You are on page 1of 12

HO: The actual productivity of subjects does not differ with respect to the size of group

they are working in, their gender, or a combination of these factors. H1: The actual productivity of subjects differ with respect to the size of group they are working in , their gender, or a combination of these factors
Here, in this situation, we have two Independent Variables and we need to analyze them in detail. One-way ANOVA is a technique which is to be used for the analysis of one independent variable when more than one population is under consideration. But it cannot manipulate two or more than two independent variables at the same time. For this purpose, Factorial Analysis of Variance (ANOVA) is used. This is an extension of one way ANOVA in that it involves the analysis of two or more independent variables. Factorial designs are labeled by the number of factors involved. Thus, a factorial design with two independent variables is called a Two-Way Factorial or Two-Way ANOVA. But, before directly applying Two-Way ANOVA, we first check data set for three assumptions. These three assumptions are Independence, Normality and

Homogeneity of variances. 1. First, we shall check data set for the assumption of Independence. Random sampling and independence are technical concerns related to the way the samples are selected. The samples must be random and independent if they are to be representative of the populations. The value of one observation should not be related to any other observation. It is typically assessed through the examination of the research design. In this situation, we do not find any clue for considering that the data collected from one sample has any impact on other samples. That is, one event does not depend on another. Thus, the assumption has been met, two way ANOVA seems an appropriate statistics to be used here.

Case Processing Summary subject gender Valid N male subject's total task1 female 60 100.0% 0 0.0% 60 100.0% 17 Percent 100.0% N 0 Cases Missing Percent 0.0% N 17 Total Percent 100.0%

Descriptives subject gender Mean 95% Confidence Interval for Mean 5% Trimmed Mean Median Variance male Std. Deviation Minimum Maximum Range Interquartile Range Skewness Kurtosis subject's total task1 Mean 95% Confidence Interval for Mean 5% Trimmed Mean Median Variance female Std. Deviation Minimum Maximum Range Interquartile Range Skewness Kurtosis Lower Bound Upper Bound 22.58 21.21 23.96 22.35 21.00 28.281 5.318 13 38 25 7 .695 .469 .309 .608 .687 Lower Bound Upper Bound Statistic 21.47 19.12 23.82 21.25 20.00 20.890 4.571 14 33 19 5 .945 1.425 .550 1.063 Std. Error 1.109

Tests of Normality subject gender Kolmogorov-Smirnov Statistic male subject's total task1 female *. This is a lower bound of the true significance. a. Lilliefors Significance Correction .134 60 .009 .960 60 .048 .160 df 17
a

Shapiro-Wilk Statistic
*

Sig. .200

df 17

Sig. .305

.939

Case Processing Summary group size Valid N size 2 subject's total task1 size 3 39 100.0% 0 0.0% 39 100.0% 38 Percent 100.0% N 0 Cases Missing Percent 0.0% N 38 Total Percent 100.0%

Descriptives group size Mean 95% Confidence Interval for Mean 5% Trimmed Mean Median Variance size 2 subject's total task1 Std. Deviation Minimum Maximum Range Interquartile Range Skewness Kurtosis Mean size 3 95% Confidence Interval for Lower Bound 21.60 Lower Bound Upper Bound Statistic 21.24 19.77 22.70 21.01 20.00 19.915 4.463 13 33 20 6 .686 .542 23.41 .383 .750 .897 Std. Error .724

Mean 5% Trimmed Mean Median Variance Std. Deviation Minimum Maximum Range Interquartile Range Skewness Kurtosis

Upper Bound

25.23 23.23 23.00 31.354 5.599 13 38 25 7 .641 .305 .378 .741

Tests of Normality group size Kolmogorov-Smirnov Statistic size 2 subject's total task1 size 3 a. Lilliefors Significance Correction .119 39 .177 .964 39 .243 .162 df 38
a

Shapiro-Wilk Statistic .949 df 38 Sig. .080

Sig. .013

In this situation we have two independent variables gender and group size. For the assumption of normality we have check only kolmogorov-Smirnov because our n is greater than 50 that is 71.The data is normally distributed in condition 1(male) i.e p>.05 but the data is not normally distributed in condition 2(female) i.e,p<.05 of the independent gender. Now we check second independent variable group size which also have 2 conditions ,condition 1 (size 2) and condition 2(size3) ,the data is not normally distributed in condition 1(size 2) i.e p <.05 [p-value of the test is .013]but the data is normally distributed in condition 2 (size 3) i.e,p>.05 [p-value of the test is .0177]. So we conclude that the assumption of normality is satisfied in both independents.

Levene's Test of Equality of Error Variances Dependent Variable: subject's total task1 F 1.619 df1 3 df2 73 Sig.

.192

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

Significant value is greater than .05 that is .192 suggesting that variances are equal.So the assumption of homogeneity of variances is satisfied.

Between-Subjects Factors Value Label 1 subject gender 2 1 group size 2 size 3 39 female size 2 60 38 male N 17

Descriptive Statistics Dependent Variable: subject's total task1 subject gender group size size 2 male size 3 Total size 2 female size 3 Total size 2 Total size 3 Total Mean 23.38 19.78 21.47 20.67 24.50 22.58 21.24 23.41 22.34 Std. Deviation 5.125 3.456 4.571 4.180 5.698 5.318 4.463 5.599 5.154 N 8 9 17 30 30 60 38 39 77

In the Between Subject Factor table we have two variables subject gender with two conditions male and female and second independent variable is group size with two conditions size 2 and size 3.Data has been taken from 16 males and 60 female. From the total data set we can see that 38 workers consider group size as size 2 and 39 workers consider their group size as size 3.The mean of subjects total task for 8 males who has group size 2 is 23.38 and the mean of subjects total task for 9 males who has group size 3 is 19.78 and their standard deviation are 5.125 and 3.456 respectively where as the mean of subject total task for 30 females who has group size 2 is 20.67 and the mean of subject total task for 30 females who has group size 3 is 22.58 and their standard deviation are 4.180 and 5.698 respectively. We can combine effects from descriptive statistic table, the mean of subjects total task for size 2 is 21.24 and for size 3 is 22.34.We can see that the means of these groups are significantly different but as a combine effects we can see that there is a significantly difference between the actual productivity of the subjects with respect to the group size and their gender. Normally, we use two-way ANOVA to make a conclusion about whether the combination of two Independent Variables has an effect on the Dependent Variable but we can use this technique only when three conditions (Independency, Normality and Homogeneity of Variances) meet. In this situation, three conditions have been met, so using two-way ANOVA seem an appropriate method to be used here.

Tests of Between-Subjects Effects Dependent Variable: subject's total task1 Source Type III Sum of Squares Corrected Model Intercept gender grpsize gender * grpsize Error Total Corrected Total 291.624
a

df

Mean Square

Sig.

Partial Eta Squared

3 1 1 1 1 73 77 76

97.208 25762.539 13.395 .184 182.355 23.666

4.108 1088.602 .566 .008 7.705

.009 .000 .454 .930 .007

.144 .937 .008 .000 .095

25762.539 13.395 .184 182.355 1727.597 40440.000 2019.221

a. R Squared = .144 (Adjusted R Squared = .109)

We use Factorial ANOVA to determine if the means are statistically different. But we do not need to use it to find out some basic information about mean differences. For this, we just simply compare our means. We have already done this task and came to know that there is significant statistical difference between means. We shall confirm this result by using ANOVA as well. In the Test of Between-Subjects Effects Box, Sig value helps us to determine if our condition means are relatively the same or if they are significantly different from one another. Put differently, this value helps us to determine if our Independent Variable has an effect. First pick the items of interest from Source column. We are interested to see the individual effect of both independent variables i.e. Gender and group size but our final aim is to analyze the combine effect of both IDV on our DV i.e. subjects total task. For Gender, the Sig. value is 0.454. As in, Sig value is greater than ( = 0.05); so we may conclude that there is no statistically significant difference between two conditions (Female & Male). It shows that mean subject total task do not significantly vary for female and male students. For grpsize ,the Sig. value is .930.As in,Sig value is greater than(=0.05);so we conclude that there is no statistically significant difference between two conditions(size 2 and size 3).It shows that mean subject total task do not significantly vary for size 2 and size 3. For Gender*Group size (Combine effect), the Sig. value is 0.007. As in, Sig value is less than ( = 0.05); so we may conclude that there is statistically significant difference among multiple conditions. It shows that when we take a combine effect of both variables (Gender and Group size), we can see that mean Subject total task significantly vary among different conditions. We have already seen this result in comparing means process where mean points were significantly different from each other.

Estimated Marginal Means

1. subject gender Dependent Variable: subject's total task1 subject gender Mean Std. Error 95% Confidence Interval Lower Bound male female 21.576 22.583 1.182 .628 19.221 21.332 Upper Bound 23.932 23.835

Pairwise Comparisons Dependent Variable: subject's total task1 (I) subject gender (J) subject gender Mean Difference (I-J) Std. Error Sig.
a

95% Confidence Interval for Difference Lower Bound


a

Upper Bound 1.661 3.674

male female

female male

-1.007 1.007

1.338 1.338

.454 .454

-3.674 -1.661

Based on estimated marginal means a. Adjustment for multiple comparisons: Least Significant Difference (equivalent to no adjustments).

Univariate Tests Dependent Variable: subject's total task1 Sum of Squares df Mean Square F Sig. Partial Eta Squared Contrast Error 13.395 1727.597 1 73 13.395 23.666 .566 .454 .008

The F tests the effect of subject gender. This test is based on the linearly independent pairwise comparisons among the estimated marginal means.

In the Univariate Tests Box, Sig value is 0.454 which is greater than (0.05). Thus, we may conclude that subject total task do not significantly differ for Female and Male students, F(1,73) = .566, p > 0.05. Thus, gender has no effect on the subject total task..

group size:

Estimates Dependent Variable: subject's total task1 group size Mean Std. Error 95% Confidence Interval Lower Bound size 2 size 3 22.021 22.139 .968 .924 20.092 20.296 Upper Bound 23.950 23.981

Pairwise Comparisons Dependent Variable: subject's total task1 (I) group size (J) group size Mean Difference (I-J) Std. Error Sig.
a

95% Confidence Interval for Difference Lower Bound


a

Upper Bound 2.549 2.786

size 2 size 3

size 3 size 2

-.118 .118

1.338 1.338

.930 .930

-2.786 -2.549

Based on estimated marginal means a. Adjustment for multiple comparisons: Least Significant Difference (equivalent to no adjustments).

Univariate Tests Dependent Variable: subject's total task1 Sum of Squares df Mean Square F Sig. Partial Eta Squared Contrast Error .184 1727.597 1 73 .184 23.666 .008 .930 .000

The F tests the effect of group size. This test is based on the linearly independent pairwise comparisons among the estimated marginal means.

In the Univariate Tests Box, Sig value is 0.930 which is greater than (0.05). Thus, we may conclude that subject total task do not significantly differ for size 2 and size 3 of group size, F(1,73) = 0.008, p >0.05. Thus, rating of group size has not a significant effect on the subject total task.

4. subject gender * group size Dependent Variable: subject's total task1 subject gender group size Mean Std. Error 95% Confidence Interval Lower Bound size 2 male size 3 size 2 female size 3 24.500 .888 22.730 26.270 19.778 20.667 1.622 .888 16.546 18.897 23.010 22.437 23.375 1.720 19.947 Upper Bound 26.803

We use Factorial ANOVA when we are interested in analyzing the combine effect of more than one independent variable on dependent variable. Although, we study the individual effect of variables as well but our ultimate aim still remains to analyze the joint effect of variables. In this situation, when we combine two independent variables Gender and group size, we see statistical significant difference in the subject total task. We have seen these results in comparing descriptive process. (Total, Size 2: Mean=21.24, S.D=4.446; Total, Size 3: Mean=23.41, S.D=5.599; ). It shows that mean subject total task differ with respect to combine effect of gender and group size. In Gender*Grpsize Box, we see that female students who consider their group size as size 2 lower mean of subject total task i.e. 20.667. Similarly, male students who consider their group size as size 3 have lower mean number of subject total task i.e.

19.778. So, these two combinations seem to be more appropriate because Subject total task is much lower than in any other combination.

Profile Plots

It can be much more elaborated with the graphical representation which has been given above. Female students have lower mean subject total task when they consider their group size as size 2 and male students have lower mean subject total task when they consider their group size as size 3.

Now, after analyzing results we come to know that Subject total task significantly differ in different conditions. As, we have find enough evidence to infer that our research hypothesis (H1) is true so we accept (H1) in the favor of Null Hypothesis (H0). Thus, we can conclude that gender and group size jointly affect the Subject total task.

You might also like