Sei sulla pagina 1di 11

Macquarie University Stat270: Applied Statistics Assignment One Solution, First Semester 2009

Abstract The head circumference of male and female students in a second year statistics unit at Macquarie University were compared. These data support the commonly held belief that some physical measurements are higher for males than for females, with average head circumferences significantly higher in males than in females. This study also compared the right forearm lengths of males and females in tutorial groups run by two different tutors. We found that there is no strong evidence of a difference in the forearm length of students in the two tutorial groups, indicating that measurements were taken in a similar way in both groups.
Introduction This report summarises the results of statistical analysis conducted to investigate some physical measurements in male and female adults. It is generally believed that physical measurements may differ in males and females. We will investigate this belief in this analysis for head circumferences. In addition, we investigate whether there was any difference in right forearm lengths in tutorials taken by two different tutors, which will indicate whether the two tutors gave similar instructions about the procedure to be followed. As these measurements may differ in males and females these comparisons are done separately for the two sexes. The data were collected in tutorials in a second year statistics unit, and a selection of these data were used for this analysis. In particular, two research questions are addressed:

1. 2.

Is there any evidence of a difference in average head circumference of males and females? For males and females separately, is there any evidence of a difference in average right forearm length for students in Lindas tutorial group (Tuesday 11am) compared with students in either of Brads tutorial groups (Monday 12 or 2)?

Methods All students in a second year statistics unit were asked to measure their head circumference, left and right handspan, right forearm length (all in centimetres) as well as recording their sex. This data was collected in tutorials in the second week of semester. Instructions about the methods for taking measurements were displayed in class, and are attached in the Appendix (page 4). Head circumference was measured just above the eyes, forearm length was measured from the elbow to the tip of the longest finger and handspan was measured from the tip of the thumb to the tip of the last figure with the hand spread on the desk. A selection of the data were made available in an Access data base for this task. Relevant data for this report was extracted from that data base and imported in to Minitab (Version 15) for statistical analysis. The students in this study were not randomly selected. They may be considered to be a convenience sample from a target population of Macquarie University students. In particular, we are not given information about how data were selected for inclusion in the dataset. We must assume that these data are representative of our target population. The statistical methods used to investigate the research questions are two sample t-tests. The assumptions of these methods are checked, and reported below.

Preliminary Data Exploration The sample contained records for fifty one students in total. The data set contained some entries for head circumference and forearm length which were obviously errors. This can be seen in the stem and leaf plot given in the Appendix (page 5). Head circumference was measured in centimetres, and results of 22, 23 and 38 were considered unreasonable for females. Similarly results of 31 and 32 were considered unreasonable for males. These measurements were recoded as missing values in the data set before any further analysis was undertaken. Some errors were also apparent in the data for right forearm measurements, as shown in the stem and leaf plots in the Appendix (page 6). One male recorded 102 cm, and two males recorded 17 and 19 cm. Each of these was recoded as a missing value, as it was felt that these were unreasonable measurements. The possible errors in the female right forearm data are less clear cut. Six observations were between 15cm and 26 cm. These were not removed at the initial stage of the analysis. Numerical summaries for the head circumference and right forearm length data are given in Table 1. Full descriptive statistics are given in the Appendix (pages 5 and 6). Figure 1 shows the distribution of Head Circumference by Sex, and Figure 2 shows the distribution of Right Forearm length by Sex and Tutorial group. Table 1: Descriptive Statistics: Head Circumference and right Forearm length (cm)
n Head Circumference - female Head Circumference - male 18 27 Mean 55.5 58.0 Median 55.8 58.0 Standard Deviation 2.3 1.9 Min 50.0 54.0 Max 59.0 62.0

Female Forearm - Brads group Female Forearm - Lindas group Male Forearm - Brads group Male Forearm - Lindas group

14 8 15 11

35.7 37.9 42.9 45.4

41.8 42.0 45.0 46.4

11.8 7.5 6.0 5.7

15.8 25.5 28.0 29.5

46 43.5 49.0 51.0

Figure 1: Head Circumference (cm) by Sex


62 60 Head Circumference Right forearm 58 56 54 52 50 Female Male 55 50 45 40 35 30 25 20 15 Tutor Sex

Figure 2: Right Forearm length by Sex and Tutorial group

Brad F

Linda

Brad M

Linda

There were slightly more males than females in the data set. In general, the head circumferences for females students are lower than those for male students, and there is less variability in the measurements in the female group. Females also tend to have slightly lower forearm measurements than males, but the male measurements are less variable than the females. There does not appear to be a large difference in the results for each tutor. The assumptions of the two sample t-test are that each sample has been drawn independently from a normal population with the same variance. We do not have sufficient information to judge independence in this case. For each two sample comparison two normal scores plots were produced

(one for each sample involved). In the case of the head circumference data, the normal scores plots appear to be linear, suggesting that it is reasonable to assume that each of the samples was drawn from a population which is normally distributed (Appendix, page 7). In the case of the forearm length data, the normal scores plots show unusual values at the lower end of the scale in each case, and it does not appear that the assumption of normality is reasonable (Appendix, page 8). It is possible that these low values are data errors, as they seem to be measurements which are very low for this physical characteristic in adults. The analysis was undertaken with these low values included and excluded, to see what effect this had on the results. In each case the assumption of equal variance was checked - see Appendix page 7 for the head circumference data, and page 8, 9 and 10 for the right forearm data. This assumption is considered reasonable for all comparisons (head circumference, male vs female: : F17, 26 = 1.37, p= 0.462, forearm, female, Linda vs Brad: F13, 7 = 2.49, p= 0.231, forearm, male, Linda vs Brad: F14, 10 = 1.10, p = 0.904). After the removal of unusual values in the forearm measurements, this assumption was re-checked, and found to be reasonable (forearm, female, Linda vs Brad: F9, 5 = 1.53, p= 0.667, forearm, male, Linda vs Brad: F12, 9 = 1.20, 0.796). Statistical Analysis After considering the results of the assumption checks, a pooled two sample t-test was carried out for the comparison of average head circumferences in males and females. It was found that males have significantly higher average head circumferences when compared with females ( t43 = 4.10, p . 0.000, Appendix page 7) For both males and females, no statistically significant difference was found between the right forearm lengths of those in the two tutorial groups (females: t20 = 0.47, p = 0.641, males: t24 = 1.04, p = 0.308, Appendix pages 8 and 9). This was also the case when the unusually small values were removed from the analysis (females: t14 = 0.84, p = 0.416, males: t21 = 1.90, p = 0.072, Appendix page 10). Conclusion Our results suggest that males do have significantly higher average head circumferences when compared with females. We are 95% confident that the average male head circumference is between 1.3 and 3.9 cm higher than that for females. Our results suggest that for both males and females there is no strong evidence of a difference in the forearm length of those in the different tutorial groups, suggesting that the procedure for taking measurements was similar in both groups. Both of these results need to be taken with caution. While these results do give some information about differences between males and females with respect to head circumference, we note that only selection of second year students at Macquarie University were used - and no information was given about how this selection was made. There were many problems with the data provided for this analysis. These may be due to a number of factors. Some values were too low to be considered reasonable and students who reported low measurements on one characteristic also tended to report low values on the other characteristic (Appendix, page 11). This may be due to students not following the instructions given, or measuring in inches rather than centimetres. To investigate the research questions further we would recommend a more tightly controlled study, using a random selection of students. This study should clearly set out how these measurements should be taken. It would be preferable to train those who would be taking the measurements, and also to ensure that they were taken as accurately as possible.

APPENDIX
Instructions for data collection, displayed in tutorials:

In this tutorial we will collect some data from you as described in the tutorial exercises. Please record your data below.
Handspan (cm) Right Left Right Forearm (cm) Head Circumference (cm) Sex

Please give this sheet to your tutor when you have completed it. handspan in cm spread hand and measure from tip of thumb to tip of small finger measure from elbow to tip of largest finger (you could put your arm down on the desk and mark the end points and then measure this) round just above the eyes

forearm length in cm

head circumference

APPENDIX
Stem-and-Leaf Display: Original Head Circumference Data
Stem-and-leaf of Head Circumference_F Leaf Unit = 1.0 3 3 3 4 4 4 8 (14) 2 2 3 3 4 4 5 5 223 8 0144 55555666667889 N = 29 N = 22

Stem-and-leaf of Head Circumference_M Leaf Unit = 1.0 2 2 2 2 3 (19) 7 3 3 4 4 5 5 6 12

4 5666666777888888899 0000012

Stem-and-Leaf Display: Head Circumference data after removal of unusual values


Stem-and-leaf of HC_F* Leaf Unit = 0.10 N* = 4 1 2 2 2 4 9 9 4 3 1 50 51 52 53 54 55 56 57 58 59 0 0 05 00005 00026 0 05 0 N = 27 N = 18

Stem-and-leaf of HC_M* Leaf Unit = 0.10 N* = 2 1 2 8 11 (7) 9 7 2 1 54 55 56 57 58 59 60 61 62 0 8 000005 005 0000125 55 00005 0 0

APPENDIX
Descriptive Statistics: HC_F*, HC_M*
Variable HC_F* HC_M* Variable HC_F* HC_M* N 18 27 N* 4 2 Mean 55.461 58.041 SE Mean 0.533 0.372 StDev 2.261 1.935 Minimum 50.000 54.000 Q1 54.875 56.000 Median 55.750 58.000 Q3 56.700 60.000

Maximum 59.000 62.000

Stem-and-Leaf Display: Original data, Right forearm


Stem-and-leaf of Right forearm_F Leaf Unit = 1.0 3 4 6 6 8 (13) 1 1 2 2 3 3 4 4 567 3 56 89 0022233333444 6 N = 29 N = 22

Stem-and-leaf of Right forearm_M Leaf Unit = 1.0 2 4 6 (20) 3 1 1 1 1 1 1 2 3 4 5 6 7 8 9 10 79 89 18 33344555556666677779 01

After removal of unusual values:


Descriptive Statistics: Right forearm_F
Variable Right forearm_F Variable Right forearm_F Tutor_F Brad Linda Tutor_F Brad Linda N 14 8 N* 0 0 Mean 35.67 37.88 Maximum 46.00 43.50 SE Mean 3.15 2.65 StDev 11.80 7.48 Minimum 15.80 25.50 Q1 21.63 29.50 Median 41.80 42.00

Q3 44.00 42.75

Descriptive Statistics: Right forearm_M


Variable Right forearm_M Variable Right forearm_M Tutor_M Brad Linda Tutor_M Brad Linda N 15 11 N* 3 0 Mean 42.93 45.35 Maximum 49.00 51.00 SE Mean 1.54 1.72 StDev 5.96 5.69 Minimum 28.00 29.50 Q1 43.00 44.40 Median 45.00 46.40

Q3 46.50 47.80

APPENDIX
Comparison of Head Circumferences:
Probability Plot of HC_F*
Normal
99 Mean StDev N AD P-Value 55.46 2.261 18 0.653 0.074 99

Probability Plot of HC_M*


Normal
Mean StDev N AD P-Value 58.04 1.935 27 0.458 0.245

95 90 80

95 90 80

Percent

70 60 50 40 30 20 10 5

Percent 50 52 54 56 HC_F* 58 60 62

70 60 50 40 30 20 10 5

54

56

58 HC_M*

60

62

Test for Equal Variances: HC_F*, HC_M*

95% Bonferroni confidence intervals for standard deviations HC_F* HC_M* N 18 27 Lower 1.63179 1.47470 StDev 2.26122 1.93495 Upper 3.61168 2.78322

F-Test (Normal Distribution) Test statistic = 1.37, p-value = 0.462 Levene's Test (Any Continuous Distribution) Test statistic = 0.01, p-value = 0.932

Two-Sample T-Test and CI: HC_F*, HC_M*


Two-sample T for HC_F* vs HC_M* HC_F* HC_M* N 18 27 Mean 55.46 58.04 StDev 2.26 1.93 SE Mean 0.53 0.37

Difference = mu (HC_F*) - mu (HC_M*) Estimate for difference: -2.580 95% CI for difference: (-3.850, -1.309) T-Test of difference = 0 (vs not =): T-Value = -4.10 Both use Pooled StDev = 2.0701

P-Value = 0.000

DF = 43

APPENDIX
Comparison of Right Forearm lengths:
Probability Plot of Right forearm_F_Brad
Normal
99 Mean StDev N AD P-Value 35.67 11.80 14 1.757 <0.005 99

Probability Plot of Right forearm_F_Linda


Normal
Mean StDev N AD P-Value 37.88 7.482 8 1.103 <0.005

95 90 80

95 90 80

Percent

70 60 50 40 30 20 10 5

Percent 0 10 20 30 40 50 Right forearm_F_Brad 60 70

70 60 50 40 30 20 10 5

20

30

40 Right forearm_F_Linda

50

60

Probability Plot of Right forearm_M_Brad


Normal
99 Mean StDev N AD P-Value 42.93 5.961 15 1.640 <0.005 99

Probability Plot of Right forearm_M_Linda


Normal
Mean StDev N AD P-Value 45.35 5.694 11 1.215 <0.005

95 90 80

95 90 80

Percent

70 60 50 40 30 20 10 5

Percent 30 35 40 45 50 Right forearm_M_Brad 55 60

70 60 50 40 30 20 10 5

30

35

40 45 50 Right forearm_M_Linda

55

60

Test for Equal Variances: Right forearm_F_Brad, Right forearm_F_Linda


95% Bonferroni confidence intervals for standard deviations Right forearm_F_Brad Right forearm_F_Linda N 14 8 Lower 8.19031 4.68091 StDev 11.8002 7.4821 Upper 20.5069 17.1377

F-Test (Normal Distribution) Test statistic = 2.49, p-value = 0.231 Levene's Test (Any Continuous Distribution) Test statistic = 0.72, p-value = 0.405

Two-Sample T-Test and CI: Right forearm_F_Brad, Right forearm_F_Linda


Two-sample T for Right forearm_F_Brad vs Right forearm_F_Linda Right forearm_F_Brad Right forearm_F_Linda N 14 8 Mean 35.7 37.88 StDev 11.8 7.48 SE Mean 3.2 2.6

Difference = mu (Right forearm_F_Brad) - mu (Right forearm_F_Linda) Estimate for difference: -2.20 95% CI for difference: (-11.90, 7.50) T-Test of difference = 0 (vs not =): T-Value = -0.47 P-Value = 0.641 Both use Pooled StDev = 10.4930

DF = 20

Test for Equal Variances: Right forearm_M_Brad, Right forearm_M_Linda

APPENDIX
95% Bonferroni confidence intervals for standard deviations Right forearm_M_Brad Right forearm_M_Linda N 15 11 Lower 4.18351 3.79081 StDev 5.96078 5.69357 Upper 10.1032 10.9427

F-Test (Normal Distribution) Test statistic = 1.10, p-value = 0.904 Levene's Test (Any Continuous Distribution) Test statistic = 0.05, p-value = 0.830

Two-Sample T-Test and CI: Right forearm_M_Brad, Right forearm_M_Linda


Two-sample T for Right forearm_M_Brad vs Right forearm_M_Linda Right forearm_M_Brad Right forearm_M_Linda N 15 11 Mean 42.93 45.35 StDev 5.96 5.69 SE Mean 1.5 1.7

Difference = mu (Right forearm_M_Brad) - mu (Right forearm_M_Linda) Estimate for difference: -2.42 95% CI for difference: (-7.21, 2.37) T-Test of difference = 0 (vs not =): T-Value = -1.04 P-Value = 0.308 Both use Pooled StDev = 5.8509

DF = 24

Comparison of Right Forearm lengths, after removal of additional unusual values: Test for Equal Variances: Right forearm_F_Brad, Right forearm_F_Linda

Probability Plot of Right forearm_F_Brad


Normal
99 Mean StDev N AD P-Value 42.71 2.165 10 0.512 0.146 99

Probability Plot of Right forearm_F_Linda


Normal
Mean StDev N AD P-Value 41.83 1.751 6 0.661 0.042

95 90 80

95 90 80

Percent

Percent

70 60 50 40 30 20 10 5

70 60 50 40 30 20 10 5

38

40

42 44 Right forearm_F_Brad

46

48

37

38

39

40 41 42 43 Right forearm_F_Linda

44

45

46

Probability Plot of Right forearm_M_Brad


Normal
99 Mean StDev N AD P-Value 45 2.525 13 0.540 0.133 99

Probability Plot of Right forearm_M_Linda


Normal
Mean StDev N AD P-Value 46.94 2.301 10 0.216 0.786

95 90 80

95 90 80

Percent

70 60 50 40 30 20 10 5

Percent 40.0 42.5 45.0 47.5 Right forearm_M_Brad 50.0 52.5

70 60 50 40 30 20 10 5

40

42

44 46 48 Right forearm_M_Linda

50

52

APPENDIX
95% Bonferroni confidence intervals for standard deviations Right forearm_F_Brad Right forearm_F_Linda N 10 6 Lower 1.41591 1.02678 StDev 2.16459 1.75119 Upper 4.35873 5.01006

10

F-Test (Normal Distribution) Test statistic = 1.53, p-value = 0.667 Levene's Test (Any Continuous Distribution) Test statistic = 0.62, p-value = 0.443

Test for Equal Variances: Right forearm_M_Brad, Right forearm_M_Linda


95% Bonferroni confidence intervals for standard deviations Right forearm_M_Brad Right forearm_M_Linda N 13 10 Lower 1.73102 1.50534 StDev 2.52488 2.30130 Upper 4.51514 4.63403

F-Test (Normal Distribution) Test statistic = 1.20, p-value = 0.796 Levene's Test (Any Continuous Distribution) Test statistic = 0.00, p-value = 0.969

Two-Sample T-Test and CI: Right forearm_F_Brad, Right forearm_F_Linda


Two-sample T for Right forearm_F_Brad vs Right forearm_F_Linda Right forearm_F_Brad Right forearm_F_Linda N 10 6 Mean 42.71 41.83 StDev 2.16 1.75 SE Mean 0.68 0.71

Difference = mu (Right forearm_F_Brad) - mu (Right forearm_F_Linda) Estimate for difference: 0.88 95% CI for difference: (-1.37, 3.12) T-Test of difference = 0 (vs not =): T-Value = 0.84 P-Value = 0.416 Both use Pooled StDev = 2.0266

DF = 14

Two-Sample T-Test and CI: Right forearm_M_Brad, Right forearm_M_Linda


Two-sample T for Right forearm_M_Brad vs Right forearm_M_Linda Right forearm_M_Brad Right forearm_M_Linda N 13 10 Mean 45.00 46.94 StDev 2.52 2.30 SE Mean 0.70 0.73

Difference = mu (Right forearm_M_Brad) - mu (Right forearm_M_Linda) Estimate for difference: -1.94 95% CI for difference: (-4.07, 0.19) T-Test of difference = 0 (vs not =): T-Value = -1.90 P-Value = 0.072 Both use Pooled StDev = 2.4316

DF = 21

Relation between Head Circumference and Right Forearm length, original data:

APPENDIX

11

Scatterplot of Head Circumference vs Right forearm


60
Sex F M

Head Circumference

50

40

30

20 20 30 40 50 60 70 Right forearm 80 90 100 110

Scatterplot of Head Circumference vs Right forearm


60
Tutor Brad Linda

Head Circumference

50

40

30

20 20 30 40 50 60 70 Right forearm 80 90 100 110

Potrebbero piacerti anche