANOVA

Model
1

Sum of Squares

df

Mean Square

Regression

13417774350,000

3354443588,000

Residual

15883130610,000

469

33865950,140

Total

29300904970,000

473

Sig.

99,051

,000

Results:
Educational level: b = 1,857,335; t(469) = 19,249; r = 0,00; R = 0,458
Time hired: b = -37,420; t(469) = -1,403; r = 0,161; R = 0,458
Previous experience: b = 17,406; t(469) = 6,538; r = 0,00; R = 0,458
Minority status: b = -1866,415; t(469) = -2,840; r = 0,005; R = 0,458

The time hired is not significant for predicting the beginning salary (r = 0,161 > 0,05 H0
accepted).
The educational level, the previous experience and the minority status are significant for
predicting the beginning salary (r = 0,00 and r = 0,005 < 0,05 H0 rejected), with the
educational level being the strongest predictor (highest t-value)
The linear regression model predicts the outcomes fairly well (R = 0,458).

Two way ANOVA


H0: no interaction between the two variables
H1: interaction between the two variables
a) Gender, employment category and current salary
Descriptive Statistics
Dependent Variable: Current Salary
Employment Category

Gender

Mean

Std. Deviation

Clerical

Custodial

Manager

Total

Female

25003,69

5812,838

206

Male

31558,15

7997,978

157

Total

27838,54

7567,995

363

Male

30938,89

2114,616

27

Total

30938,89

2114,616

27

Female

47213,50

8501,253

10

Male

66243,24

18051,570

74

Total

63977,80

18244,776

84

Female

26031,92

7558,021

216

Male

41441,78

19499,214

258

Total

34419,57

17075,661

474

Levene's Test of Equality of Error Variancesa


Dependent Variable: Current Salary
F

df1

33,383

df2
4

Sig.
469

,000

The variances are not the same, so we have a bias.


Tests of Between-Subjects Effects
Dependent Variable: Current Salary
Type III Sum of
Source

Squares

df

Mean Square

Sig.

Partial Eta Squared

Corrected Model

96456357290,000a

24114089320,000

272,780

,000

,699

Intercept

177271943100,000

177271943100,00

2005,313

,000

,810

0
jobcat

32316332040,000

16158166020,000

182,782

,000

,438

gender

5247440732,000

5247440732,000

59,359

,000

,112

jobcat * gender

1247682867,000

1247682867,000

14,114

,000

,029

Error

41460138150,000

469

88401147,440

Total

699467436900,000

474

Corrected Total

137916495400,000

473

Result: F(1,469) = 14,144; p = 0,00

The interaction between gender and job category is significant (p = 0,00 < 0,05 H0
rejected).

5. Conclusions

Does the HR payment policy look fair, given the employees' characteristics?
Should it be fair and if yes, why?

Results
1) T-Test

Men earn more than women


Minorities earn less than non-minorities

2) One way ANOVA

Also, we recognize that employment category has an impact on current salary


policy. Its a proper result because every employee has different position, so all
employees should get paid in a positive function of workload and responsibility.
Mean salary Clerical: 27828,54
Mean salary Custodial: 30938,89
Mean salary Manager: 63977,80

3) Bivariate Correlation and simple regression

Beginning salary is not correlated with previous experience. The simple


regression also shows us that previous experiences is not a significant predictor.
Current salary and time since hire is not correlated. When we realized the simple
regression test, we saw that time hired in company is not a significant predictor.

4) Chi-square test

Gender and employment category are statistically associated with each other;
women are less likely to hold a manager position.

Minority status and employment category are associated with each other; people
with no minority are more likely to be in a higher position.

5) Multilinear regression

Educational level is the strongest predictor for both beginning salary and current
salary, but a lot stronger on current salary
The minority status is a significant predictor of the beginning and current salary
The time hired is not a significant predictor

Recommendations

When we consider todays conditions in business life, HR Department must be


fair in his salary policy regardless of job-unrelated characteristics such as gender
or minority. The perception of fairness is essential for employees engagement
and can therefore enhance productivity.

Source: Rasch, R. and Szypko, M. (2013). Perception is Reality: The Importance of Pay Fairness to Employees and
Organizations. WorldatWork Journal, pp. 1-74. Retrieved from: https://hr-vendordirectory.worldatwork.org/html/pub/PerceptionIsReality_Q4Journal.pdf

HR departments should hire males and females and minorities on all positions
(diversity!)
Rather evaluate qualifications and experiences of candidates instead of gender or
minority status (equal chances!)
Link beginning salary to previous experience (a little bit more strongly), because
employees with a lot of experience are more valuable than inexperienced
employees, so they should be rewarded for their experience.
Link current salary to time hired (a little bit more strongly), because employees are
gaining new visions and experiences within time and they should be rewarded for
loyalty.

Das könnte Ihnen auch gefallen