Six Sigma DMAIC Process - Analyze Phase - Analysis Examples

One-Sample t-Test Example:
Definition: A one-sample t-test compares a population mean with a specified reference value.
Situation: We have 50 turnaround-time observations and want to know whether our population mean is greater than the competitor's benchmark of 4.5 minutes.

Null: Mean TAT <= 4.5 minutes
Alternate: Mean TAT > 4.5 minutes

Row	TAT (in Mins)	TAT (in Mins)	TAT (in Mins)	TAT (in Mins)	TAT (in Mins)
1	4.82	4.41	4.60	4.68	4.80
2	4.76	4.62	4.83	4.49	4.67
3	4.52	5.04	5.09	4.48	4.91
4	4.71	4.62	5.46	5.01	4.44
5	5.09	4.68	4.67	4.63	4.58
6	5.03	4.77	4.55	4.20	4.97
7	4.86	5.16	4.87	4.94	4.44
8	4.68	4.98	5.04	4.62	4.47
9	4.71	4.58	4.91	4.55	4.84
10	4.49	4.93	4.74	5.50	4.94

Open Full-Size Diagram ↗

Turn Around Time (in Mins)

Open Full-Size Diagram ↗

Analysis Result

Result: The one-sided p-value is below 0.05, so we reject the null hypothesis. The data provide evidence that our mean turnaround time exceeds 4.5 minutes: our process is slower than the competitor's benchmark.


Two-Sample t-Test Example:
Definition: A pooled two-sample t-test compares two population means under an equal-variance model.
Situation: We have morning and night shifts and want to know whether their population mean turnaround times differ.

Null: The morning and night shifts have equal population mean turnaround times.
Alternate: The morning and night shifts have different population mean turnaround times.

Row	Morning Shift	Morning Shift	Morning Shift	Morning Shift	Morning Shift	Night Shift	Night Shift	Night Shift	Night Shift
1	13.79	15.58	16.70	13.92	14.00	15.56	16.37	21.67	13.37
2	15.29	14.68	18.71	14.11	9.14	11.28	17.62	20.78	20.95
3	13.40	12.37	20.12	8.83	18.97	15.04	23.15	18.33	18.02
4	12.10	12.76	12.99	17.04	16.53	22.29	14.92	19.14	12.11
5	16.00	17.41	18.67	12.38	10.02	16.78	20.96	17.75	17.34
6	12.53	17.03	18.68	15.25	12.64	20.12	20.83	14.67	16.65
7	14.75	18.77	11.83	15.49	13.58	19.27	18.23	19.39	9.89
8	21.10	15.56	10.69	11.51	15.89	16.65	16.84	16.76	17.81
9	14.18	13.79	17.26	14.84	15.56	14.97	18.19	13.65	15.07
10	12.06	15.53	11.85	10.70	9.81	16.38	19.30	19.90	16.72
11	15.25	15.74	17.77	11.44	14.92	15.81	13.09	15.54	18.46
12	13.46	12.85	14.83	14.33	17.45	19.96	15.97	15.17	21.92
13	12.06	15.41	20.58	19.73	18.34	17.31	16.90	19.45	15.23
14	14.86	12.16	13.06	11.67	14.02	16.66	22.23	10.25	19.07
15	14.25	16.78	12.68	15.05	19.83	14.35	21.28	16.63	14.65
16	13.97	8.36	9.85	17.59	21.92	18.58	21.77	19.34
17	14.76	14.65	11.58	17.41	11.80	17.06	16.92	17.37
18	13.83	18.58	20.19	15.82	11.21	14.18	14.40	17.38
19	15.63	12.42	12.09	18.24	10.97	16.45	14.91	20.09
20	15.34	16.46	15.91	16.19	15.46	13.59	13.69	22.73

Open Full-Size Diagram ↗

Morning Shift and Night Shift Data

Open Full-Size Diagram ↗

Analysis Result

Result: The two-sided p-value is below 0.05, so we reject equal population means. The morning and night shifts differ in mean turnaround time; the morning shift has the smaller sample mean. The confidence interval is for morning minus night.


One-Way ANOVA Example:
Definition: One-way ANOVA tests whether three or more population means are all equal.
Situation: We compare the mean turnaround times of Staff1, Staff2, Staff3 and Staff4.

Null: All four population mean turnaround times are equal.
Alternate: Not all four population mean turnaround times are equal.

Row	Staff1	Staff1	Staff2	Staff2	Staff3	Staff3	Staff4	Staff4
1	20.56	13.43	18.18	16.17	13.55	8.93	19.43	21.40
2	13.84	14.43	16.86	14.61	6.55	9.82	19.15	14.67
3	14.30	12.51	15.27	20.12	15.84	11.88	15.55	17.75
4	15.66	12.49	18.80	19.62	15.83	13.16	17.18	14.29
5	15.26	18.64	15.44	14.15	14.07	14.20	18.84	19.72
6	12.77	13.80	15.19	13.72	13.08	12.82	18.03	18.98
7	8.38	14.08	16.81	15.71	15.94	15.84	19.08	22.14
8	14.01	16.49	16.21	18.98	8.55	11.83	19.35	15.58
9	13.34	15.25	19.54	14.40	14.00	12.82	16.10	16.30
10	13.28	14.02	16.01	18.63	9.92	17.53	18.59	15.17
11	14.95	13.12	19.18	18.34	15.63	11.96	19.05	15.48
12	18.64	10.43	16.86	15.29	16.04	12.08	21.64	17.42
13	14.22	12.08	14.87	14.17	14.20	14.13	20.26	16.74
14	15.30	13.33	17.34	17.53	13.24	13.61	20.94	17.82
15	15.77	10.63	16.98	16.06	16.26	19.32	20.29	11.04
16	17.27		16.84		15.59		14.12
17	19.22		20.73		11.76		19.52
18	17.25		18.77		11.01		14.36
19	12.90		17.28		12.40		19.40
20	15.80		16.98		10.17		19.85

Open Full-Size Diagram ↗

Staff 1, Staff 2, Staff 3 and Staff 4 Data

Open Full-Size Diagram ↗

Analysis Result

Result: The p-value is below 0.05, providing evidence that not all four population means are equal. We reject the equal-means null hypothesis. ANOVA alone does not identify which staff pairs differ; pairwise comparisons are needed for that conclusion.


Mood's Median Test Example:
Definition: Mood's median test compares the population medians of two or more groups.
Situation: We compare median turnaround times for Staff1, Staff2, Staff3 and Staff4, using the same 35 displayed observations per staff member as in the ANOVA example.

Null: All four population median turnaround times are equal.
Alternate: Not all four population median turnaround times are equal.

Row	Staff1	Staff1	Staff2	Staff2	Staff3	Staff3	Staff4	Staff4
1	20.56	13.43	18.18	16.17	13.55	8.93	19.43	21.40
2	13.84	14.43	16.86	14.61	6.55	9.82	19.15	14.67
3	14.30	12.51	15.27	20.12	15.84	11.88	15.55	17.75
4	15.66	12.49	18.80	19.62	15.83	13.16	17.18	14.29
5	15.26	18.64	15.44	14.15	14.07	14.20	18.84	19.72
6	12.77	13.80	15.19	13.72	13.08	12.82	18.03	18.98
7	8.38	14.08	16.81	15.71	15.94	15.84	19.08	22.14
8	14.01	16.49	16.21	18.98	8.55	11.83	19.35	15.58
9	13.34	15.25	19.54	14.40	14.00	12.82	16.10	16.30
10	13.28	14.02	16.01	18.63	9.92	17.53	18.59	15.17
11	14.95	13.12	19.18	18.34	15.63	11.96	19.05	15.48
12	18.64	10.43	16.86	15.29	16.04	12.08	21.64	17.42
13	14.22	12.08	14.87	14.17	14.20	14.13	20.26	16.74
14	15.30	13.33	17.34	17.53	13.24	13.61	20.94	17.82
15	15.77	10.63	16.98	16.06	16.26	19.32	20.29	11.04
16	17.27		16.84		15.59		14.12
17	19.22		20.73		11.76		19.52
18	17.25		18.77		11.01		14.36
19	12.90		17.28		12.40		19.40
20	15.80		16.98		10.17		19.85

Open Full-Size Diagram ↗

Staff 1, Staff 2, Staff 3 and Staff 4 Data

Open Full-Size Diagram ↗

Analysis Result

Result: The p-value is below 0.05, so we reject equal population medians. The overall test does not establish that every staff pair differs. These counts, medians and confidence intervals use the 35 observations shown for each staff member.


Chi-Square Test Example:
Definition: A chi-square test compares categorical counts with the counts expected under the null hypothesis. Here, we compare agreement proportions in two independent groups.
Situation: We want to know whether the proportions agreeing and disagreeing differ between the Men and Women groups.

Null: Agreement proportions are the same in the two groups.
Alternate: Agreement proportions differ between the two groups.

	Men	Women
Agree	15	20
Disagree	30	35

Open Full-Size Diagram ↗

Data for Chi-Square Test

Open Full-Size Diagram ↗

Analysis Result

Result: The p-value exceeds 0.05, so we fail to reject equal agreement proportions. These data do not provide evidence of an association between group and response. This result does not prove that the population proportions are identical.