Anova

NOTES ON ANALYSIS OF VARIANCE (ANOVA) COMPILED BY RAM SINGH LECTURER SCHOOL OF MANAGEMENT SCIENCES LUCKNOW
1. What are the assumptions required to conduct a One-Way ANOVA? Explain each one. Answer:
-Independence of observations -Homogeneity of variance -Scores/observations normally distributed in each group (and in each parent population). See book/notes for more details.
2. To assess the likely truth or falsity of the null hypothesis, we compute two estimates of population variance. What are they and which is affected by the truth or falsity o the null hypothesis? Answer
MSwithin/error MSbetween(group) this depend on whether the null is true or false. If true, should be the same as MSwithin/error 3. Suppose I conducted an experiment to determine which TV network can make people laugh more on Thursday nights. I have three groups: One group watches NCB, the second group watched ABC, and the third group watched CBS. All participants watch TV from 8pm-10pm with an audio recorder. The experimenter then listens to the recording and counts the number of time the participants laugh.
6a. What statistical test should I use? A) independent sample t-test B) related-sample t-test C) one-way ANOVA D) Repeated-measures ANOVA
6b. In this experiment, what are some of the reasons for between group variability? That is, what are some of the reasons that the groups in an experiment may have different levels of laughter? (In other words, what are some of the reasons that people in the NBC group might have higher levels of laughter relative to the other two groups?)
Group difference - One station produced more laughter than the other (i.e., there was an effect of TV station). Experimental errorsome laughs might have been quiet. Individual difference participants might have varied in the extent to which they laugh out loud.
6c.In this experiment, what are some of the reasons for within group variability? That is, what are some of the reasons that the subjects/participants may have different levels of laughter?
Experimental errorsome laughs might have been quiet. Individual difference participants might have varied in the extent to which they laugh out loud.
4. What is a post-hoc test and why do we conduct these?
Statistical procedures conducted to determine whether 2 specific groups differ significantly. We want to know which groups are different from one another, not just that at least one group is different from at least one other group in our experiment.
5. When is it appropriate to conduct a post-hoc test?
When we have a significant F-test
A pool of participants was randomly divided into FIVE treatment groups. The groups were administered daily doses of vitamin C over a 12-month period. The data in the table represents the number of cold and flu viruses reported by the participant as a function of their vitamin C dosage. Using = .05, analyze the data using the correct statistical procedure.
0mg 6 5 3 2
250mg 3 4 5 4
500mg 3 3 4 2
1000mg 4 1 0 3
2000mg 1 0 2 1
X1 = 16 X12 = 74 XTOT = 56 XTOT2 = 210
X2 = 16 X22 = 66
X3 = 12 X32 = 38
X4 = 8 X5 = 4 X42 = 26 X52 = 6
What kind of an ANOVA will you need to conduct?
One-way ANOVA
STEP 1: State the null and alternative hypotheses.
Ho : The means are the same for each condition H1 : At least one of these group means is different from another
STEP 2: Set up the criteria for making a decision about your null hypothesis.
dfbetween = K 1 = 5 1 = 4 dfwi/error = N K = 20-5 = 15
so Fcrit = 3.06
STEP 3: Summarize the data into the appropriate tests statistic.
SStot = 210 (562/20) = 53.2
SSwithin/error = (74 [162/4]) + (66 [162/4]) + (38 [122/4]) + (26 [82/4]) + (6 [42/4]) = 10+2+2+10+2 = 26
SSbetween/group = ([162/4] + [162/4] + [122/4] + [82/4] + [42/4]) - (562/20) = (64+64+36+16+4)-156.8 = 27.2
MSbetween/group = 27.2/4 = 6.8
MSwithin/error = 26/15 = 1.73
F = 6.8/1.73 = 3.93 STEP 4: Evaluate the null hypothesis: Reject for Fail to Reject? 6. A new treatment meant to help those with chronic arthritis pain was developed and tested for its long-tern effectiveness. Participants in the experiment rated their level of pain on a 0 (no pain) to 9 (extreme pain) scale at three-month intervals. Was the treatment effective? = .05.
Participant 1 2 3 4
Before 8 6 7 5 X1 = 26 pTOT = 89
3mo 7 5 6 5 X2 = 23
6mo 6 5 6 4 X3 = 21
9mo 6 4 5 4 X4 = 19 p1 = 27 p2 = 20 p3 = 24 p4 = 18
X12 = 174 XTOT = 89 XTOT2 = 515
X22 = 135
X32 = 113
X42 = 93
What kind of an ANOVA will you need to conduct?
Repeated-measures ANOVA
STEP 1: State the null and alternative hypotheses.
Ho : The means are the same for each condition H1 : At least one of these group means is different from another OR people experienced less pain after receiving at least one of these doses compared to at least one other dose.
STEP 2: Set up the criteria for making a decision about your null hypothesis.
dfbetween = K 1 = 4 1 = 3 dferror = (k-1)(s-1) = 3*3 = 9 so Fcrit = 3.86
STEP 3: Summarize the data into the appropriate tests statistic.
SStot = 515 (892/16) = 19.94
SSbetween/group = ([262/4] + [232/4] + [212/4] + [192/4]) - (892/16) = (169+132.25+110.25+90.25)-495.06 = 6.69
SSwithin/error = (174 [262/4]) + (135 [232/4]) + (113 [212/4]) + (93 [192/4]) = 5+2.75+2.75+2.75 = 13.25
SSsubjects/within pure = (272/4 + 202/4 + 242/4 + 182/4) (892/16) = (182.25+100+144+81) 495.06 = 12.19
SSerror = SSwithin/error - SSsubjects/within pure = 13.25-12.19 = 1.06
MSbetween/group = 6.69/3 = 2.23
MSerror = 1.06/9 = .118
F = 2.23/.118 = 18.90
STEP 4: Evaluate the null hypothesis: Reject for Fail to Reject?
Interpret your decision. That is, what do you conclude from this experiment.
The dosage did affect the amount of pain participant felt.
If you wanted, could you conduct a post-hoc test on this data?
YES
Q1. Which
is the best coffee (most cups ordered): Blue-Label Green-Label Red-Label 13 1 5 4 1 2 10 2 2 13 2 2 11 2 6 3 4 4 Sum 54 12 21 87 Sum X2 584 30 89 703 N 6 6 6 18 SS 98 6 15.50 119.50
Between Within Total
SS 163.00 119.50 282.50
df 2 15 17
ms 81.50 7.97 16.62
F = 10.23 Critical value of F = 3.68. F is significant. Which is the best coffee. Youll have to do t-tests to find out. But looking at the numbers above, you can guess that Blue Label is most likely to be the big winner. But youll know more after you do the post tests.
Item 4 Which is candle lasts more days? Non-Sented Low-Sented High-Sented 3 5 8 9 1 2 5 2 6 11 5 4 Sum 28 13 20 61 2 Sum X 236 55 120 411 N 4 4 4 12 SS 40 12.75 20 72.75 Between Within Total SS 27.17 72.75 100.92 df 2 9 11 ms 14.08 8.08 9.17
F = 1.74 Critical value of F = 4.26. F is not significant. Which is candle lasts more days? No significant difference between them. Choose any.

Anova

Uploaded by

Anova

Uploaded by

NOTES ON ANALYSIS OF VARIANCE (ANOVA) COMPILED BY RAM SINGH LECTURER SCHOOL OF MANAGEMENT SCIENCES LUCKNOW

4. What is a post-hoc test and why do we conduct these?

5. When is it appropriate to conduct a post-hoc test?

When we have a significant F-test

X1 = 16 X12 = 74 XTOT = 56 XTOT2 = 210

What kind of an ANOVA will you need to conduct?

STEP 1: State the null and alternative hypotheses.

dfbetween = K 1 = 5 1 = 4 dfwi/error = N K = 20-5 = 15

STEP 3: Summarize the data into the appropriate tests statistic.

SStot = 210 (562/20) = 53.2

SSbetween/group = ([162/4] + [162/4] + [122/4] + [82/4] + [42/4]) - (562/20) = (64+64+36+16+4)-156.8 = 27.2

MSbetween/group = 27.2/4 = 6.8

MSwithin/error = 26/15 = 1.73

X12 = 174 XTOT = 89 XTOT2 = 515

What kind of an ANOVA will you need to conduct?

STEP 1: State the null and alternative hypotheses.

dfbetween = K 1 = 4 1 = 3 dferror = (k-1)(s-1) = 3*3 = 9 so Fcrit = 3.86

STEP 3: Summarize the data into the appropriate tests statistic.

SStot = 515 (892/16) = 19.94

SSbetween/group = ([262/4] + [232/4] + [212/4] + [192/4]) - (892/16) = (169+132.25+110.25+90.25)-495.06 = 6.69

SSerror = SSwithin/error - SSsubjects/within pure = 13.25-12.19 = 1.06

MSbetween/group = 6.69/3 = 2.23

MSerror = 1.06/9 = .118

STEP 4: Evaluate the null hypothesis: Reject for Fail to Reject?

The dosage did affect the amount of pain participant felt.

If you wanted, could you conduct a post-hoc test on this data?

Between Within Total

SS 163.00 119.50 282.50

ms 81.50 7.97 16.62

You might also like