This document discusses key concepts in inferential statistics and hypothesis testing. It explains that inferential statistics allow estimating population characteristics from sample data and are used to answer questions about comparisons or relationships. Hypothesis testing involves forming a null hypothesis, which is statistically tested to determine if an observed difference is likely due to chance or a real treatment effect. Type 1 and type 2 errors in hypothesis testing are defined. The document also outlines factors like level of significance, power, and choice of parametric vs non-parametric tests based on the study design and data.
2. Inferential statistic allows us to estimate
population characterstics from sample data.
It is used to answer questions concerning
comparison or relationships.
eg. – Is one treatment more effective than another?
- Is there a relationship between length of time
treatment was given and improvement in the
patient.
3. We would see some difference between groups
even when a treatment is not effective.
This is because of chance difference in subject
characterstics.
We need to decide if an observed effect is because
of chance or due to real treatment effects.
Eg. – 2 frozen shoulder patients with different pain
tolerance.
We do this through hypothesis testing.
4. Hypothesis is the probable explanations of
outcome, which can be 2.
The researcher’s goal is always be to statistically
test the null hypothesis.
The difference between the
groups has occurred by
chance as a result of
sampling error.
NULL HYPOTHESIS (H₀)
There is a true difference
between the groups, and
the treatment was
effective.
ALTERNATIVE
HYPOTHESIS (H₁)
5. H₀ :- μₐ = μb OR μₐ – μb = 0
Where μₐ & μb are the population means.
Even when there is no true treatment effect, groups
will not have equal means.
So rather than saying means are equal we say they
are not “significantly different”, that is observed
difference would be the result of chance.
6. Therefore we say, we reject or do not reject/ retain
the null hypothesis
The purpose of an experiment is to give the data a
chance of disproving the null hypothesis.
7. Observed difference is too large to be considered as
a result of chance alone.
stated as
Directional and Non directional hypothesis –
alternative hypothesis is directional, when it
specifies that one group mean would be larger than
other.
Eg. – stating that the group recieving mulligan
mobilzation will show greater improvement in ROM
than one recieving maitland mobilization.
H1 : μₐ ≠ μb μa - μb ≠ 0
H1 : μₐ > μb μa - μb > 0
8. Type 1 error : Occurs due to rejection of statement
which should have been accepted.
Type 2 error : Occurs due to acceptance of statement
which should have been rejected.
9. TYPE 1 ERROR : H₀ is rejected
Researcher that there is difference between the groups,
when in reality it doesn’t.
False +ve.
Can result in ineffective treatment to be administered.
TYPE 2 ERROR : H₀ is accepted
Researcher concludes that there is no difference
between the groups.
False –ve
Researcher may ignore an effective treatment.
May occur due to inadequate sample size.
Denoted as α
Denoted as β.
10. It is a criterion for judging that the observed
difference is by chance(sampling error) or there for
real.
Larger the observed difference less likely it is to be
due to error.
We know there is some chance that observed
difference may be a result of sampling error.
α is the probablity of committing type 1 error –
risk of stating that a treatment is effective when it is
not.
11. Selected level of α will be maximum acceptable risk.
Standard value : α = 5% = 0.05
Level of significance : p < 0.05
13. Probablity of detecting a difference if true
difference exists.
Since probablity of type 2 error is β,
Conventional value 80-90%
POWER = 1-β
β
14. If you have rejected the null hypothesis but
level of significance (p = 0.1), what type of error
have you committed ?
While stating the null hypothesis, the treatment is
said to be not effective when the group means are
not
_________ ____________.
Power is probablity of ___________ the null
hypothesis when true difference exists.
Type 2 error is denoted as __________.
Conventional value of type 1 error is α = _________.
16. Based on whether data is parametric or non parametric.
PAIRED tTEST / DEPENDENT tTEST UNPAIRED tTEST /
INDEPENDENT tTEST
2 groups matched or same group tested twice.
Eg - Study on twins.
- Pre post design
2 unmatched groups
Group 1 OUTCOMES
Group 2 COMPARED
GROUP : PRETEST POSTTEST
COMPARED
Group 1
INTERVENTION COMPA
Group 2 - RE
PEFR of Interns in sitting and standing position 2 groups of 10 Frozen shoulder
each given mulligan
mobilization.
MATCHING
INTERVENTION
INTERVENTION
17. DEPENDENTTEST INDEPENDENT TEST
1 group tested ≥ 3 times ≥ 3 groups tested and compared
Outcome 1
1 Group Outcome 2
Outcome 3
Group 1
Group 2
Group 3
1 group of 10 OA patients given different
mobilizations to find its effects on ROM.
3 different groups of OA given three
different mobilizations and compared.
18. Exam result
of 5th grade
students in
1st and 2nd
unit test.
P F
Unit1
Unit2
Exam result of
5th and 6th
grade in the
same unit test.
2 x 3
3 x 3
3 x 2
19. Q 1. To determine the effect of 6 weeks of breathing exercises
on 6MWD in 30 healthy participants.
Q. 2To find out the difference between the effect of plank, curl-
ups and diet modifications onWaist hip ratio among 3
different groups of mildly obese people.
Q.3 Effect of 6 weeks of E. S on muscle strength in 10 patients
of tendon transfer.
Q.4 Difference between effects of maitland, mulligan, and
kalterborne glide on Pain measured with NRS, in 10 patients
of frozen shoulder.
Q.5To determine if 1wk of suryanamaskar is as effective as
coventional therapy in increasing lumbar flexibility of LBP
patients
20. Q.6 Gender specific effect on flexibility of a group of frozen
shoulder patients after administration of maitland, mulligan
and kalterborne glides
Q 7.What is the difference between response of Amputation
and CRPS patients towards mirror therapy.
Q 8.To find out whether traffic policemen posted at 3 different
areas of Mumbai – Busiest road, toll naka and a non crowded
intersection have low back pain